vvEPA
United States Environmental Monitoring Systems EPA 600 4 82-060
Environmental Protection Laboidtory f-eh !983
Agency Research Triangle Park NC 2771 i
Research and Development
Quality Assurance
Handbook for Air Pollution
Measurement Systems:
Volume IV. Meteorological
Measurements
-------
United States Environmental Monitoring Systems
Environmental Protection Laboratory
Agency Research Triangle Park NC 27711
Research and Development EPA-600/4-82-060 Feb. 1983
c/EPA Quality Assurance
Handbook for Air Pollution
Measurement Systems:
Volume IV. Meteorological
Measurements
Peter L. Finkelstein, Daniel A. Mazzarella, Thomas J. Lockhart,
William J. King, and Joseph H. White
-------
Notice
This document has been reviewed in accordance with U.S. Environmental
Protection Agency policy and approved for publication. Mention ol trade names or
commercial products does not constitute endorsement or recommendation for use.
Peter L Fmkelstem is a physical scientist in the Environmental Monitoring Systems
Laboratory. He is on assignment from the National Oceanic and Atmospheric
Administration. U.S. Department of Commerce
-------
Foreword
Measurement and monitoring research efforts are designed to anticipate
potential environmental problems, to support regulatory actions by developing an
m-depth understanding of the nature and processes that impact health and the
ecology, to provide innovative means of monitoring compliance with regulations
and to evaluate the effectiveness of health and environmental protection efforts
through the monitoring of long-term trends. The Environmental Monitoring
Systems Laboratory, Research Triangle Park, North Carolina, hasthe responsibility
for' assessment of environmental monitoring technology and systems;
implementation of agency-wide quality assurance programs for air pollution
measurement systems; and supplying technical support to other groups in the
Agency including the Office of Air, Noise and Radiation, the Office of Toxic
Substances and the Office of Enforcement.
This document was developed to help fulfill the need for a quality assurance
manual for meteorological measurements that are frequently made in conjunction
with air quality studies. A quality control and assurance program is a necessary part
of any environmental measurement system. Meteorological measurements are
made in conjunction with pollutant dispersion studies, modelvalidation studies and
legally mandated air pollution monitoring activities. This manual will guide and
assist in the development of specific quality assurance plans for those
meteorological measurement programs.
Thomas R Mauser
Director
Environmental Monitoring
Systems Laboratory
-------
Abstract
There is little information available on suggested quality assurance procedures to
be used with the collection of meteorological data. This manual is an attempt tof ill a
small part of that gap by suggesting QA procedures that can be used with
meteorological measurement programs designed to supplement air quality studies.
This manual is organized in three main sections. The first discusses the
measurement process for each meteorological variable from the perspective of
quality assurance. This includes general guidance and suggestions for judging the
adequacy of the process of selection of the measurement system (procurement
specifications and receiving inspections against the specifications), installation,
documentation of the initial system calibration, and the ongoing operating
procedures (calibration, preventive maintenance, repair and routine inspection.
The result of this first section is a method for estimating the accuracy and precision
of the measurement process. The second section gives guidance on the proper
siting of the various meteorological sensors to be compatible with the intended
application of the data. Where possible, this guidance follows that of the World
Meteorological Organization (WMO, 1971) although some changes are made
because it is recognized that the goals of WMO'schmatological program may not be
the same as those of an air pollution control agency. The final section reviews
various approaches to data validation and makes some recommendations for a
program that may be suitable for the needs and resources of an air quality study.
-------
Table of Contents
Section Page
Foreword ill
Abstract iv
List of Figures viii
List of Tables ix
1.0 Introduction 1
1 1 Organization 1
1.2 Implementation 1
2.0 Quality Assurance of the Measurement Process 2
2 1 Quality Assurance Program for Meteorological Measurements 2
2 1 1 Planning for a Quality Assurance Program 2
2.1 2 Operation and Management of a Quality Assurance
Program 2
2.1.2.1 Systems and Performance Audits 3
2.1.2.2 Interpretation of Audit Results 3
2.1.3 Quality Assurance Program Reports 4
2.2 Quality Assurance Considerations Common to All Variables 4
2.2.1 Instrument Procurement 5
2.2.2 Acceptance Testing 5
2.2.3 Calibrations 5
2.2.4 Audits 6
2.2.5 Operational Checks and Preventive Maintenance 6
2.2.6 Preparation for Field Installation 6
2.3 Wind Measurements 6
2 3 1 System Description 6
2.3 1 1 Wind Sensor Characteristics 9
2 3.1.2 Wind Data Requirements 10
2 3.2 Procurement 10
2.3.3 Acceptance Testing 10
2.3.4 Calibration 11
2.3.5 Installation 12
236 Operation of a Wind Measuring System 13
2.3 7 Preventive Maintenance (PM) 13
2.3.8 Audit Procedures 13
2.4 Temperature Measurements 14
2.4.1 Introduction 14
2.4.1.1 Sensor Characteristics—Accuracy 14
2.4.1 2 Solar Radiation Shields 15
2.4.1.3 Temperature Data Requirements 16
2.4.2 Procurement 16
2.4.3 Acceptance Testing 16
2.4.4 Calibration 16
2 4.5 Installation 17
2.4.6 Field Operation of a Thermometry System 17
2.4.7 Preventive Maintenance 17
2.4.8 Audit Procedures 17
-------
Table of Contents
(continued)
Section Page
2.5 Humidity/Dew Point Measurements 17
2.5.1 Introduction 17
2.5.1.1 Sensor Characteristics 19
2.5.1.2 Sensor Housings and Shields 19
2.5.1.3 Data Requirements 20
2.5.2 Procurement 20
2.5.3 Acceptance Testing 20
2.5.4 Calibration 20
2.5.5 Installation 21
2.5.6 Field Operation and Preventive Maintenance 21
2.5.7 Audit Procedures 21
2.6 Solar Radiation 21
2.6.1 Introduction 21
2.6.1.1 Instrument Characteristics 22
2.6.1.2 Recorders and Integrators for Pyranometers and
Net Radiometers 22
2.6.2 Procurement 22
2.6.3 Acceptance Testing 23
2.6.4 Calibration 23
2.6.5 Installation 23
2.6.6 Field Operation of a Solar Radiation System 24
2.6.7 Preventive Maintenance 24
2.6.8 Audit Procedures 24
2.7 Precipitation Measurements 25
2.7.1 Introduction 25
271.1 Instrument Characteristics 26
2.7.1.2 Windshields and Heaters 26
2.7.1.3 Precipitation Data Requirements 27
2.7.2 Procurement 27
2.7.3 Acceptance Testing 27
2.7.4 Calibration 27
2.7.5 Installation 27
2.7.6 Field Operation of a Precipitation Measurement
System 27
2.7.7 Preventive Maintenance 27
2.7.8 Audit Procedures 28
3.0 Methods for Judging Suitability of Sensor Siting 29
3.1 Introduction 29
3.2 Instrument Siting 29
3.2.1 Wind Speed and Direction 29
3.2.2 Temperature and Humidity 29
3 2.3 Radiation 29
3.2.4 Precipitation 30
3.2 5 Meteorological Towers 30
3.3 Station Siting 30
-------
Table of Contents
(continued)
Section Page
4.0 Meteorological Data Validation 32
4.1 Methods 32
4 2 Recommendations 33
5.0 Bibliography 36
VII
-------
List of Figures
Figure No. Title Page
2.3.1 Typical wind vanes by Climet Instruments, Inc. (left) and
R. M. Young Co. (right) 6
2.3.2 Propeller wind vanes by Bendix Fnez Instrument Division (upper
left) by R. M. Young Co. (upper right), and by Meteorology
Research, Inc. (lower) 7
2.3.3 Typical three-cup anemometers featuring conical cups. Manufacturers
are Belfort Instrument Company (upper left), Climet Instruments, Inc.
(upper right), and C. W. Thornthwaite Associates (lower) 9
2.3.4 Example of solar noon orientation form and tables 11
2.4.1 Examples of various radiation shields (McTaggart-Cowen and
McKay, 1 976) 15
2.4.2 A simple bath for calibrating thermometers (Middleton and
Sphilhaus, 1953) 16
2.5.1 An official National Weather Service Service sling psychrometer (a)
and an Assmann psychrometer (b) 18
2.5.2 A typical Dewcel sensor housing and transmitter 19
2.5.3 A pair of tower-mounted Gill aspirated radiation shields for
housing temperature and dew point sensors (R. M. Young Co.) 20
2.6.1 The features of a typical pyranometer (Carter, et al., 1977) 22
2.6.2 A Campbell-Stokes sunshine recorder (Department of the
Army, 1975) 22
2.6.3 Features of a typical pyrheliometer and tracking mount (Carter,
et al., 1977) 23
2.6.4 A Moll-Gorczynski solanmeter (Department of the Army, 1975) 24
2.7.1 Atypical standard ram gage (Belfort Instrument Company) 25
2.7.2 Automatic wet/dry precipitation collector 25
2.7.3 A typical weighing rain gage (left and typical tipping bucket ram
gage (Belfort Instrument Company) 26
3.2.1 Siting wind instruments; a 10-m tower, located at least
10 times the height of obstructions away from those obstructions
(not to scale) 29
4.2.1 Schematic flow of decisions in EMSL data validation scheme 35
VIII
-------
List of Tables
Table No. Title Page
2.1.1 Summary of Performance Audit Methods for Common
Meteorological Monitoring Instruments m the Field 3
2.1.2 Recommended Tolerance Limits for Audit Results 4
2.3.1 Types of Anemometers and Their Operating Principles 8
2.5.1 Principles of Humidity Measurement 18
2.6.1 Classification of Pyranometers According to Physical Response
Characteristics 23
3.2.1 Limits on Terrain and Obstacles Near Towers 30
41.1 Examples of Data Editing Criteria 34
4.2.1 Suggested EMSL Screening Criteria 35
-------
Section 1.0
Introduction
Most air pollution control agencies
have been measuring meteorological
variables for as long as they have been
measuring atmospheric contaminants.
Wind speed, direction, and temperature
are the usual measurements. For the
most part, these data have been filed
away and forgotten, to be retrieved
occasionally to help track down the
source of a noxious odor, for use in air
pollution episodes, or the accidental
spill of a toxic chemical.
Serious attempts to use the monitored
meteorological information to help
analyze air quality data or as input into a
diffusion model are usually met with
limited success Upon inquiry into the
status or quality of the data, one is
frequently told such things as' "Oh, that
station used to be O K., but since we put
it up m 1964 some trees have grown up
around it," or "Well, you can have that
data if you want, but we don't know if it
is any good—the propeller is still
spinning so I suppose it's 0 K." With
uncertainties such as these, it is no
wonder that the diffusion meteorologist
turns so often to the nearest National
Weather Service office for data This is
not meant as criticism of any air
pollution control agency For the most
part, these agencies have been under-
staffed, underfunded, and overworked
Meteorological data have had a uniquely
low spot on priority lists, if they were
thought about at all Federal agencies
have also been lax in setting good
examples of how to run meteorological
quality assurance programs
Recent events, however, have com-
bined to bring about a needed change in
this situation Increased emphasis on
models for decision-making has high-
lighted the need for better urban-scale
meteorological data Photochemical
models, especially, have a need for a
dense network of meteorological moni-
toring stations. Agencies have rapidly
come to realize that they have a need for
the meteorological data they have been
gathering all along In addition, the
recently published "Ambient Monitoring
Guidelines for Prevention of Significant
Deterioration (PSD)" (USEPA, 1980)
document requires new sources to
monitor meteorological and atmospheric
variables at many proposed sites for
new industrial facilities in order to
evaluate their air quality impact Finally,
a committee of the American Meteoro-
logical Society (Hanna, 1977) has
recommended that dispersion param-
eters used in models should be meas-
ured directly on site rather than con-
tinue use of standard values from tables
as has been done in the past.
In response to this growing interest in
and need for higher quality meteoro-
logical data, various air quality monitor-
ing groups have perceived the needfor a
formal quality assurance plan for the
collection of meteorological data. This
report addresses this need by suggesting
a quality assurance program that will be
useful for meteorological monitoring
done in support of air pollution studies
It is hoped that such a program would
benefit short- and long-term field
measurement programs Others in the
meteorological monitoring business
may also find it useful in the develop-
ment of their own quality assurance
programs
1.1 Organization
This report is organized in three main
parts Section 2 discusses the meteoro-
logical measurement system. It reviews
methods of measurement, procurement,
maintenance and calibration from the
perspective of quality assurance (OA).
The practical advice comes principally
from experience of the authors. The QA
procedures follow the principles of the
EPA Quality Assurance Handbook
(USEPA, 1976).
Section 3 provides guidance for
judging the suitability of the siting of the
various meteorological sensors Where
possible this guidance follows that of
the World Meteorological Organization
(WMO, 1971) although some changes
are made since it is recognized that the
goals of WMO's climatological program
may not be the same as those of an air
pollution control agency
The final section (Section 4) reviews
various approaches to data validation
procedures and makes some recom-
mendations for one example program
Both automatic and manual validation
procedures are discussed. A mixture of
both is suggested, but each monitoring
group will have to evaluate its own
situation before designing its procedures
1.2 Implementation
Before implementing a quality assur-
ance program, an agency should try to
evaluate the projected uses of the data it
is gathering and the value of that data
or, conversely, the cost of invalid data.
This exercise should help in the budget
decisions connected with the QA
program.
As with any other new program, the
implementation of a quality assurance
program in a laboratory that did not
previously have one will require an
additional expenditure of resources.
The adoption of a program on paper,
without allowing for the additional
manpower and money that are needed,
will not result in substantial improve-
ment in data quality. The challenge is to
develop a program that meets an
agency's data needs and is cost-
effective. This manual hasbeen prepared
to assist in achieving that goal.
-------
Section 2.0
Quality Assurance of the Measurement Process
All too often the meteorological
system is considered an infallible
instrument or group of instruments
that, once installed, yields accurate
measurements until catastrophic failure,
an easily recognized event, occurs. This
is not true. Meteorological instrumenta-
tion requires calibration, preventive
maintenance, and constant checking if
the data acquired are to be accurate and
complete
This section provides guidelines for
proper meteorological instrument oper-
ation to obtain good quality data. The
variables discussed here are: wind
speed, wind direction, temperature,
humidity/dew point, precipitation, and
total sky radiation.
For each instrument, selection and
acceptance criteria (procurement), lab
and field calibration procedures, re-
commended maintenance procedures,
and audit procedures are presented
Recommendations for the imple-
mentation and management of a quality
assurance (QA) program for meteoro-
logical instrumentation are presented
2.1 Quality Assurance
Program for Meteorological
Measurements
Quality assurance (QA) applied to
environmental monitoring consists of
both "the system of activities to provide
a quality product" (traditional quality
control) and "the system of activities to
provide assurance that the quality
control system is performing adequately"
(traditional quality assurance) The first
of these quality control (QC) functions
consists of those activities performed by
station operators directly on the instru-
ments, e g , preventive maintenance,
calibrations, etc. These activities are
described for each of the individual
instruments in Sections 2 3 through
2 8. The purpose of the second set of
activities is to manage the quality of the
data and administer corrective actions
as necessary to ensure that the data
quality requirements are met. This
section describes the planning, operation,
and reporting of the quality assurance
effort
2.1.1 Planning for a Quality
Assurance Program
A formal quality assurance program
should be designed into the monitoring
system so that provisions may be made
in the system design for desired quality
control checks and for better monitoring
of system operations. If these activities
are planned and provided for by incor-
poration of necessary readouts, calibra-
tion sources, etc., then they are more
likely to be performed in a satisfactory
manner
The quality assurance activities
necessary for a monitoring program are
determined by the program data quality
requirements which are, in turn,
determined by the purpose for which
the data are to be used. Consideration
must also be given to possible future
applications of the data.
The formal plans for quality assurance
must be presented in a document called
the QA Plan This document lists all
necessary quality-related procedures
and the frequency with which they
should be performed. Specific informa-
tion to be included is described below:
Project personnel responsibilities'
Responsibilities of personnel per-
forming tasks that affect data quality
Data reporting procedures: Brief de-
scription of how data are produced
delineating functions performed
during each step of the data processing
sequence.
Data validation procedures. Detailed
listing of criteria to be applied to data
for testing their validity, how the
validation process is to be carried
out, and the treatment of data found
to be questionable or invalid
Audit procedures Detailed description
of what audits are to be performed,
how often they are to be performed,
and an audit procedure (referencing
document procedures whenever
possible). Also, description of internal
and external systems audits including
site inspections by supervisory per-
sonnel or others.
Calibration procedures Detailed de-
scription of calibration techniques
and frequency for each of the sensors
or instruments being used. Both full
calibrations and zero and span checks
should be defined
Preventive maintenance schedule:
Detailed listing of specific preventive
maintenance functions and the fre-
quency at which they should be per-
formed Includes not only routine
equipment inspection and wearable
parts replacement but also functional
tests to be performed on equipment
Quality reports: Schedule and content
of reports for submission to manage-
ment describing status of quality
assurance program.
More details on the requirements for
a QA plan which is to be submitted to
EPA are available in "Interim Guidelines
and Specifications for Preparing Quality
Assurance Project Plans," prepared by
the Office of Monitoring Systems and
Quality Assurance (1980).
2.1.2 Operation and Management of
a Quality Assurance Program
The quality assurance program in-
cludes the implementation of all func-
tions specified in the QA Plan. This
implementation involves personnel at
all levels of the organization Technicians
who operate equipment must perform
preventive maintenance and QC checks
on the measurements systems for
which they are responsible They must
perform calibrations and, when required,
participate in internal audits of stations
run by other technicians. Their immedi-
ate supervisors should check to see that
all specified QA tasks are performed,
and should review logs and control
charts to ensure that potential problems
are corrected before significant data
loss occurs.
The overall QA program responsibility
lies with the person responsible for
quality assurance. His responsibility
includes the assessment of the quality
of the data acquired, the preparation of
QA status reports, and management of
the quality assurance effort in a cost-
effective manner The quality assurance
coordinator must track the implementa-
tion and effectiveness of the QA plan
through reports from subordinates,
personal communications, on-site in-
spections, and review of audit data
Data validation procedures also provide
indications of degradations of data
quality; however, these indications are
not produced until data problems occur
and data are being lost Jt is more cost
effective, considering the value of lost
data, to implement adequate assess-
ment and preventive measures to
correct data quality problems before
significant amounts of data are lost.
The audits—both performance and
system—provide the manager with the
best information for the quantitative
and qualitative assessment of the status
of the QA program The following
sections present considerations for the
-------
performance of audits and evaluation of
results
2.1.2.1 Systems and Performance
Audits-Systems and performance audits
should be the most quantitative and
unbiased measures of data quality
available to the QA coordinator The
systems audit is an inspection of
monitoring stations for indications of
proper quality control procedures and
adequacy of the instrumentation for
making the desired measurement
Inspections are made to determine
• adequacy of recordkeepmg
• level of preventive maintenance
• suitability of equipment used for
calibration and operational checks
• adequacy of operating procedures
Systems audits should be performed
at the beginning of a monitoring
program shortly before data acquisition
has begun and yearly thereafter If data
quality does not meet the requirements
of the program, the systems audit
should be performed more frequently
System audits should ideally be
performed by an impartial group com-
pletely independent of the group operat-
ing the monitoring program This is
especially important if the audits are to
be used to establish credibility for the
measurements being taken (i.e., for
demonstrating data quality to the
agency requiring the measurements or
for possible use in a court of law) If the
systems audit is simply to function as an
on-site inspection as part of the man-
agement review process, then it may be
performed by the supervisor or manager
responsible for the station operation
Performance audits provide a quanti-
tative indication of the accuracy of
measurements being made by the
physical verification of the instrument
calibration. Suggested audit methods
are described for each of the variables in
Sections 2 3 through 2.8 and summarized
in Table 211 Having independent
investigators execute the audit applies
also to performance audits
2.1.2.2 Interpretation of Audit Results-
The interpretation of audit results for
meteorological instrumentation will
necessarily be somewhat different from
interpretation for ambient air monitoring
instruments because of differences in
the audit techniques Air monitoring
equipment is accessible because it is
usually installed in air conditioned
shelters with the sample drawn into the
instrument through a tube Calibrations
or audits are performed by generating
known concentrations of the monitored
gas in a manifold and letting the
instrument draw its sample from that
manifold in the same manner in which it
would draw the ambient air sample. The
test conditions, therefore, accurately
simulate the measured conditions with
the measured quantity being carefully
controlled
Meteorological instrumentation, how-
ever, usually consists of a sensing
element that must be located directly in
the field being measured without
altering that field. Producing a known
field of meteorological variables at the
monitoring site is exceedingly difficult,
if not impossible. Once installed, the
instrument may only be checked by
generating an artificial field (e.g ,
spinning anemometer cups with a
motor, orienting the wind vane toward
aiming stakes, or submersion of a
temperature sensor m a bath) Checks
made in this manner do not detect or
quantify coupling errors that occur at
the sensor/air interface (slippage of
cups in wind because of increased
bearing friction, errors in indicated
direction due to geometrical asymmetry
of the vane, or effects of radiation on
temperature measurements). The checks
represent only a calibration of a portion
of the system and, consequently, may
not be routinely used to evaluate
accuracy on the entire measurement
system.
The instrument may also be checked
by the technique of collocated sensors,
in which the audit sensor is subject to
the same air/sensor interface errors as
the audited sensor. When audits of this
type are conducted, the ambient field as
measured by the audit device is assumed
Table 2.1.1. Summary of Performance Audit Methods for Common Meteorological Monitoring Instruments in the Field
Audit method
Variable
Collocated transfer standard
Measurement system
audit or calibration
Operational check
Wind speed
Wind
direction
Sigma theta
Temperature
Humidity
Solar
radiation
Precipitation
Anemometer with calibration
related to NBS
Wind vane with independent
orientation
Calculate from direction
samples
Thermometer with calibration
related to NBS in acceptable
aspirated radiation shield
Assmann psychrometer,
cooled-mirror dewpomter
Calibrated pyranometer
Not applicable
Known rate of rotations driving
sensor shaft. Starting torque
measurement for threshold
Alignment to orientation target
and rotate in three or more
known angular steps
Calculate from direction samples
Compare with transfer standard at
two or more temperatures in
stable thermal masses For
differential temperature test
zero difference at two
temperatures
No practical field method.
Return for lab calibration
Cover sensor for zero point.
Return for lab calibration
Introduce known volume of water
at rate about 1 inch/hour into
known inlet area (tipping
bucket} Use calibration
weights (weighing gage)
Substitute known frequency
or voltage
Substitute known
resistance for
potentiometer type
Substitute known
amplitude and wave shape
in frequency window
Substitute resistance or
voltage for sensors
Substitute resistance for
temperature
Substitute millivolts or
microvolts for sensor
Substitute event simulator
(tipping bucket} or
resistance (electric
weighing gage)
-------
Table 2.1.2. Recommended Tolerance Limits for Audit Results
Variable'
Calibration via
artificial field*
Collocated
sensors**
Wind speed
Wind direction
Temperature
Temperature difference
Humidity/dew point
Total sky radiation
Precipitation
±0 2 m/s
±2°
±0 25°C
±0 1°C
Return to manufacturer}
Return to manufacturer}
±01"
±0 5 m/s
±5°
±05°C
±02°C
±i.s°c***
±5% Full scale (70 W/m2)
±10% of reading
^Artificial field implies the simulation of the measured variable by artificial means (e.g., spinning anemometer cups by
controlled speed motor).
**Sensors utilizing the same measurement technique operated in similar housings located adjacent to the audited sensor
** Dependent on measurement technique; - some methods may be better.
tO/" an impartial standards laboratory. See "A Directory of Standards Laboratories" by National Conference of Standards
Laboratories, NBS, Boulder, CO 80303
'10- to 20-minute averaging time when averaging is required.
to be the known field. Care must be
taken when this method is used for
calibration or audit purposes because
the characteristics of the atmosphere
are not controlled and unknown spatial
and temporal variations may exist
The audit data indicate the state of
instrument calibration because most
audit procedures verify the total system
except for the air/sensor interface It is
possible for a system to receive a good
audit report and still make poor mea-
surements of meteorological conditions
because of problems at the air/sensor
interface or due to poor siting. However,
this is not likely if good sensors are used
and proper siting precautions are taken.
To further minimize the chances of
inaccurate measurements going un-
noticed, performance audits should
always be accompanied by inspections
for potential problems that might cause
the measured values to be unrepre-
sentative of ambient conditions.
The calibration errors indicated by a
performance audit should fall within the
limits specified in the QA Plan Table
2 1 2 lists suggested tolerance levels for
each of the measurement techniques
Exact limits should be determined by
monitoring program requirements Any
audit results falling outside these limits
indicate a problem that should be
rectified by normal maintenance proce-
dures and followed by reauditing of the
system. The audits are only checks and
should not be used to determine
calibration equation coefficients for the
instrument or for other types of data
correction or adjustment.
For meteorological measurement
systems that are audited by the collocated
instrument technique, two sets of data
will result. These are the normal
measured values and the data recorded
from the audit system If the monitoring
system is properly calibrated and
maintained, then the auditing system is
susceptible to the same levels of errors
as the monitoring system. To make a
quantitative assessment of the audit
data, two techniques are available; i.e ,
manual inspection and statistical-evalu-
ation:
Manual examination of audit vs sample
data for differences within acceptance
limits.* Either instantaneous values or
average values over the period of audit
may be compared. Also, if a collocated
test were run over a period sufficiently
long that widely varying levels of the
measured variable were encountered,
then results of a regression between the
audit and sample values will yield
difference and bias information
Statistical evaluation of audit results
either from audits performed on multiple
stations or multiple audits on a single
station. Simple tests are described in
the QA Handbook, Vol I, Section G,
which detect bias in system measure-
ments through audit data taken over a
network These tests include the sign
test and the paired t test More complex
tests are available
2.1.3 Quality Assurance Program
Reports
Periodically, the quality assurance
coordinator should prepare a report for
management describing the status of
the quality assurance effort. This report
should provide quantitative information
on the quality of data, activities performed
to improve data quality, and the cost of
the QA efforts Specifically, the report
should cover
• results of performance audits
• results of systems audits
• percentage of data reported
• cost of QA effort
• problems that degrade data quality
and corrective action taken
•Note that acceptance limits consider not only the
variations expected in the measurement system but
also variations in the audit measurement system
No general format exists that will
serve for all quality reports. The exact
contents will depend on the level of QA
activity for the organization and the
extent of the monitoring network It is
important that the document be clear
and concise Voluminous, highly detailed
information should be summarized or, if
necessary, placed in a separate appendix
Reports that are complex and difficult to
comprehend will not effectively com-
municate their message
2.2 Quality Assurance
Considerations Common to
All Variables
Two areas common to all systems
must be considered One is procedural
including procurement specifications,
acceptance testing, recordkeeping and
reporting The other is related to the
hardware of the system including power
supplies, recording systems, multi-
plexers, operating power source, and
some cables
The amount of the contributions to
the error of the common hardware parts
of a measurement system depend on
the type of system If the system is a
digital logging system where voltages
are provided by the signal conditioners
of each sensor to a common multiplexer,
analog to digital (A to D) converter and
digital recorder, the contribution of the
system to the error budget of the
measurement is usually small A good
quality A to D converter will be accurate
to 0 1 percent of full scale. Once
converted, the digital handling should
be error free except for the last digit
uncertainty If the resolution of the
digital data is also 0.1 percent of full
scale or the least significant digit is
properly rounded from such a resolution,
the contribution to the measurement
error is less than 0.2 percent Rounding
in digital systems can cause an un-
acceptable bias when it is not done
correctly This is easily checked by
-------
looking at a frequency distribution of
real data at the resolution of the
recorded data. When auditing a system
with a collocated transfer standard
where synchronous measurement
differences are examined, it is important
to restrict the differences to those
caused by the measurement system To
find any detectable bias from rounding,
filtering or averaging which might be
unexpected, examine the shape of the
frequency distribution (not cumulative)
for reasonableness. Othertestsdesigned
to challenge these functions of a digital
data system may also be used. It is most
important for the auditor of a digital
system to be satisfied that the system
works without bias by whatever method
might be appropriate
When the common part of the system
is something like a multipoint recorder
or a series of single-channel chart
recorders, the error from the recording
system may be as large or larger than
the error from the rest of the measure-
ment system for the variable This error
is tolerable if it is a backup recording. If it
is the primary data source, recording
error may bea seriousproblem; particu-
larly if the application requires a high
degree of accuracy.
The quality control effort common to
all variables must be documented from
observations taken directly from the
charts and logs maintained by the
station operators and their supervisors
The following reports should be kept for
the purpose of efficient station operation,
and they can be a useful source for any
good QA plan
• Calibration Log. The calibration log
contains detailed calibration informa-
tion including date and time, name of
person calibrating, calibration tech-
nique used, problems noted, sensor
adjustments made, and calibration
values and sensor responses (before
and after adjustments)
• Operational Log. The operational log
contains records of activities per-
formed on the meteorological mea-
suring system including preventive
maintenance, repairs, inspections,
and special test data not recorded
elsewhere. Some test parameters
should be maintained in the opera-
tional log ona regular basis to provide
baseline information on proper
operating conditions of the instru-
mentation. This information can be
invaluable in troubleshooting once
problems are noted by providing
nominal operating values and the
amount of expected variation in those
values. Examples of data that should
be recorded are readings associated
with any test positions on the
instrument, bridge excitation voltage
(if accessible), and line voltage.
Not all quality assurance recording
functions performed by the station
operator are strictly for the purpose of
furnishing information to management.
Some of the recordkeepmg functions
enable the station operator to spot
indications of potential problems before
data are lost, e g., the routine recording
of operational values described above.
Some data are subject to day-to-day
variations and it is difficult to evaluate if
individual values are within acceptable
limits Data of this nature are best
recorded and plotted using control
charts Typical control charts include
response to zero/span checks and
digital versus strip chart readings
2.2.1 Instrument Procurement
The function of QA in the procurement
process is to verify the presence in
purchase orders or contracts of per-
formance specifications in unambiguous
terms along with some indications of
how verification tests will be conducted
The importance of this function is
directly proportional to the importance
of the instrument having the specified
characteristics
In choosing any meteorological in-
strument or array of instruments, the
potential user should carefully evaluate
his needs for data Data requirements
should be identified and a list made of
suitable instrument configurations that
meet these requirements For most air
pollution monitoring applications,
meteorological instrument systems will
be commercial, off-the-shelf items. The
suitability of each instrument to perform
in the anticipated field environment is
evaluated by examining such charac-
teristics as stability, linearity, response
time, and durability
Budget considerations may appear to
eliminate certain candidates but should
be carefully examined Purchasing a
meteorological instrument is more than
a simple capital equipment outlay A
scientific instrument requires a certain
amount of attention Man-hours invested
in data reduction, preventive mainte-
nance, calibration and repair, and the
value of lost data are real budget
considerations Maintenance records of
candidate instruments should be checked
because the time a component spends
in the manufacturer's service depart-
ment represents wasted resources and
lost data. Generally it is best to purchase
components from one manufacturer
Although it is possible to obtain sensors,
bridge circuitry, housings and mounting
apparatus from different sources,
purchasing in this way splits responsi-
bility, invites problems of component
mismatching, and frequently results in
delays. As a rule of thumb, a delivery
schedule of approximately 3 months is
reasonable for most meteorological
instrumentation; 4 months is average.
2.2.2 Acceptance Testing
Never assume that the manufacturer
has sent a complete, working instrument.
Since warranties may expire in as short
a period as 90 days, the instrument
should be examined upon receipt,
particularly if it is not slated for
immediate deployment. Any damage to
the shipping carton should be docu-
mented with the shipper The instrument
should then be unpacked and assembled
according to the manufacturer's written
instructions. Each component should be
inspected for physical damage Open
access plates and inspect for loose
connections and other visible effects
All cables should be checked for
continuity Bench test the instrument
under controlled conditions Using the
operations and maintenance manuals,
check as many performance specifica-
tions as possible. Document all accep-
tance testing and bring problems to the
attention of the supplier or manufacturer
immediately
When an important performance
characteristic is specified, it is prudent
to verify that performance with an
acceptance test. If it is a design feature
(like overshoot or damping ratio on a
wind vane), it can usually be verified by
a test on a randomly selected sample of
the units purchased. The cost of this
type of documentation should be con-
sidered as a part of the cost of specifica-
tion. If verification acceptance testing is
not planned or budgeted, it should be
understood that the specification be-
comes an expression of desire or
expectation and it should be used to
describe the data collected only with a
footnote to qualify it as a manufacturer's
claim or by reference to tests by others
on the same kind of instrument
2.2.3 Calibrations
The calibration of the part of the
system common to all variables is
usually a simple task and should be
done initially in conjunction with the
acceptance inspection and testing It
may require simply challenging the
recording system with a voltage or
current within the operating range and
verifying that the output is within the
accuracy specification. If the recording
subsystem does not have a specific
specification, these test results will
define it Of course, it must be smaller
than the total measurement accuracy
specification
After this initial test, it is often
unnecessary to repeat the test or
calibration except in cases where
-------
repairs or replacements are involved.
That part of the system common to all
variables is usually included in the total
system calibration or audit done for
each variable and need not be considered
as an independent subsystem.
2.2.4 Audits
The audit is an integral part of the
quality assurance program for any field
measurement exercise. To minimize
bias, audits are usually performed by an
independent investigator not adminis-
tratively connected with the organization
conducting the measurement program.
The audit is similar in some respects toa
field calibration check in that a major
objective is to determine if the measure-
ment system has retained its calibration.
During an audit, however, no changes
are ever made to the instrument or to
the calibration coefficients.
The system audit is common to all
variables and is described m Section
2.1.2. It involves an examination of all
site logs and a review of instrument
siting and installation, daily operating
procedures, preventive maintenance
(PM) activities, and calibration methods.
A systems audit should be scheduled
near the beginning of a measurement
program
The performance audit is specific to
each variable and is discussed with
each variable in the following paragraphs'
of this section.
2.2.5 Operational Checks and
Preventive Maintenance
To ensure the highest possible data
quality, the user should establish a
preventive maintenance routine and
perform operational checks daily or as
often as the site is visited Except under
extremely unusual circumstances, no
piece of meteorological equipment
should go unattended for more than 30
days. Some instruments, such as
pyranometers, require daily attention to
assure data quality. Operational checks
include timing checks and adjustments,
servicing the recorder needs, logging
activities and observations, and cursory
inspection of data for anomalous
behavior. Preventive maintenance (PM)
involves specified routine activities
such as cleaning, tightening, lubricating,
and scheduled replacing of minor parts.
The user should maintain a log of all PM
and operations checks.
2.2.6 Preparation for Field
Installation
Certain routine preparation is required
before a meteorological sensor system
is deployed. If the instrument is not
newly purchased and has not recently
been operating, it should be checked
following the basic procedure used for
acceptance testing (paragraph 2 2.2).
The checkout is followed by bench
calibration, after which it is desirable to
set up the instrument in field configura-
tion near the main laboratory for a
shakedown period. If several instruments
are involved, side-by-side comparisons
can be made, strengthening the quality
assurance program The shakedown
period may vary in length but should at
least verify that the instrument is in
good working order. Following shake-
down, the system is dismantled and
carefully packed for shipment to the
field. A physical damage inspection
should be made at the destination
2.3 Wind Measurements
2.3.1 System Description
Wind measurements are of primary
importance in studies of the diffusion
and tra nsport of atmospheric pollutants.
These measurements include wind
speed, wind direction, and turbulence or
gustiness
There are many wind-measuring
systems commercially available. Some
are ruggedly constructed, designedfor a
wide range of applications, and require
minimum attention. The more delicate
instruments, such as those used for
measuring small-scale turbulence, can
only be used during selected periods of
favorable weather.
Almost any anemometerorwind vane
will provide some information on wind
characteristics. However, the quality of
wind data depends directly upon how
well the sensor is maintained and how
well the measuring equipment functions
as a system. Not only must the dynamic
characteristics of the sensor match
program data requirements, but the
sensor must also interface without
degradation of important performance
characteristics with the total data
acquisition system, which may include
the transducer, signal conditioner,
telemetry, data processor, and readout
device
Anemometers - A number of wind
speed sensors operating on a variety of
physical principles are available com-
mercially. However, the rotational cup
and propeller anemometers are the
most commonly employed wind speed
sensors. The more esoteric designs are
generally used in very specialized
studies.
Wind Vanes - Wmd-direction-mea-
suring sensors are operated by wind
exeitmg pressure on a surface that
rotates about a fulcrum The standard
vane measures only the horizontal wind
direction, but the bidirectional vane is
free to move through 360 degrees
horizontally as well as ±50 degrees or
more from the horizontal. The shape
and design of the vane surface may vary
with the manufacturer (Figure 2.3.1).
Combination Wind Sensors - Two
types of sensors incorporate both
Figure 2.3.1. Typical wind vanes by Climet Instruments, Inc., (left) and R. M. Young Company (right).
6
-------
Figure 2.3.2. Propeller wind vanes by Bendix Friez Instrument Division (upper left], by R. A/7. Young Company »
(upper right) and by Meteorology Research, Inc. (lower).
-------
direction and speed measuring capabil-
ity in a single mechanical device. The
propeller vane sensor measures two-
dimensional flow (Figure 2.3.2), and the
propeller bivane sensor measures
three-dimensional flow.
Wind component anemometers can
be used to determine the wind speed
and direction(s), using simple trigono-
metry. These instruments include the
x,y,z prop (often called u, v,w), the sonic,
the vortex shedding, and the ion flow
anemometers (see Table 2.3 1).
Transducers - Transducers convert
the weather element measured by a
sensor into an electrical or recordable
signal. This discussion will be limited to
transducers used in rotation-type wind
speed sensors and vane-type direction
sensors.
The rotary motion of cups and
propellers is most often converted to a
voltage or a frequency. Both alternating-
current generators and direct-current
generators are used, but the latter are
used more often. Frequency-type devices,
sometimes called light choppers, have
advantages in that they are almost
frictionless, operate at lower wind
speeds, and produce signals that can be
transmitted without loss over long
distances Typically, this type of trans-
ducer is made to interrupt the light of an
LED (light-emitting diode) at a rate of 1
to 132 times for each rotation of the
sensor. Units that interrupt the light
once per revolution of the anemometer
shaft are usually used to measure wind
run thus producing longer time averages.
A single sealed-m-glass switch, used in
combination with a magnet on the shaft
of the anemometer, is also a frequency-
type transducer for wind speed or wind
run Another frequency-type device
now used in anemometers is the Hall-
effect generator which uses the elec-
trical polarization of a conducting plate
moving through a magnetic field
Mechanical anemometers with wiper-
type contacts, still in use for climato-
logical studies, are attractive because of
their simplicity but are of limited use in
pollution studies
Wind direction transducers are of
several basic types wiper or sealed
contact switches, single or double
potentiometers, and DC or AC syn-
chronous motors. Some of the most
sophisticated transducers operate on
the principle of capacitance with
outputs in frequency form. Wire-wound
and carbon-deposited potentiometers
are used most frequently
Signal Conditioners - In the most
elementary electronic systems, the
signal from the variable being measured
is transmitted from transducer to signal
conditioner to readout device, with
power applied at some or all of these
steps. The signal conditioner converts
the transducer output into an electrical
quantity suitable for the proper operation
of the readout equipment The signal
conditioner may vary from a simple
resistance network or impedance-
matching device to an amplifier, an
analog-to-digital converter, or, as in the
case of the photoelectric speed trans-
ducer, a frequency-to-voltage converter.
Devices that convert from 360 degrees
to 540 degrees of wind direction and
devices that provide "average" wind or
wind with "time-weighted" average,
are signal conditioners Sealers that
compress, expand, or change the
transducer signal from electrical to
engineering-equivalent units are also
signal conditioners.
Readout Devices - Readout devices
are classified as mechanical or electrical.
Few of the mechanical, drum or disc
drive recorders are now used for
recording wind. Electrical recorders are
(1) galvanometric, direct-writing, strip-
chart type, (2) null-balance, servo-
operated, (may be potentiometric), DC
bridge, or force-balance servo type; (3)
event or operations; (4) digital; and (5)
other special types
Direct writing (D'Arsonnal movement)
galvanometric recorders with strip
chaits are used most frequently for
wind measurements They are used
because of their reliability and their
speed of response. The chart drive
mechanisms are available as hand
wound, spring driven, battery powered,
or AC powered It is important to select a
recorder in which the damping charac-
teristics of the galvanometer do not
degrade the response of the sensor and
are in compliance with the frequency
response required in the study
Most galvanometric recorders used
for wind measurements are continuous
curve-tracing recorders The chopper-
bar type, designedtomakean imprint on
pressure-sensitive paper each time the
meter pointer is clamped against a
sharp-edged platen, produces a record
that is noncontmuous. Imprints are
usually made every 2 seconds, so in
rapidly varying winds the record appears
very scattered Some manufacturers of
meteorological instruments supply a
recorder with a built-in signal conditioner,
that reduces the scatter through the use
of either unspecified or selective time
constants on the response of the
galvanometer This conditioner must be
viewed critically if calculation of the
standard deviation of the measurement
is an objective in analyzing the record
Potentiometric recorders, used fre-
quently for wind measurement, may not
have the rapid response characteristics
of galvanometric recorders As a null-
balance device, the input impedance is
Table 2. 3.1.
Physical
principle
Rotation
Pressure
Cooling
Sound
Vortex shedding
Ion flow
Types of Anemometers and Their Operating Principles
Anemometer type
Vane-oriented propeller
Bivane-oriented propeller
Fixed propellers
Cups
Plate
Tube
Br/dled cups
Hot wire
Hot thermopile
Hot film
Sonic
Vane-oriented shape
Transport
Measurement
Horizontal speed
Total speed
Three-dimensional components
on perpendicular axes
Horizontal speed
Horizontal speed
Horizontal speed
Horizontal speed
Directional flow components
Directional flow components
Directional flow components
Directional flow components
Horizontal speed
Horizontal speed
-------
very high when the stylus is at equili-
brium, therefore errors due to electrical
loading of the sensor output and errors
due to voltage drops in long signal leads
are minimized.
Multipoint potentiometnc recorders,
which sequence through a series of
inputs or scheduled cycles, have no
value in recording wind. Instantaneous
samples of wind direction or wind speed
once every few minutes are useless in
most air pollution investigations. Curve-
tracing potentiometnc recorders are
useful, and are available with charts
that require ink or with inkless charts
that operate with a heated stylus.
Digital systems are becoming more
popular for logging and displaying wind
data. They come in a wide range of
designs and are usually elaborate
systems that include analog to digital
conversion, integration over specified
intervals, and memory or storage
capability. The output format of these
systems should be compatible with the
computer used for data processing.
2.3.1.1 Wind Sensor Characteristics-
Cup anemometers, propeller anemom-
eters, and wind vanes continue to be the
best sensors for measuring wind speed
and determining wind direction over a
broad range of applications. Other types
of sensors have unique characteristics
for special projects or research This
discussion is limited to the character-
istics of devices used most frequently in
operational programs.
Cup anemometers arecomplexshapes
where the net torque (lift greater than
drag) causes a rate of rotation roughly
proportional to wind speed They
respond to any horizontal wind direction
which is an advantage but they are also
responsive to the vertical component of
the wind. In turbulent flow, the output
(average speed) may be closer to the
total speed than to the presumed hori-
zontal component (MacCready, 1966).
The design of the anemometer cup
assembly and the material from which it
is constructed are important in deter-
mining durability, linearity, starting
threshold, and dynamic response of the
instrument. Three conical cups give
better performance than hemispherical
cups (Figure 2.3 3) and are preferred
over the older four-cup design The ratio
between cup diameter and cup wheel
diameter influences the calibration
curve (Gill, 1973) Starting threshold is
defined as the lowest wind speed at
which the rotating cups meet the
accuracy specification (Lockhart, 1970)
J
Figure 2.3.3. Typical three-cup anemometers featuring conical cups. Manufacturers are Belfort Instrument Company
(upper left), Climet Instruments. Inc. (upper right) and C. W. Thornthwaite Associates (lower).
-------
In order to define the smallest eddy
size to which cups will be responsive, a
dynamic characteristic known as the
distance constant must be known. This
is determined in a wind tunnel by
measuring the time for the cups to reach
63 percent of the tunnel speed after
being released from a nonrotating
condition. The distance constant in
meters may be expressed as a time
constant in seconds at a given wind
speed by dividing by that wind speed in
meters per second.
Propeller anemometers, especially
helicoid types, are primary sensors with
rotation rates linearly proportional to
the wind speed over a wide speed range
(Gill, 1973). Propeller anemometers
must be oriented into the wind. The
error from the failure of the vane to
perfectly orient the propeller is small
since propellers have a nearly cosine
response; i.e., the propeller turns at a
rate almost directly proportional to the
wind component parallel to its axis. Like
the cup anemometer, the propeller
anemometer is a first-order, nonoscilla-
tory system whose dynamic charac-
teristics can be described by the
distance constant referenced above.
Fixed axis propeller anemometers are
designed to measure two or three
components of wind simultaneously at
a point in space They represent a
special type of propeller anemometer
for the direct measurement of turbulence
(Gill, 1975). Three helicoid anemometers
in an orthogonal array measure the
wind for the axes U-V-W. Each propeller
turns at a rate almost proportional to the
wind component parallel to its axis.
Cosine response, although not yet
perfected, is critical in this equipment;
but this device does not have the static
balance limitations of the bivane (a two
axis wind vane), which is also used for
turbulence measurements. Under condi-
tions of ram, snow, and heavy dew, the
bivane imbalance may produce un-
acceptable errors.
Wind vanes have a damped and
oscillatory motion. This characteristic
second-order response is the result of
such factors as weight of materials,
shape and size of vane, and location and
weight of counter-balance. One indica-
tion of the performance of a vane is the
starting threshold. As described by
Fmkelstem (1981), this is the lowest
speed at which a vane released from a
position 10 degrees off the centerlme in
a wind tunnel moves to within 5degrees
of center
There are several dynamic character-
istics, identifiable as constants, that can
be used to define the performance of a
wind vane (MacCready, 1965, and
Wiermga, 1967) in response to a step
function. These include damping ratio,
damped wavelength, undamped wave-
length, and delay distance. Undamped
wavelength is used in determining the
dynamic response of a wind vane to
sinusoidal wind direction fluctuations.
Damping ratio is a constant that is
dimensionless and independent of wind
speed. It is calculated from the relative
amount of overshoot on each of two
successive half cycles of a decaying
oscillation If the ratio of two successive
swings is designated as Q, then the
damping ratio £" is calculated from the
equation:
ml
n
sj
For most operational programs, a
damping ratio of 0.4 or greater is
recognized as satisfactory.
The damped wavelength is easily
determined by multiplying the time for
one complete oscillation by the wind
speed in a tunnel
Delay distance is another observed
measure of the response of a vane to a
step change. The time required for a
vane to reach 50 percent of the distance
from an initial displacement of 10
degrees toward the centerline on the
first swing is the delay time. This is
multiplied by the tunnel speed to obtain
the delay distance.
2.3.1.2 Wind Data Requirements-Any
data requirement should be expressed
in the context of all applications for
which the data may be used. Wind data
are used in environmental monitoring
for source location, transport and
dilution modeling and as a diffusivity
indicator. The principal specifications
are dynamic range (most important is
the threshold) and dynamic performance
(distance constant for speed and damping
ratio and delay distance for direction) It
is also important to specify the averaging
time and method in order to judge the
adequacy of the measuring and recording
system. For most atmospheric dispersion
studies, a starting speed of 0.5 m/s or
less is appropriate for both vanes and
anemometers. Wind vanes should have
a damping ratio of 0 4 or greater and a
delay distance of 5 m or less. Anemom-
eters should have a distance constant of
5 m or less. For climatological studies,
less sensitive instruments may be used.
Averaging for wind speed may be
done by scalar methods (dilution) or
vector methods (transport) and should
represent one hour Wind direction
should be averaged by obtaining the
resultant vector direction for the hour.
Sigma theta (standard deviation of the
wind direction) should represent 3-10
minutes if stability categories are to be
selected but should represent the hour
when the preferable direct calculations
are made for diffusion (Strimaitis,
1981).
2.3.2 Procurement
In purchasing a wind measuring
system, follow the general guidelines
advanced in Section 2.2.1, "Instrument
Procurement." It might be possible in
research projects to spend more for
instruments with the best specifications,
or instruments of most recent design;
but in operational programs, only the
field-tested and time-proven instruments
with known performance records should
be purchased.
Caution should be used in purchasing
components as opposed to systems,
especially if the procurement of wind
equipment is part of an installation
involving other instruments. It is
important to match the dynamic charac-
teristics of the wind sensors, and to
match the electrical characteristics of
the transducers with the readout
device. The omission of such a minor
consideration as mounting hardware
could delay completion of an installation.
In digital systems, special attention
should be given to sampling and
averaging times as well as to instan-
taneous as opposed to integrated
values.
2.3.3 Acceptance Testing
Follow the general guidelines set
forth in Section 2.22, "Acceptance
Testing " Be certain that the supplier
has provided all calibration certification
data, including curves and specifications.
Be particularly sure that there is
provided a table or formula relating rate
of rotation of the anemometer shaft (or
frequency for light-chopper sensors if
the number of pulses per revolution is
also given) to wind speed (or output
voltage given a voltage to speed range
relationship). This will usually relate to a
nominal propeller or cup. The specific
propeller or cup assembly should have a
permanent identification code (serial
number). In those cases when wind
tunnel calibration data are provided,
this identification is required.
The acceptance test for the direction
vane should include a measure of how
well the sensor represents the relative
poSiition of the vane to the sensor
housing Four points 90 degrees apart
can be easily bench tested by drawing
perpendicular lines crossing at the vane
rotation axis and holding the vane shaft
parallel to the lines Also check to see
that the manufacturer's method of
coping with the discontinuity between
10
-------
360 degrees and 001 degrees performs
as specified
2.3.4 Calibration
Refer to the general notes in Section
2.2 3, "Calibrations."
Bench Calibrations- Specific dynamic
response characteristics such as thresh-
old speeds, damping ratios, delay dis-
tances and distance constants can only
be checked in a wind tunnel. Instru-
ments should be returned to the manu-
facturer or a properly equipped wind
tunnel facility for major calibration
The system manual will generally
have calibration methods specified For
wind speed the usual signal conditioner
will require a DC voltage input which
represents some output value in volts
and wind speed units or it will require a
frequency to represent some output
value m volts and wind speed Full scale
is often used Sometimes the sensor is
used in calibration by turning the shaft
at a known rate of rotation Sometimes
the system will have some level of built-
in "calibration" which tests some parts
of the system Use the manufacturer's
method for bench calibration Other
methods may also be used but they
should supplement the manufacturer's
instruction rather than replacing it, at
least initially
For wind direction sensors with
potentiometers the manufacturer may
specify substitution of known resistance
for various angles relative to the sensor
housing Positioning of the connected
sensor vane or built-in substitution
resistance may also be specified. Use
the method described by the manufac-
turer augmented by whatever additional
tests will add useful information. These
might include a measurement of the
angular size of the "open space" where
the potentiometer winding is left out to
avoid shorting by the wiper, a measure-
ment of hysteresis by approaching
known positions from both clockwise
and counterclockwise rotation, and
testing any discontinuity avoiding
operation (540 degree range types) Any
adjustments, mechanical or electrical,
must be in accordance with the manu-
facturer's instruction (either from the
instrument manual or direct commu-
nication) to avoid damage or voiding the
warranty
The common point of sensor deterio-
ration is in the bearings of the bearing
shaft assemblies of both the speed and
direction sensors When bearings start
to fail as they inevitably will do, the
starting threshold will increase. The low
wind speed data are most important for
air quality applications because the
initial dilution is small and the concen-
tration therefore is high The deteriora-
tion of bearing performance is also hard
to see in the data until the problem
becomes quite serious. For example the
data cannot discriminate (by usual
inspection) between a system with an
acceptable threshold of 0 5 m/s and
one that has deteriorated to 0 7 m/s
There is a method by which the
bearing condition can be measured
(Lockhart, 1978). The use of an instru-
ment called a torque watch is not new
but it is not widely applied as yet It is a
measurement which requires consider-
able care and experience For wind
speed the cups or propeller must be
removed to avoid wind influencing the
measurement. The torque watch adapter
is attached to the shaft and the torque
watch is inserted into the adapter and
rotated until the shaft begins to turn
The maximum torque from repeated
tests is the starting torque. If manufac-
turers do not have a torque specification
for their sensor, they should be asked to
provide one
Orientation
by
Local Apparent Noon
Date of Observation
Observer(s)
Day/Time of WWV Synchronization-
Station Name
Station Longitude
Ref
Standard Meridian
Difference in degrees
Difference in time (x4)
Mean Time at Local Apparent Noon
Ref.
True Noon Time date
mm
min.
-sec.
.sec.
_hr.
mm.
.sec.
Sitings True Heading
Figure 2.3.4. Example of solar noon orientation form and tables.
Target
11
-------
Table 169.
Ephemeris of the Sun,
All data are for O' Greenwich Civil Time in the year 1950. Variations of these data from
year to year are negligible for most meteorological purposes, the largest variation occurs
through the 4-year leap-year cycle. The year 1950 was selected to represent a mean condition
in this cycle.
The declination of the sun is its angular distance north ( + ) or south (—) of the celestial
equator.
The longitude of the sun is the angular distance of the meridian of sun from the vernal
equinox (mean equinox of 1950.0) measured eastward along the ecliptic.
The equation of time (apparent — mean) is the correction to be applied to mean solar time
in order to obtain apparent (true) solar time.
The radius vector of the earth is the distance from the center of the earth to the center of
the sun expressed in terms of the length of the semimajor axis of the earth's orbit
1 U. S. Naval Observatory, The American ephemeria and nautical almanac for the year 1950, Washington,
1948.
EPHEMERIS OF THE SUN
Date
Jan 1
jan. i
9
13
17
21
25
29
Mar. 1
5
9
13
17
21
25
29
May 1
5
9
13
17
21
25
29
July 1
5
9
13
17
21
25
29
Sept 1
5
9
13
17
21
25
29
Nov. 1
5
9
13
17
21
25
29
Decli- Longi-
nation tude
•
-23
22
22
21
20
20
19
18
- 7
6
4
3
1
— 0
4. 1
3
+ 14
16
17
18
19
20
20
21
+23
22
22
21
21
20
19
18
+ 8
5
4
2
+ 1
- 0
2
-14
15
16
17
18
19
20
21
* •
4 280
42 284
13 288
37 292
54 296
5 300
9 304
8 308
53 339
21 343
48 347
14 351
39 355
5 359
30 3
4 7
50 40
2 43
9 47
11 51
9 55
2 59
49 63
30 67
10 98
52 102
28 106
57 110
21 113
38 117
50 121
57 125
35 157
7 161
37 165
6 169
34 173
1 177
32 181
6 185
11 217
27 222
38 226
45 230
48 234
45 238
36 242
21 246
f
1
5
10
14
19
23
27
31
51
51
51
51
50
49
47
44
4
56
48
40
32
23
14
4
36
24
13
2
51
40
29
19
59
52
45
38
32
26
21
16
59
0
1
2
4
6
8
11
Equation
of time
m. •.
- 3 14
5 6
6 50
8 27
9 54
11 10
12 14
13 5
-12 38
11 48
10 51
9 49
8 42
7 32
6 20
5 7
+ 2 50
3 17
3 35
3 44
3 44
3 34
3 16
2 51
- 3 31
4 16
4 56
5 30
5 57
6 15
6 24
6 23
- 0 15
+ 1 2
2 22
3 45
5 10
6 35
8 0
9 22
+16 21
16 23
16 12
15 47
15 10
14 18
13 15
11 59
Radius
vector
0.98324
.98324
.98333
.98352
.98378
.98410
.98448
.98493
0.99084
.99182
.99287
.99396
.99508
.99619
.99731
.99843
1.00759
1.00859
1.00957
1.01051
1.01138
1.01218
1.01291
1.01358
1.01667
1.01671
1.01669
1.01659
1.01639
1.01610
1.01573
1.01530
1.00917
1.00822
1.00723
1.00619
100510
1.00397
1.00283
1.00170
0.99249
.99150
.99054
.98960
.98869
.98784
.98706
.98636
Date
Feb. 1
5
9
13
17
21
25
Apr. 1
5
9
13
17
21
25
29
June 1
5
9
13
17
21
25
29
Aug. 1
5
9
13
17
21
25
29
Oct. 1
5
9
13
17
21
25
29
Dec. 1
5
9
13
17
21
25
29
Decli-
nation
• i
-17 19
16 10
14 55
13 37
12 15
10 50
9 23
+ 4 14
5 46
7 17
8 46
10 12
11 35
12 56
14 13
+21 57
22 28
22 52
23 10
23 22
23 27
23 25
23 17
+ 18 14
17 12
16 6
14 55
13 41
12 23
11 2
9 39
- 2 53
4 26
5 58
7 29
8 58
10 25
11 50
13 12
-21 41
22 16
22 45
23 6
23 20
23 26
23 25
23 17
Longi-
tude
* '
311 34
315 37
319 40
323 43
327 46
331 48
335 49
10 42
14 39
18 35
22 30
26 25
30 20
34 14
38 7
69 56
73 46
77 36
81 25
85 15
89 4
92 53
96 41
128 11
132 0
135 50
139 41
143 31
147 22
151 14
155 5
187 14
191 11
195 7
199 5
203 3
207 1
211 0
214 59
248 13
252 16
256 20
260 24
264 28
268 32
272 37
276 41
Equation
of tune
m.
-13
14
14
14
14
13
13
- 4
3
1
- 0
+ 0
1
1
2
+ 2
1
1
+ 0
- 0
1
2
3
- 6
5
5
4
4
3
2
1
+10
11
12
13
14
15
15
16
+11
9
8
6
4
2
+.0
•.
34
2
17
20
10
50
19
12
1
52
47
13
6
53
33
27
49
6
18
33
25
17
7
17
59
33
57
12
19
18
10
1
17
27
30
25
10
46
10
16
43
1
12
17
19
20
39
Radius
rector
0.98533
.98593
.98662
.98738
.98819
.98903
.98991
0.99928
1.00043
1.00160
1.00276
1.00390
1.00500
1.00606
1.00708
1.01405
1.01465
1.01518
1.01564
1.01602
1.01630
1.01649
1.01662
1.01494
1.01442
1.01384
1.01318
1.01244
1.01163
1.01076
1.00986
1.00114
1.00001
099888
.99774
.99659
.99544
.99433
.99326
0.98604
.98546
.98494
.98446
.98405
.98372
.98348
.98334
Figure 2.3.4 (continued) (Reproduced with permission)
12
For wind direction the procedure is
the same but because the sensor
usually has a potentiometer attached to
the shaft, the range of the torque
required to tuin both the shaft and the
potentiometer is higher than for wind
speed This usually requires a second
torque watch with a higher operating
range For sensors where the vane
cannot be easily or completely removed,
the test must be done in a still room or
enclosure with the sensor shaft vertical
or perfectly balanced
It should be noted thatan experienced
instrument technician can detect the
condition of a bearing assembly by
"feel" with perhaps as much sensitivity
as the torque watch. This talent should
be used as a routine method for
checking bearing condition for those
programs which do not require docu-
mentation of data quality and do not
have the torque watch capability
Others should develop the capability
In general field calibrations should be
as much like the bench calibrations as
the field conditions and available
equipment will allow, In addition,
however, the field calibration must
include those additional tests which
relate to the system being installed Two
examples of the additional tests are the
orientation of the direction vane or fixed
propeller assembly with respect to
TRUE NORTH and the alignment of the
sensors in the vertical operating position
Field calibrations should be done at
least every six months
2.3.5 Installation
Follow the general guidelines set
forth in Section 226, "Preparation for
Field Installation," and in Section 3,
"Methods for Judging Suitability of
Sensor Siting
Wind vanes require orientation to
true north Some manufacturers elec-
trically orient the transducer in the
housing and identify the northfacing
side of the housing with an engraved
mark, or they provide a flat surface for
sighting toward north Other manufac-
turers supply, as an option, a mounting
jig for keeping the vane in proper
alignment for orientation
The most accepted procedure for
determining true north involves shooting
the North Star with a first-order
theodolite Any textbook or handbook on
land surveying will describe the tech-
nique and will contain all the necessary
tables Best results may come from
hiring a registered land surveyor to
establish true north
Another acceptable method of estab-
lishing true north is by the location of
the sun at true solar noon (see Figure
2.3 4)
-------
T,sn =(12:00:00) +
[B(Long-15n) -A]
where TtSn = local standard time of
true solar noon
A = "Equation of Time"from
"Ephemens of Sun"
tables (List, 1971)
B = 4 minutes per degree of
longitude
Long = local longitude
n = number of time zones
from Greenwich
n =5 Eastern
n =6 Central
n = 7 Mountain
n =8 Pacific
n =9 Yukon
n = 10 Alaska/Hawaii
n =11 Bering
Alignment in the vertical is equally
important. Studies have indicated that
vertical misalignment of 1 degree may
yield data errors of 10percentorgreater
in measurement of turbulent parameters
(Pond, 1968; Deacon, 1968; and Kraus,
1968) Vertical alignment should be
established with a good carpenter's
level or torpedo level at two points 90°
apart in the horizontal
2.3.6 Operation of a Wind
Measuring System
An operational check and calibration of
the wind system is recommended after
the installation has been completed and
at intervals of at least every six months
after operations begin
The use of tic-marks on chart rolls is
very effective to indicate time of visit or
some operation These are often made
with event pens purchased with the
recorders Whether manually or auto-
matically actuated, the event pens can
be energized to produce distinguishing
marks corresponding to different times
If the sensor is readily accessible,
another simple means of establishing a
tic-mark is to rotate the vane 360
degrees and note the time on the strip
chart. Where strip chart recorders are
used, time checks every 24 hours are
desirable
Of course, documentation relating to
date, time, etc , should be entered on
each strip chart when it is installed and
removed. The amount of data to be
reduced or retrieved from properly
annotated and logged strip charts will
depend on program objectives in con-
junction with the amount of the strip
chart data needed All operational
activities should be logged including
those written on strip charts It is not
possible to log too much information as
long as it is consistent The problem is
always with too little information,
particularly time and date
2.3.7 Preventive Maintenance (PM)
Follow the general guidelines in
Section 2.2.5, "Operational Checks
and Preventive Maintenance." Physical
checks of the equipment should be
made as often as possible, and at least
monthly These checks include an
examination of the sensors and the
readout equipment. At locations where
pollution is heavy, it may be necessary
to routinely change the bearings in both
the anemometer and vane housing
Oilless bearings should never be oiled.
Bearings can be checked with a torque
watch (see Section 2 3.4).
Recording equipment with pens and
ink should be cleaned on a regular
monthly schedule If the recorder is
equipped with disposable felt tip pens,
careful monitoring is recommended
The fluctuations of the wind, as recorded
by the pen, result in an average pen life
of 2 or 3 days, so there may be a loss of
data
Light freezing ram with little wind is
the most detrimental condition for cups
and especially for propellers It has been
found that a light spray coating of Pam®
(a household substitute for grease) or an
equivalent non-sticking spray, helps to
retard the formation of ice especially on
propeller anemometers For extremely
low temperature conditions, the use of
external heat lamps or heater strips is
helpful Internal heaters, now available
with some wind sensors, keep bearings
free but have low wattage and are of
limited use in severe conditions. Spare
parts for anemometers and vanes
should include replacement cup assem-
blies or propellers, vanes, and bearings
A log should be kept of all PM activities
2.3.8 A udit Procedures
Refer to Section 2.24, "Audits," for
general guidelines A complete systems
audit is necessary near the beginning of
any wind measuring program Particular
attention should be paid to sensor
siting
Performance audits may include but
should not be limited to repeating the
field calibration on the sensors and
associated cabling while they are still in
place (see Section 2.3.4) The wind
speed calibration procedure may in-
volve rotation of the anemometer with a
constant speed motor. When a syn-
chronous motor is used for an audit
there needs to be a methodfor measuring
the rate of rotation and there needs to be
AC power, usually 60 Hertz It may not
be possible to get the power at the
sensor location or even the site location
It also may be troublesome to measure
the rate of rotation if the transducer is a
generator since generators often have
electrical noise and the averaging of
voltage samples is less accurate than
counting revolutions in an accurately
known time gate.
An alternative method is to challenge
the anemometer shaft with at least two
known average rates of rotation The
rates should represent common low
wind speeds; somewhere in the ranges
1 -2 m/s and 5-7 m/s for example One
should audit the low end of the speed
range because that is where the data
are most important for air quality
applications and where error estimates
are most valuable. It isalsoeasytoknow
the accurate average rate of rotation by
counting revolutions (by switch or
manually) over a period of time signifi-
cant to the data period (2 to 1 5 minutes
typically) where the number of rotations
is large enough that the uncertainty
over the time period is less than one
percent. The elapsed time measurement
will always be accurate enough so that
the average rate of rotation will be
within one percent of the reading Since
an average is used, the motor driving
the shaft does not need to be syn-
chronous and a DC motor driven by a
battery will give freedom from the
commercial power requirement It is
important to have nearly or exactly the
same time interval used by both the
motor drive and the output reading The
only part of the system challenged by
this method is the measurement and
recording of the rate of rotation expressed
as wind speed. There is no test of the
rate of rotation of the sensor as a
function of wind speed, that relationship
is either given by the manufacturer as a
nominal value or was determined by a
wind tunnel test. If a wind tunnel test
was made initially and the sensor is
identified by serial number and shows
no sign of deformation in shape, it is
reasonable to assume the sensor rate of
rotation vs. wind speed relationship has
not changed It is not necessary to wind
tunnel test every six months or year If,
on the other hand, a wind tunnel test
has not been done and the program
requires documentation of accuracy, a
wind tunnel test should be done or an
audit with a collocated transfer standard
(CTS) The CTS has the advantage of
challenging the entire system in naturally
turbulent flow The CTS must be well
sited to measure representative flow
without mutual interference A period of
at least 1 2 hours (half of a diurnal cycle)
should be used to obtain as much range
comparison as is practical
The wind direction audit procedure
should concentrate on output when the
vane is aimed at known sites The
direction to and from a distant landmark
should be determined by independent
methods Map location is best if a good
map is used (USGS, airways or equiva-
13
-------
lent) and the site can be accurately
located. Another method is to use a
north finding technique such as described
in Section 2.3.5 and to measure the
angle between north and the landmark
with a theodolite or a sextant. The
direction vane should be taped or
clamped in the direction desired so a
steady reading can be accurately made.
If a sigma theta measurement is
included in the system, this would be a
good time to verify zero.
If a CIS is used, the comparison of
averages will be subject to orientation
differences (bias) and perhaps others
caused by different dynamic perform-
ance characteristics and different aver-
aging methods. When properly exposed,
the CIS gives the best estimate of
accuracy when operating in a naturally
turbulent flow. Collocated sensors are
within 10 m of each other in the
horizontal and one meter in the vertical
according to Hoehne (1971). Data taken
while the wind is within ±30° of the line
between the sensor being audited and
the CIS should not be used.
2.4 Temperature
Measurements
2.4.1 Introduction
Temperature is not simply an isolated
piece of meteorological data but is used
in the measurement and determination
of a number of other atmospheric
parameters such as vertical temperature
gradient (stability), relative humidity,
and gaseous pollution concentrations.
Several types of sensors and recorders
are used routinely in a variety of
combinations to acquire temperature
data The following sensors are used
most often for environmental measure-
ments
Linear Thermistors - Thermally sen-
sitive resistors or thermistors are
electronic semiconductors that are
made from certain metallic oxides The
resistance of a thermistor varies inverse-
ly with its absolute temperature. Linear
thermistors are a composite of two or
more thermistors and fixed resistors
designed to produce a linear response to
a wide temperature range. In system
configuration, the thermistor is connect-
ed to a bridge circuit or some suitable
signal conditioning circuit. When a low
excitation voltage is applied to the
thermistor, the output from a bridge
circuit is a voltage that varies directly
with the temperature of the sensor.
Linear thermistors are particularly well
suited to remote sensing applications
because of their large resistance change
to temperature change ratio. Coeffi-
cients on the order of 125 Q/°C are
common (Lockhart and Gannon, 1978),
reducing the impact of lead resistance
errors. A lead resistance of 12.5 O
would be required to produce a 0.1 °C
temperature measurement error. The
thermistor or thermistor composite may
be packaged as a glass covered bead or
potted in a stainless steel sheath. The
latter is usual and best for monitoring
applications
Resistance Temperature Detectors
(RTD) - The RTD operates on the
principle that the electrical resistance of
a pure metal increases with temperature.
Although RTD sensors are made using
silver, copper, and nickel wire, platinum
wire is the best material because of its
superior linearity and stability charac-
teristics, high sensitivity, and resistance
to corrosion. The RTD probe is encased
in a protective stainless steel housing
composed of an insulating core wrapped
with platinum wire The resistance of
the wire is measured by a bridge circuit
in a signal conditioner. The RTD
operates at a much lower resistance to
temperature ratio than the linear
thermistor. RTD's used most often m
meteorological work have a coefficient
of resistance change on the order of 0.4
ohms per degree Celsius, and are more
sensitive to lead resistance errors than
are thermistors. "Three wire" and "four
wire" systems are configurations that
automatically compensate for lead
resistance.
Liquid-ln-Glass-Thermometers - Most
of these thermometers are 10-1 /2-mch
glass tubes with a uniform capillary
bore and a liquid reservoir or bulb at the
bottom. Thermometers are usually
mounted on a stainless steel back with
the bulb protruding to allow free air
circulation Volumetric expansion and
contraction of the operating liquid
provides the measure of temperature
The liquid is usually mercury but maybe
a mixture of ethyl alcohol and red dye.
Thermometers filled with alcohol are
called spirit thermometers Though
inherently less accurate than a mercury
thermometer, the spirit thermometer
must be used where extremely cold
temperatures are anticipated because
mercury will freeze at -38°C
Thermocouples - Because of its
complex circuitry and somewhat limited
accuracy, the thermocouple is not a
popular operational meteorological
sensor, although its vast capabilities
make its use in other temperature
measurement applications quite com-
mon. The thermocouple operates on the
principle that when two different metals
are joined, a small voltage with a
temperature-dependent magnitude is
generated By comparing this voltage to
the voltage generated by the same two
metals joined at a temperature-con-
trolled reference junction (ice point), the
ambient temperature can be determined.
Deformation-Type Thermometers -
This category of thermometer depends
upon the thermal expansion and con-
traction of metals or liquids. The
response to temperature changes is
actual physical movement of the sensor
mechanism, which is mechanically
transmitted through a system of levers
to a thermograph, where a continuous
temperature trace is produced. The
most common deformation-type thermom-
eters are the bimetallic thermometer,
where vastly different thermal expansion
properties of two mated metal strips
produce deformation; the Bourdon tube
thermometer, where a coiled, hollow,
spirit-filled, flexible metal tube changes
shape with the thermal expansion and
contraction of the internal fluid; and the
mercury-in-steel thermometer, where
the pressure produced by thermal
expansion of the mercury is transmitted
to a Bourdon tube. With the coming of
age of the quick-response electronic
recording sensor system, slow-reacting
thermographs with their requirement of
manual data reduction are slipping in
popularity.
2.4 1.1 Sensor Charactenstics-/4c-
curacy - A high-q uality mercury thermom-
eter properly calibrated will provide
temperature data of sufficient accuracy
for most atmospheric measurement
programs. Over the years, the capillary
bore will tend to contract slightly,
causing the zero point to rise. Because
the thermal response of mercury is
essentially linear, this error is correctable
through calibration. The most serious
errors incurred in using a mercury
ther mometer are usually due to improper
exposure and/or observer error (e.g.,
panallax error in reading the tempera-
ture) Typical commercial platinum RTD
sensors follow the platinum character-
istic that represents a resistance
accuracy of ±0 26°C at 0°C and
±0.38°C at 50°C Linear thermistors
are available commercially with an
accuracy of +0.15°C over the range of
temperatures normally encountered in
environmental measurements (Lockhart
and Gannon, 1978). Ordinary commer-
cial thermocouples are only accurate to
±1DC (Wang, 1975) Accuracy of a
measurement system is limited more by
such factors as sensor exposure,
improper coupling, and signal inter-
ference than by accuracy of the sensor.
For air quality monitoring applications,
an accuracy (including coupling error)of
better than 0 5°C is easily achievable
and adequate (Strimaitis, 1981)
Linearity - There are no significant
linearity problems with any of the
electronic or liquid-m-glass sensors
-------
within the extremes of tropospheric
temperature. Some linearity problems
occur in the deformation-type sensors,
particularly at extreme temperatures.
Response Time - A certain amount of
caution must be exercised in assessing
the value of quick response time in
making stationary environmental tem-
perature measurements. At the warmest
time of the day, a continuous Eulenan
temperature trace may show variations
of up to 1 °C in a 30-second period. Care
must be taken in evaluating instanta-
neous temperature data from a thermis-
tor or RTD, both of which may have time
constants as short as 5 seconds or less
For both of these sensors, temperature
values should be the average of a series
of readings taken over a minimum
period of 1 minute to avoid "noise"
errors. Usually the stainless steel
sheath which protects the sensitive
element for field monitoring use adds to
the time constant mechanically thereby
providing both ruggedness and a more
useful time constant
Precision - Most good quality electrical
temperature systems used for moni-
toring in the atmosphere will have an
output with better resolution and
precision than accuracy The application
of the data may take advantage of this
fact. For example, a temperature dif-
ference between two levels on a tower
may (and should) be calibrated around
the precision of each sensor yielding an
accuracy of a difference on the order of
the precision of the sensor (0.1 °C for
example). This statement must be
qualified to either ignore the coupling
error or assume each sensor will suffer
the same coupling error and, therefore,
not influence the difference measure-
ment. The latter is a reasonably good
assumption when identical radiation
shields are used.
The accuracy of the temperature
differential measurement system ex-
cluding coupling errors may be as good
as the resolution or precision if an effort
is made to either adjust the circuits or
correct the data to achieve it The
nominal values quoted above reflect an
interchangeability with circuits adjusted
to the nominal values given in the
manufacturer's manual Given sensor
stability, the accuracy will be as good as
the calibration.
Durability - All of the electronic
temperature sensors are fairly rugged
and stable. Liquid-in-glass thermom-
eters obviously require careful handling.
The deformation thermographs also
require delicate handling as any physical
damage will alter the calibration.
2.4.1.2 Solar Radiation Shields-The
most serious problem encountered in
environmental thermometry is radiation
error, which can amount to several
degrees Celsius at midday. Early attempts
to combat radiation error took the form
of instrument shelters, such as the all-
wood, louvered, double-roofed cotton-
region type. Much better results in
reducing radiation error have been
achieved with aspirated radiation
shields (Figure 2.4.1)
Naturally Ventilated Radiation Shields
- The most common of these radiation
shields employs a vertically mounted
sensor with a 360° ventilation exposure
The housing is topped with single or
multiple polished domes for solar
radiation shielding and has lower
circular shields to block terrestrial and
reflected radiation. The vane oriented
radiation shield directs the ambient
wind to the temperature sensor. A
swivel-mounted, white, horizontal,
double-walled tube is equipped with a
large vane that keeps the housing
opening pointed into the wind However,
these shields give large errors when the
wind is light. Both of these naturally
ventilated radiation shields will accept a
variety of electronic temperature sensors,
including the thermistor, thermocouple,
and RTD
Mechanically Aspirated Radiation
Shields - This type of radiation shield is
used where power is readily available to
AES Stevenson Screen
IS
CCIW Parallel Pie Plate Shield
AES Parallel Plate
Teledyne Aspirated
Radiation Shield
Israeli Thaller Shield
Kahl Self-Aspirating Shield
TT
Gill Naturally
Ventilated Shield
Climat Motor Aspirated Shield
AES Marine Shield
AES Dual Aspirated Psychrometer
Curved Plate Shield
EG &G Dew Point
Hygrometer Shield
Modified Ramos Shield
Gill Aspirated Shield
AES Aspirated
Stevenson Screen
Figure 2.4.1. Examples of various radiation shields (McTaggart-Cowan and
Mcr.ay. 1976).
15
-------
drive the aspirator motor, which draws
ambient air across the sensor at an
average flow of 5 m/s The radiation
shield may take a number of forms, but
common features include double-wall
construction and white paint. Some use
Thermos® bottle type wall construction.
Some are mounted horizontally with the
air intake facing north, while on others,
the tube is mounted vertically with the
air intake facing downward. Some
vertical units also employ domed
polished radiation shields for both sky
and ground radiation shielding (Mc-
Taggart-Cowan and McKay, 1976)
These mechanically aspirated radiation
shields will accommodate the full range
of electronic temperature sensors along
with a variety of dew point and humidity
sensors. It is desirable that aspirator
motors be wired to signal whether or not
they are operating, particularly in tower
installations
2.4.1.3 Temperature Data Require-
ments-ln general, measuring
temperature with an accuracy greater
than 0 5°C may not be necessary.
Extreme daytime horizontal tempera-
ture gradients are well documented
(Hoffmann, 1965, and Department of
the Army, 1975) However, certain
circumstances require more accuracy
or relative accuracy for differential
temperature measurements For
example, a temperature gradient
measurement requires a relative
accuracy of temperature or an absolute
accuracy of temperature difference of
0.1 °C. This is true for the low level local
surface gradient measurement
between 2 m and 10 m above the
ground or for the more conventional
measurement between 10 m and 60 m
Also, if humidity is measured by the
wet-bulb and dry-bulb difference,
sensors should be matched so that
small differences are meaningful.
The very fact that the atmosphere is a
turbulent, differentially heated fluid
should remind any investigator that a
temperature recorded to the nearest
tenth of a degree is representative of
only a very small volume of air for a very
short period of time. All routine
monitoring applications of temperature
data, be they air temperature or
temperature gradient, will be expressed
as a longer term average.
2.4.2 Procurement
In purchasing a suitable temperature
sensor system, follow the general
guidelines advanced in Section 221,
"Instrument Procurement " Pay
particular attention to data requirements,
including management and reduction.
Evaluate the measurement site or sites,
giving consideration to power
requirements, cable lengths, possible
signal interference, sensor mounting
and exposure, and housing requirements
for ancillary equipment
2.4.3 Acceptance Testing
Follow the general guidelines set
forth m Section 2.2.2, "Acceptance
Testing " The general response of the
sensor can be checked by placing it
alternately in warm and cold water and
observing the output. This unnatural
exposure may damage some types of
sensors It is prudent to keep the sensor
assemblies dry by covering them with a
thin plastic bag. This prevents any
"wicking" of water into electronic
assemblies.
If the time constant is to be tested
(important primarily for research
applications), it should be done m the
aspirator at the normal operating flow
rate. The aspirator should provide a flow
by the sensor of about 5 m/s The flow
can be checked with a pitot tube or hot
wire anemometer. If the flow seems
weak, check the motor output, the fan
mechanisms for binding or da mage, and
the air passages for obstructions
2.4.4 Calibration
Refer to the general notes in Section
2.2 3, "Calibrations "
Bench Calibration - In bench
calibrating a temperature sensor, one
may use material changes of state to
obtain an absolute reference
temperature For meteorological
purposes, the triple point of water is one
such absolute reference that falls
within the normal range of ambient
boundary layer temperatures It
requires special equipment to produce
this temperature In calibrating a
remote electronic sensor, all ancillary
equipment, including cables, signal
conditioners, and recorders, should be
hooked up asclosetofield configuration
as is possible to achieve m the
laboratory Deformation-type instru-
ments are usually returned to the
manufacturer for calibration A
minimum of two reference points is
required All calibrations should be
performed in stable thermal masses in
insulated containers, such as Dewar
flasks or Thermos® bottles (Figure
242) A precision mercury-m-glass
thermometer or other calibrated
thermometer is required. This
instrument should be certified as having
been calibrated according to National
Bureau of Standards (NBS) procedures
(Lockhart and Gannon, 1978). A bath
must be continuously agitated either by
hand or with a mechanical stirrer that is
thermally insulated so as not to act as a
heat sink or source Asolid mass m near
thermal equilibrium may also be used
The sensor must be completely
submerged, held with an insulated
clamp, and not allowed to contact either
the stirrer, the reference thermometer,
or the sides of the flask. If an ice bath is
used, it should contain distilled water
and ice slurry made from distilled water
(Wang, 1975) For the second reference,
best results will be obtained by using
water at near-ambient temperature to
reduce heat exchange problems. When
temperatures have stabilized, record
the readings of both the sensor and the
reference thermometer
If the calculated error at each
calibration point falls within the
accepted bounds for the particular
sensor, the calibration can be certified
If adjustments are required, the
calibration checks must be repeated
The bench calibration of a differential
temperature system (often called delta
T) requires both sensors to be at exactly
the same temperature For this, thermal
stability is much more important than
accurate knowledge of the temperature.
If the zero difference is verified at two
temperatures (such as near 0°C and
40°C), then it is also verified that each
sensor is reacting with the same output
vs temperature transfer function
When two are alike it is reasonable to
expect they are following the
manufacturer's nominal transfer
function. The differencing circuit can be
calibrated by using substitute sensors
Figure 2.4.2
A simple bath for cali-
brating thermometers
(Middleton and Spil-
haus, 1953).
16
-------
(resistance boxes, etc.) with an accuracy
consistent with the 01°C goal It is
difficult to maintain differential thermal
masses of known temperature to
achieve this calibration
Field Calibration - If at all possible, the
bench calibration procedures should be
duplicated in the field Obviously, for
this to be possible, the user must have a
mobile laboratory or on-site access to a
sheltered work area To perform a
proper calibration, the sensor must be
removed from its housing For
electronic sensors, all cables must
remain attached If no shelter is
available, calibration must be done
during a period of calm winds under
either cloudy skies or at night under
inversion conditions If these minimum
conditions cannot be achieved, then
field calibration is not possible and a
calibration check will be performed
instead If the check indicates that the
calibration is off, then the instrument
must be brought in for a bench
calibration
In addition to the sensor checks
outlined above, circuitry checks of
remote sensing systems are often made
by replacing the electronic sensor with
resistors of known value Some systems
include buift-m reference resistors
allowing easy "internal" calibration
checks
Both the bench and field calibration
methods described above deal with the
transducer and measurement system
but ignore the coupling by the radiation
shield A collocated transfer standard
(CTS) of known accuracy may be used in
an operational way for a period of time
at least one half a diurnal cycle to
maximize the range being tested With
hourly (or shorter) averages taken over
exactly the same periods of time as the
monitoring system and with assurance
of noninterference with the air which
reaches each sensor, an estimate of
accuracy for the complete measurement
system can be made Errors in coupling
from solar or terrestrial radiation cannot
be adjusted for The solution for this
problem will usually require a better
radiation shield
2.4.5 Installation
Follow the general guidelines set
forth m Section 226, "Preparation for
Field Installation " The shakedown
period for a temperature sensor need
only last long enough for the diurnal
temperature cycle to be defined and a
number of comparisons against an
NBS-certified thermometer to be made,
usually a few days Following bench
calibration and shakedown, the sensor
is shipped to the field Proper siting of
temperature sensors is described in
Section 3, "Methods for Judging
Suitability of Sensor Siting." Louvered
shelter doors and mechanical aspirator
openings should be oriented north in
the northern hemisphere Tower-
mounted sensors should have
downward facing aspirated shields
cantilevered on a boom not less than
one tower diameter in length to
minimize the effects of sensible heat
from the tower structure Installations
on the sides of buildings or through
windows of buildings cannot be
expected to provide unbiased data All
wire connections should be made as
soldered joints, without acid-type
fluxes, using approved soldering
techniques; "cold joints" are one of the
greatest sources of difficulty in
resistance thermometry Proper
grounding procedures will minimize
problems associated with potential
gradients and thunderstorm activity
Once installed, a calibration check
should be made At least one further
cursory operational check should be
made 24 hours after the instrument has
been placed on line
2.4.6 Field Operation of a
Thermometry System
As part of the quality assurance
program, a field calibration should be
performed a minimum of once every 6
months Calibration checks should be
made monthly A technician should visit
the site as often as possible, even daily
No site should go unattended for longer
than a month Each time the site is
visited, a comparison should be made
against a reference thermometer A
side-by-side comparison is desirable
although this may not be practical with
certain tower installations The data
should be inspected for a reasonable
diurnal temperature pattern Where
strip chart recorders are used, time
checks every 24 hours are desirable
Data retrieval will depend upon program
objectives, but even for climatological
programs, data should be retrieved
monthly All operational activities
during a site visit should be logged
2.4.7 Preventive Maintenance
Temperature sensors are basically
maintenance free Most PM on thermom-
etry systems is concerned with hous-
ings Instrument shelters must be
cleaned as often as local weathering
conditions require The air passageways
and screens in radiation shields, both
aspirated and naturally ventilated,
should be cleaned out at least once
every month The aspirator motor
should be checked during each visit. For
remote sites or where data collection is
critical, the aspirator motor should be
changed periodically as recommended
by the manufacturer. Lubricate the
aspirator system as required, but not
oilless bearings. The spare parts inven-
tory should include a sensor and
aspirator motor The most common
cause of component failure is lightning.
Obviously a system damaged by lightning
will require recalibration after repairs
are made All PM activities should be
logged
2.4.8 Audit Procedures
Refer to Section 2 2.4, "Audits," for
general guidelines.
A performance audit on a thermometry
system is simply a calibration check
following the procedures previously
outlined in Section 2.4.4. If a field
calibration check is not physically
possible or practical, then the audit may
consist of running a side-by-side
comparison using a second complete
sensor system with known accuracy
and response characteristics Inserting
a calibrated audit sensor into the same
housing with the sensor being audited
may be a poor procedure, as the second
sensor may alter the flow characteristics
and sensible heat regime within the
housing. There is no requirement
dictated by temperature sensor char-
acteristics for a regular schedule of
temperature instrument audits Of
course, a system audit should be
performed near the beginning of a
measurement program. Other audits
should be geared to program goals and
may be random events.
2.5 Humidity/Dew Point
Measurements
2.5.1 Introduction
Humidity is a general term for the
water-vapor content of air. Other, more
specific, terms for humidity include'
absolute humidity, relative humidity,
specific humidity, mixing ratio, and dew
point (Huschke, 1959). This section
discusses the measurement of relative
humidity and dew point
Relative humidity (RH) is a dimen-
sionless ratio of the actual vapor
pressure of air to the saturation vapor
pressure at a given dry bulb temperature
Dew point is the temperature to which
air must be cooled, at constant pressure
and constant water vapor content, to be
saturated with respect to liquid water.
Frost point is the temperature below
0°C at which air is saturated with
respect to ice
There are many ways to measure the
water vapor content of the atmosphere.
These can be classified in terms of the
six physical principles (Middleton and
Spilhaus, 1953) listed in Table 25 1
Examples of instruments for each
technique are provided.
77
-------
Table 2.5.1. Principles of Humidity Measurement
• Reduction of temperature by evaporation
• Dimensional changes due to absorption
of moisture, based on hygroscopic
properties of materials
• Chemical or electrical changes due to
absorption or adsorption
Formation of dew or frost by
artificial cooling
Diffusion of moisture through porous
membranes
Absorption spectra of water vapor
psychrometer
hygrometers with
sensors of hair,
wood, natural and
synthetic fibers
electric hygrometers
such as Dunmore Cell;
lithium, carbon, and
aluminum oxide strips;
capacitance film
• cooled mirror
surfaces
• diffusion hygrometers
infrared and UV
absorption; Lyman-
alpha radiation
hygrometers
Instruments such as diffusion hygrom-
eters that involve the diffusion of
moisture through porous membranes
are used primarily in research programs.
The same is true of instruments that
utilize the absorption spectra of water
vapor, such as infrared and ultraviolet
hygrometers, and Lyman-alpha radiation
hygrometers This class of instrument
requires frequent attention and repre-
sents a major investment in procure-
ment and maintenance costs.
Psychrometry identifies a basic tech-
nique for deriving both relative humidity
and dew point temperature from a pair
of thermometers—a dry bulb thermom-
eter that measures the ambient temper-
ature, and a wet bulb thermometer. The
reservoir of the wet bulb thermometer is
covered with a muslin wick When the
wick is moistened and the thermometer
ventilated, the indicated temperature is
related to the amount of evaporative
cooling that can take place at the
existing ambient temperature, water
vapor partial pressure, and the atmos-
pheric pressure (Figure 251).
The temperature sensors in a sling
psychrometer are usually mercury- or
alcohol-filled thermometers The same
is true of portable motor-operated
psychrometers, but the psychrometric
principle has been used with sensors
made of thermocouples, wire-wound
resistance thermometers, thermistors,
and bimetal thermometers Relative
humidity and dew point are easily
determined by observing the difference
between the dry bulb and the wet bulb—
the wet bulb depression—and then
referring to psychrometric tables,
charts, or calculators One must be
certain to use computed values for the
atmospheric pressure range of the
location where the observation is taken
More measurements of atmospheric
water vapor have probably been made
with the sling psychrometer than byany
other manual method When properly
used and read, the technique can be
reasonably accurate, but it is easily
misused. The most important errors are
from radiation, changes during reading,
and parallax The Assmann psychrom-
eter continuously aspirates the ther-
mometers and protects them from
(a)
radiation which allows time and accessi-
bility for a careful reading to avoid
parallax (a parallax avoiding guide to
keep the eye perpendicular to the
meniscus is best). For good accuracy,
particularly where a variety of observers
are taking measurements, an Assmann
or equivalent type psychrometer is
recommended. One should use the
psychrometric tables with dew point
values for the altitude (pressure) where
measurements are being made.
Hygrographs, which record relative
humidity, or hygrothermographs, which
record both relative humidity and
temperature, usually incorporate human
hair as the moisture-absorbing sensor.
Other instruments with sensors that
respond to water vapor by exhibiting
dimensional changes are available
They are made with elements such as
wood, goldbeater's skin (an animal
membrane), and synthetic materials,
especially nylon.
Instruments that utilize the hygro-
scopic characteristics of human hair are
used most frequently, primarily because
of availability The hygrograph provides
a direct measure of relative humidity in
a portable instrument that is uncompli-
cated and is relatively inexpensive
There are limitations in accuracy below
20 percent relative humidity and above
80 percent that may be unacceptable, as
Figure 2.5.1
An official National Weather Service sling psychrometer (a) and
an Assmann psychrometer (b).
18
-------
well as limitations for applications at
low temperatures. Atmospheric Envi-
ronment Services of Canada has found
that Pernix, a specially treated and
flattened hair element, can be used at
temperatures below freezing without
serious errors. The hygrothermograph
made to an NWS specification incor-
porates human hair as the humidity
sensor and bourdon tube (a curved
capsule filled with alcohol) as the
temperature sensor
Dew point hygrometers with con-
tinuous electrical outputs are in com-
mon use for monitoring One dew point
hygrometer was originally developed for
air conditioning control applications
under the trade name Dewcel (Hickes,
1947) and was adopted to meteorological
use (Conover, 1950) From the trade
name, the generic term dew cell has
evolved that now identifies an instru-
ment made by several manufacturers
This device determines moisture based
on the principle that for every water
vapor pressure in contact with a
saturated salt solution, there is an
equilibrium temperature at which this
solution neither absorbs nor gives up
moisture to the surrounding atmosphere.
The dew cell, also known by the trade
name Dew Probe, consists of bifilar wire
electrodes wrapped around a woven
glass cloth sleeve that covers a hollow
tube or bobbin. The sleeve is impregnated
with a lithium chloride solution (Figure
252) Low-voltage a.c is supplied to
the electrodes, which are not inter-
connected but depend on the conduc-
tivity of the atmospherically moistened
lithium chloride for current flow The
temperature sensor in the tube is
usually a resistance thermometer, but
can be a thermistor, thermocouple,
bimetal thermometer, capillary system,
or any sensor calibrated for the proper
temperature-to-dew-point relationship
In the early 1960's, the technique of
detecting the dew point on a cooled
mirror surface evolved into a production-
type unit This unit was automatically
operated and had an optical dew-
sensing system that incorporated ther-
moelectric cooling (Francisco and
Beaubien, 1963) Four manufacturers
now produce a meteorological type,
thermoelectric, cooled-mirror dew point
instrument (Mazzarella, 1977) Three of
these instruments cover the range of
-50° to +50°C Linear thermistors are
used to measure the mirror temperature
m three of the units, a platinum wire
sensor is used in the other All are
designed with simultaneous linear
output signals for TdP (dew point
temperature) and T (ambient tempera-
ture) Two of the manufacturers make
claims to NBS-traceability with stated
dew point accuracies ranging from
±0.2° to +0.4°C and ambient tempera-
ture accuracies ranging from ±0 1° to
±0.5°C All incorporate some form of
standardization that involves clearing
the mirror by heating, either automati-
cally or manually. Although complex in
design and operation, this type of
cooled-mirror hygrometer is considered
to be a functional standard
In recent years, two other sensors for
humidity have been used on tower
installations for atmospheric pollution
studies. One involves the use of an
organic seed, cut and coupled to a strain
gage. In principle, absorption of moisture
in the seed results in distortion, which is
converted to an electrical signal by the
strain gage assembly. Reports on
performance are mixed. Certainly the
applications are limited, and the approach
does not represent a technological
advance By contrast, the thin film
capacitor, designed primarily for radio-
sonde applications, incorporates ad-
vanced technology (Suntola and Antson,
1973) Reports of users have been
mixed, with a common complaint of
poor performance in polluted atmos-
pheres. The capacitor may still be
considered a sensor for some research
applications on towers
2.5.1.1 Sensor Charactenstics-AI-
though the psychrometer is considered
8 1/4"
7 1/4"
\~-35/16"-\
Figure 2.5.2. A typical Dewcel
sensor housing and
transmitter.
the most practical and widely used
instrument for measuring humidity, two
major problems are associated with wet
and dry bulb psychrometry involving the
accuracy of the thermometers and the
cumulative errors related to operating
technique (Qumn, 1968). An accuracy
of ±1 percent at 23°C and 50 percent
RH requires thermometers with relative
accuracy of ±0.1 °C The commonly
used 0.5°C division thermometers
introduce an uncertainty of ±5 percent
RH at this condition This assumes that
the readings were taken atthe maximum
wet bulb depression, a difficulttask with
a sling psychrometer
It has long been recognized that there
are some limitations in using the dew
cell instrument (Acheson, 1962) The
lowest relative humidity it can measure
at a given temperature is the ratio of the
vapor pressure of a saturated solution of
LiCI to that of pure water. This is
calculated to be 118 percent RH. A
second limitation is that at -65.6°,
-20.5°, +19.0°, and +94.0°C, LiCI in
equilibrium with its saturated solution
undergoes a phase change Errors in
dew point measurements occur at -69°,
-39°, -12°, and +40°C. This problem is
inherent in the useof LiCI and cannot be
eliminated. It is estimated that the
accuracy of the LiCI saturated salt
technique is 1.5°C over the range of
-30°to30°C.
The optical chilled (cooled) mirror
technique of measuring dew point is a
fundamental measurement. No calibra-
tion is required for the fundamental dew
generating process. The measurement
however is the temperature of the
surface at which the dew forms and as
with any electrical temperature mea-
surement system, calibration is required
The process of periodically heating the
mirror to a temperature above the dew
(or frost) point is followed by a balancing
of the optical system to correct for
changing the dry mirror reflectance that
might result from contamination In the
better instruments, automatic balancing
is programmable in terms of frequency
and length of time. It can also be
accomplished manually.
2.5.1.2 Sensor Housings and Shields-
Psychrometers of all types should be
acclimated to the environmental condi-
tions in which the measurements are to
be made. In most cases, psychrometers
should be stored m a standard instru-
ment shelter so that the mass of the
thermometers, and especially the mass
of the housing, adjusts to the tempera-
ture of the air Psychrometers with a
stored water supply, such as those on a
tower, must be shielded from solar
radiation
19
-------
For meteorological applications, the
dew cell element should be enclosed in
a weatherhood to protect it from
precipitation, wind, and radiation effects.
This type of element functions best in
still air. Some aspirated radiation
shields are designed, in keeping with
these specifications, to house both a
temperature sensor, which requires
ventilation, and a dew cell, which
requires only the smallest amount of air
flow (Figure 2.5.3). The miniaturization
of the dew cell has created some
problems related to excessive air flow
and solar radiation that remain only
partially solved
All manufacturers of optical cooled-
mirror dew point and temperature
monitoring equipment provide housings
for the sensors, which include forced
ventilation and shielding from solar
radiation.
2.5.1.3 Data Requirements-Electrical
hygrometers for monitoring applications
have time constants generally longer
than air temperature systems. The
usual data of interest are hourly average
values. Data should be reported in terms
of the condition measured, dew point
temperature, relative humidity or wet-
bulb and dry-bulb temperature Programs
may be used which convert among
these if all the relevant variables are
known The station elevation may be
used to estimate a nominal pressure if a
measurement is not available The
temperature needed to convert a relative
humidity measurement to dew point
temperature is that temperature at the
relative humidity sensor surface This
may not be the same temperature as
that measured at some other location
On the other hand, the dew point
temperature is a fundamental measure
of the amount of water vapor in the air
and is independent of air temperature
Relative humidity calculations can
therefore be made given the dew point
temperature and any temperature
measurement point in the samegeneral
air mass
Psychrometers are convenient devices
for making spot checks of the perform-
ance of other devices, especially those
that are permanently installed, providing
the checking is done under reasonably
steady overcast conditions The psychro-
metric technique built into tower
installations presents servicing problems,
especially at temperature extremes
High temperatures cause rapid evapora
tion, and low temperatures cause
freezing
Both the dew cell and the cooled-
mirror type instruments have applica-
tions on 10-meter or taller tower
installations for pollution studies,
providing the sensors are housed in the
Figure 2.5.3. A pair of tower-mounted Gill aspirated radiation shields for
housing temperature and dew point sensors (R M Young Co.).
recommended shields with little, if any,
aspiration for the dew cell and the
recommended rate of aspiration for the
cooled-mirror design is selected
2.5.2 Procurement
In purchasing a suitable humidity/dew
point measuring system, follow the
general guidelines advanced in Section
221, "Instrument Procurement "
Sling psychrometers and aspirated
psychrometers with thermometers
shorter than 10 inches do not have
sufficient resolution for the accuracies
required for checking other instruments
Equally important, the thermometers
should have etched stems, i e , the scale
markings should be etched on the glass
Reliable thermometers are factory
calibrated at a minimum of two tem-
peratures,and usuallyatthree Thermom-
eters calibrated with NBS-traceable
standards are preferred
When patents expired on the original
Dewcel, a number of similar units
appeared on the market In light of
problems which have existed in the
past, it is prudent to specify accuracy of
the humidity system when it is operating
as a system in the atmosphere Problems
with ventilation rates will be quickly
exposed by this requirement It is not
recommended to purchase components
to patch together m a system Corrosion
in polluted atmospheres can be avoided
by selecting optional 24-carat gold
windings, provided cost is not prohibitive.
It dew point alone is tobe measured, the
standard weatherhood is a proper
choice If both temperature and dew
point are to be measured, it may be
advantageous to purchase a standard
shield that provides a housing for the
dew cell and a separate aspirated
compartment for the temperature
probe
Optical cooled-mirror dew point
systems are now available commercially
from four manufacturers, all of which
incorporate either linear thermistors or
platinum resistance temperature devices
2.EI.3 Acceptance Testing
Follow the general guidelines set
forth m Section 2.2.2, 'Acceptance
Testing " Test at least the ambient
atmosphere at one point in normal wind
and radiation
2.5.4 Calibration
Refer to the general notes m Section
223, "Calibrations " The procedure for
calibrating the thermometers in a
psychrometer is essentially the same as
any thermometer calibration (see Section
244)
Both the dew cell and the cooled-
mirror hygrometer can be checked for
approximate calibration accuracy with a
motor-operated psychrometer Their
performance should be verified under
stable conditions at night or under
cloudy conditions during the day
Several readings taken at the intake of
the aspirator or shield are recommended
Bench calibrations of these more
sophisticated units must be made by the
manufacturer The electronics portion
of some instruments may be calibrated
by substitution of known resistances m
place of the temperature sensor This
procedure, if appropriate, isdescribed in
the manufacturer's operating manual
for the instrument
20
-------
2.5.5 Installation
Follow the general guidelines set
forth in Section 2.2.6, "Preparation for
Field Installation" and in Section 3,
"Methods for Judging Suitability of
Sensor Siting "
Dew point measuring equipment on a
tower should be installed with the same
considerations given to temperature
sensors. Reference has already been
made to the weatherhood as a shield for
the dew cell and to an aspirated shield
for the cooled-mirror instrument. At
some installations, success has been
reported in mounting these housings so
that they are close to the tower
framework on the north-facing side.
This minimizes the effects of direct solar
radiation and provides a rigid support,
especially for the cooled-mirror sensor,
which requires a stable mounting
surface Another consideration in
mounting these devices inboard involves
servicing Inboard mounting makes
recharging the dew cell with lithium
chloride and cleaning the reflective
surface of the cooled-mirror hygrometer
much easier
2.5.6 Field Operation and Preventive
Maintenance
Note the guidelines in Section 225,
"Operational Checks and Preventive
Maintenance "
Field calibration checks should be
made at least monthly on dew cell type
units The use of gold wire windings
around the LiCI cylinder minimizes
corrosion problems in polluted atmos-
pheres. Periodic removal and washing
of old lithium chloride, followed by
recharging with a fresh solution,
improves data reliability
Once a mercury or alcohol liquid-m-
glass thermometer is calibrated, there is
no need for recalibration, unless it is to
be used for reference or as a transfer
standard Errors in wet bulb tempera-
tures are most frequently the result of
an improperly installed or dirty muslin
wick, the repeated use of tap water
instead of distilled water, or human
error in reading Wicking material used
on psychrometers must be washed to
remove traces of sizing and fingerprints
Once cleaned, the material is tied at the
top of the thermometer bulb and a loop
of thread placed around the bottom so
the thermometer bulb istightly covered
To prevent solid matenalsfrom collecting
on the cloth and preventing proper
evaporation, the wick should be wet
with distilled water Of course, slinging
or motor aspiration should be done in
the shade, away from reflected or
scattered radiation, at a ventilation rate
of about 3 to 5 m/s Many technique-
related errors are minimized by using an
Assmann-type, motor-operated psychrom-
eter, providing the instrument is allow-
ed to assume near ambient conditions
prior to use.
The cooled-mirror instruments require
no calibration except for the minor
temperature sensor. Depending on
environmental conditions, the mirror is
easily cleaned with a Q-Tip® dipped in
the recommended cleaning fluid, usually
a liquid with an alcohol base. While the
accuracy of a psychrometer is inferior to
that of the optical chilled mirror system,
an occasional check at the intake to the
sensor shield is recommended under
the provisions specified earlier
All operational and PM activities
should be logged. Data retrieval will be
dependent upon program objectives.
2.5.7 A udit Procedures
Refer to Section 2.2.4, "Audits," for
general guidelines. Instrument audit
procedures for hygrometry systems
follow calibration procedures. A systems
audit should be performed near the
beginning of a field measurement
program
The performance audit of a humidity
measuring system should be based
upon a comparison with a collocated
transfer standard (CTS) Parts of the
system can be tested by conventional
electronic tests, but this avoids so much
of the measurement process that it
should only be used to augment the tola I
system test The CTS may be any
qualified instrument The most accurate
type is the cooled-mirror dew point
instrument. The Assmann-type psychrom-
eter with calibrated thermometers trace-
able to NBS is acceptable for most data
applications. It is also most convenient
since it does not require commercial
power and can be carried to elevated
levels on a tower.
Given the qualifier that humidity is a
very difficult measurement to make, a
rule of thumb for judging the accuracy of
a humidity monitoring system with an
Assmann as the CTS is as follows-when
the CTS and the challenged system
agree in dew point temperature to
within 1°C, the challenged system is
assumed to be within 0 5°C of the true
value This arbitrarily assigns an uncer-
tainty in dew point temperature of
±0 5°C for the Assmann which is true
for most of the range
2.6 Solar Radiation
2.6.1 Introduction
Solar energy is the driving force of
large-scale atmospheric motion, indeed,
of the general circulation of the atmos-
phere. Although air pollution investi-
gators normally consider the measure-
ment of solar radiation secondary to
wind and temperature measurements,
solar radiation is directly related to
atmospheric stability. It is measured as
total incoming global radiation, as
outgoing reflected and terrestrial radia-
tion and as net total radiation.
Quantitatively, solar radiation is
described in units of energy flux, either
W/m2 or cal/cm2-min. When measured
in specific, narrow wavelength bands,
solar radiation may be used to evaluate
such air pollution indicators asturbidity,
amount of precipitable water, and rates
of photochemical reactions. However,
this manual will cover only broadband
measurements and sunshine.
The generic term, radiometer, refers
to any instrument that measures
radiation, regardless of wavelength.
Shortwave radiation has wavelengths
less than 4 micrometers (/jm) and is
subdivided as follows.
Ultraviolet (UV) 0.20 fjm to 0.38 /urn
Visible 0.38/um to 0.75/um
Near-infrared 0.75 /jm to 4 00 //m
The infrared (IR) radiation band from 4
urn to 100 /urn is considered longwave
radiation. The instruments most com-
monly used for environmental monitor-
ing are discussed below
Pyranometers are instruments that
measure the solar radiation received
from the hemispherical part of the
atmosphere it sees, including the total
sun and sky shortwave radiation (Figure
2.6.1). Most pyranometers incorporate
a thermopile as sensor Others that
measure a narrower, shortwave band-
width, use a silicon photovoltaic cell as a
sensor The spectral pyranometer with
two hemispherical domes is designed to
measure sun and sky totally or in
defined wavelengths. This is achieved
by substituting one of several colored
Schott glass filter domes for the clear
glass outer dome
Bimetallic recording pyranometers,
also known as actmometers, were
designed by Robitzsch of Germany
These mechanical sensors consist of
two or three bimetallic strips, alternately
painted black and white, that respectively
absorb and reflect solar radiation The
resulting differential heating producesa
deformation that is transmitted me-
chanically through levers and a pen arm
to a clock-wound drum recorder Although
of limited accuracy, these instruments
are useful for locations with no com-
mercial power
Net radiometers or netpyrradiometers
are designed to measure the difference
between downward and upward total
radiation, including the total incoming
shortwave and longwave radiation and
the total outgoing shortwave and
longwave radiation. There are two basic
types of net radiometers The ventilated
plate type, often referred to by the name
21
-------
Sensing
Element
Guard Disc
\
Precision Ground &
Polished Glass Dome
(Frequently Double)
^Leveling
II. II III Screw
Figure 2.6.1. The features of a typical pyranorneter (Carter, et a/., 1977)
of the designers (Gier and Dunkle), is
more popular in research applications
than the type with hemispherical
polyethylene domes originally designed
by Funk Both incorporate thermopiles
with blackened surfaces. Because net
radiometers produce a signal with a
positive sign when the incoming radia-
tion exceeds the outgoing, the recording
equipment must be designed with an
offset zero
Sunshine recorders are designed to
provide information on the hourly or
daily duration of sunshine. Only one
commercially available, off-the-shelf
type of sunshine recorder is now
available This is the Campbell-Stokes
design (Figure 2.6 2), designated as the
interim reference sunshine recorder
Figure 2.6.2.
A Campbell-Stokes
Sunshine recorder
Department of the
Army, 1975).
"IRSR" by the World Meteorological
Organization. The device consists of a
glass sphere 10 cm in diameter mounted
in a spherical bowl. The sun's rays are
focused on a card that absorbs radiation
and changes color m the presence of
sunlight. The recorder is used infre-
quently in the United States but
extensively abroad, primarily for the
collection of climatological data. The
National Weather Service routinely
uses a Sunshine Switch, which incor-
porates one shaded photocell and one
exposed photocell
2.6.1.1 Instrument Characteristics-
Only the characteristics of pyranometers
and net radiometers, the two types of
instruments used most frequently in
pollution-related programs, will be
discussed in this section
Pyranorneter Characteristics - The
pyranorneter is not to be confused with
the pyrheliometer, "an instrument for
measuring the intensity of direct solar
radiation at normal incidence" (WMO,
1971) The pyrheliometer is mounted in
a solar tracker, or equatorial mount,
automatically pointing to the sun as it
traverses from east to west (Figure
263) By contrast, the pyranorneter is
mounted facing toward the zenith
Ideally, the response of the thermopile
sensor in the pyranorneter is propor-
tional to the cosine of the angle of the
solar beam and is constant at all
azimuth angles This characteristic is
known as the Lambert Cosine Response,
an important characteristic of pyranom-
eters. For the majority of applications
related to atmospheric pollution, Class 2
and Class 3 are satisfactory
Net Radiometer Characteristics -
Most net radiometers now available
commercially are made with a small
disc-shaped thermopile covered by
polyethylene hemispheres In most
units the material used for shielding the
element from the wind and weather is
very thin and is transparent to wave-
lengths of 0.3 to 60 //m. Until recently,
the internal ventilation and positive
pressure required to maintain the
hemispheres of net radiometers in their
proper shape was considered critical;
however, new designs have eliminated
this problem. The plate-type net radiom-
eter, most often the modified Gier and
Dunkle design sold commercially in the
United States, is occasionally used in
routine air pollution investigations The
thermopile heat flow transducer is
blackened with a material that is easily
cleaned with water or naphtha. Because
the thermopile is uncovered for total
spectrum response, a built-in blower,
available for operation on 11 5 V 50/60
Hz or 12 V d c., draws air across the
element at a constant rate eliminating
the effects of varying natural winds. The
device is temperature-compensated
and typically has a sensitivity of 2 2 uM
per W/m2, a response time of 10
seconds, and a "relative" accuracy of
two percent in calibration When
supplied with a reflective shield on its
lower surface, this plate type net
radiometer of the Gier and Dunkle
design becomes a total hemispherical
radiometer or unshielded pyranorneter
2.6.1.2 Recorders and Integrators for
FVranometers and Net Radiometers -
The relatively high impedance and low
signal of thermopile sensors, excluding
silicon photovoltaic cells, limits their
use with both indicating meters and
recording meters Electronic strip chart
millivolt potentiometric recorders in-
corporating variable-range rheostats
are preferred The variable-range
rheostat permits the exact matching of
the recorder scale to interchangeable
sensors so that deflections of the meter
represent engineering units, i e., W/m2,
cal/cm2-mm, etc The alternative is a
standard millivolt-meter potentiometric
recorder where the data, in millivolts,
must be translated to units of energy,
corresponding to full-scale values of
1370 W/m2 or 1 96 cal/cm2-mm With
some sensors, it is necessary to use a
preamplifier to increase the level of the
signal It may also be necessary,
especially if the signal is to be used as
an input to a computer, to combine
preamplification with scaling
2.6.2 Procurement
In purchasing a solar radiation
measurement system, followthegeneral
guidelines advanced in Section 221,
"Instrument Procurement " Many types
of radiation instruments have been
developed, especially in recent years,
because of an increasing interest in
environmental considerations (Gates,
1962), meteorological research (Mon-
teith, 1 972), and solar energy (Carter, et
22
-------
Pyrheliometer
Declination Adjustment
24 hr. Dial
Latitude Adjust
Figure 2.6.3.
Features of a typicalpyrheliometer and tracking mount (Carter et
a/., 1977)
al., 1977). Except for special studies, the
requirements for relating radiation to
stability can be satisfied by purchasing
sensors of Class 2 or Class 3 as
identified by the WMO Class 2 sensors
offer the advantage of providing data
comparable to that collected at National
Weather Service stations and at key
locations of DOE. The sensors to be
specified should be commercially avail-
able, field proven by the manufacturer
for several years, and have the technical
requirements established by WMO
standards An American Society for
Testing and Materials (ASTM) standard
is now under consideration When
purchasing a recorder or integrator, one
must match the calibration factor or
sensitivity of the sensor to the readout
equipment It must be recognized that
the signals from net radiometers, in
contrast to pyranometers, require zero-
offset capability to accommodate both
negative and positive voltage outputs
2.6.3 Acceptance Testing
Follow the general guidelines set
forth in Section 2 2.2, "Acceptance
Testing " Physical inspection of the
relatively fragile pyranometers or net
radiometers immediately after delivery
of the instrument is important. One
must be sure that the calibration data
have been received and that the data
correspond to the serial number of the
instrument. Storage of this information
will prove helpful when the time comes
to have the calibration of the instrument
checked, or to replace the sensor or
readout device Few organizations are
equipped or staffed to bench-test a
radiometer to verify calibration, but a
quick determination can be made
indoors as to whether the sensor and
recorder or integrator system is operating
by exposing the sensor to the light of a
tungsten lamp It may be necessary to
place the instrument fairly close to the
lamp Covering the sensor for several
hours will ensure that the system is not
"dark counting "
2.6.4 Calibration
Refer to the general notes in Section
2.2 3, "Calibrations " The user of a
pyranometer or net radiometer is
normally not equipped to calibrate the
sensor The best the user can do is to
perform field calibration checks on two
clear sky days These checks involve a
side-by-side comparison of the sensor
to a sensor of similar design, the
calibration of which can be traced to a
transfer standard The primary standard
pyrheliometer is an instrument selected
by an international committee. It was
made by Jet Propulsion Lab (JPL)
where custody is maintained. If this is
not possible, the device should be
returned to the manufacturer or to a
laboratory with the facilities to check
the calibration. The frequency of making
comparative readings or having factory
calibrations will depend on environ-
mental conditions. Most certainly, any
indication of discoloration or peeling of
the blackened surface or of scratches on
the hemispheres of a pyranometer
warrants recalibration and/or service
Net radiometers are more delicate
and require more frequent attention
than pyranometers Pyranometers of
high quality in a clean atmosphere may
require recalibration at 2-year intervals,
net radiometers should be recalibrated
at least yearly. Calibrating the recorder
or integrator is an easy task. The
standard method involves the use of a
precision potentiometer to impress
known voltages into the circuit The
linearity of the readout instrument may
be checked by introducing a series of
voltages covering the full scale, checking
first up-scale and then downscale.
Adjustments should be made as neces-
sary. In the absence of a precision
potentiometer, it may be possible to
introduce a calibrated millivolt source
that covers one or two points. Integrators
can be checked the same way, except
that the input value must also be timed
2.6.5 Installation
Follow the general guidelines set
forth in Section 2.2.6, "Preparation for
Field Installation." The site selected for
an upward-looking pyranometer should
be free from any obstruction above the
plane of the sensor and should be
readily accessible for cleaning and
maintenance. It should be located so
that no shadows will be cast on the
device, so that it is not too close to light-
colored walls or other objects likely to
reflect sunlight, and so that it is not
exposed to artificial radiation sources A
flat roof is usually a good choice, but if
such a site is not possible, a rigid stand
with a horizontal surface some distance
from buildings or other obstructions
should be used. A site survey of the
angular elevation above the plane of the
radiometer surface should be made
through 360 degrees (The Eppley
Laboratory, Inc ).
Table 2.6.1. Classification of Pyranometers According to Physical Response Characteristics
7sf Class
2nd Class
3rd Class
Sensitivity
(mW/cm*)
±0 1
±05
±1 0
Temperature
±7
±2
±5
Linearity
±1
±2
±3
Time Const.
(max.)
25 s
1 m
4 m
Cosine Resp.
±3.0
±5.7
±10
23
-------
The same procedures and precau-
tions should be followed for net radiom-
eters that are both upward- and down-
ward-looking. However, the instrument
must be supported on an arm extending
from a vertical support about 1 m above
the ground. Except for net radiometers
with heavy-duty domes, which are
installed with a desiccant tube in series
with the sensor chamber, most other
hemispherical net radiometers require
the positive pressure of a gas—usually
nitrogen—to both maintain the shape of
the polyethylene domes and purge the
area surrounding the thermopile. In one
increasingly popular design, there is a
requirement for internal purging with
nitrogen and external ventilation with
compressed air through holes on the
frame. The compressed air supply mini-
mizes fogging and condensation
Precautions must be taken to avoid
subjecting radiometers to mechanical
shock during installation. They should
be installed securely and leveled using
the circular spirit level attached to the
instrument. Net radiometers are difficult
to mount and to maintain free of
vibration Pyranometers of the Moll-
Gorzynski design, used extensively by
Atmospheric Environmental Sciences
(AES) of Canada, are oriented sothat the
emerging leads face north (Figure
2.6.4) This minimizes solar heat on the
electrical connections of an instrument
that is not temperature compensated.
The thermopiles of these instruments
should be oriented so that the long side
of the thermopile points east and west
(Latimer, 1972). The cable used to
connect the pyranometer to the readout
device, recorder, or integrator should be
between 16 and 20 gauge and made of
shielded, waterproofed 2-conductor
copper wire. The sensor, shield, and
readout device should be connected to a
common ground Potentiometnc milli-
volt recorders are to be used with most
high-impedance, low-signal thermopile
radiometers Cable lengths of 300 m or
more are practical. Galvanometric
recorders can be used with silicon cell
radiometers. Soldered, copper-to-copper
junctions between instrument connec-
tors and/or cables are essential.
Pyranographs or actmographs should
be installed on a level surface immune
to shadows These instruments should
be placed in such a way that the
sensitive bimetallic strips lie within 2
degrees of true east and west with the
glass inspection window facing north
2.6.6 Field Operation of a Solar
Radiation System
As part of the quality assurance
program, a field calibration check
should be performed at least once every
6 months according to the procedures
outlined m Section 264 Solar radiation
instruments require almost daily atten-
tion. The data should be inspected for a
reasonable diurnal pattern and the
absence of dark counting Where strip
chart or digital printers are used, daily
time checks are desirable Data retrieval
will depend upon program objectives,
but even for climatological programs,
data should be collected monthly. All
operational activities during a site visit
should be logged
2.6.7 Preventive Maintenance
All types of radiometers require
frequent cleaning to remove any mate-
10 in.
Figure 2.6.4. A Moll-Gorczynski solanmeter (Department of the Army, 1975)
24
rial deposited on the surface that will
intercept the radiation In most cases,
this is a daily operation. The outer
hemisphere should be wiped clean and
dry with a lint-free soft cloth Any
scratching of the surface will alter the
transmission properties of the glass, so
cleaning must be done with care. If
frozen snow, glazed ice, hoarfrost or
rime ice is present, an attempt should be
made to remove the deposit carefully
with warmed cloths.
Should the internal surface of a
pyranometer's outer hemisphere be-
come coated with moisture, it can be
cleaned by carefully removing the outer
hemisphere on a dry day and allowing
the air to evaporate the moisture If
removal of a hemisphere exposes the
thermopile element, extreme care
should be taken because it is fragile and
easily damaged About once each
month, the desiccant installed m most
pyranometers should be inspected
Whenever the silica gel drying agent is
pink or white instead of blue, it should
be replaced or rejuvenated by drying it
out on a pan in 135°C oven. The level
should be checked after each servicing
of the radiometer, or at least monthly
Significant errors can result from
misalignment.
Net radiometers require more fre-
quent maintenance attention than
pyranometers. It is necessary to replace
the polyethylene domes as often as
twice a year or more before the domes
become discolored, distorted, or cracked
More frequent replacement is necessary
in polluted environments due to accele-
rated degradation of plastic hemispheres
when exposed to pollulants A daily
maintenance schedule is essential to
check on the proper flow of gas in
instruments that are inflated and
purged with nitrogen. All PM activities
should be recorded in a log
2.6.8 Audit Procedures
Refer to Section 224, "Audits," for
general guidelines A performance
audit on a solar radiation system is only
practical with a CTS. The CTS must have
the spectral response and exposure
equivalent to the instrument being
audited One diurnal cycle will establish
an estimate of accuracy sufficient for
most air quality monitoring applications
The method of reporting the data from
the monitoring instrument (daily inte-
grated value, hourly integrated value,
average intensity per hour, etc ) must be
used m reducing the data from the CTS
to provide a meaningful comparison An
audit frequency of at least six months is
recommended
-------
2.7 Precipitation
Measurements
2.7.1 Introduction
By definition, "The total amount of
precipitation which reaches the ground
in a stated period is expressed as the
depth to which it would cover a
horizontal projection of the earth's
surface if there were no loss by
evaporation or run-off and if any part of
the precipitation falling as snow or ice
were melted" (WMO, 1969) In any
method of precipitation measurement,
the aim should be to obtain a sample
that is representative of the fall in the
area At the outset, it should be
recognized that the extrapolation of
precipitation amounts from a single
location to represent an entire region is
an assumption that is statistically
questionable. A network of stations
with a density suitable to the inves-
tigation is preferable.
Precipitation collectors are of two
basic types' nonrecording and record-
ing
Nonrecording Gages - In its simplest
form, a precipitation gage consists of a
cylinder, such as a can with straight
sides, closed at one end and open at the
other. The depth of the liquid in the can
can be measured with a measuring stick
calibrated in subdivisions of centimeters
or inches (Figure 271)
To obtain greater resolution, as in the
case of the standard 8-mch gage made
to NWS Specification No 450.2301, the
gage is constructed with a ratio of 101
between the area of the outside collec-
torcylmder and the inside measuring
tube. The funnel attached to the
collector both directs the precipitation
into the tube and minimizes evaporation
loss Amounts in excess of two inches of
rainfall overflow into the outer can, and
all measurements of liquid and melted
precipitation are made m the measuring
tube with a measuring stick
The automatic wet/dry precipitation
collector, available in several designs,
represents a specialized nonrecording
instrument designed for programs
involving the chemical and/or radio-
active analysis of precipitation. The
collector is built with a sensor that
detects the onset and cessation of
precipitation and automatically releases
a lid to open and cover the collector In
one design, the lid can be made to
remain open during either wet or dry
periods. Another model is made with
two collectors, the lid is made to cover
one bucket during periods of ram and
snow (Figure 272) In equipment of this
kind involving precipitation chemistry,
the volume of water in proportion to the
Figure 2.7.1. A typical standard rain gage (Be/fort Instrument Company).
Figure 2.7.2. Automatic wet/dry precipitation collector.
constituents collected with the water is
important, so evaporation must be kept
to a minimum
Recording Gages - Recording gages
are of two basic designs based on their
operating principles the weighing-type
gage and the tipping bucket-type gage
(Figure 273) The former, when made
to NWS Specification No 450.2201, is
known as the Universal gage, indicating
usage for both liquid and frozen precipi-
tation There are options for the remote
transmission of signals from this type of
gage The standard National Weather
Service Tipping Bucket Ram Gage is
designed with a 12-inch collector
funnel that directs the precipitation to a
small outlet directly over two equal
compartments, or buckets, which tilt in
sequence with each 001 inches of
rainfall The motion of the buckets
causes a mercury switch closure
Normally operated on 6 V d c , the
contact closure can be monitored on a
visual counter and/or one of several
recorders The digital-type impulse can
also be used with computer-compatible
equipment
25
-------
Bucket
Platform
Tipping
Bucket
Figure 2.7.3. A typical weighing rain gage (left) and typical tipping bucket rain gage (Belfort Instrument Company).
2.7.1.1 Instrument Characteristics -
The most accurate precipitation gage is
the indicating-type gage However, the
recording-type gage measures the time
of beginning and ending of rainfall and
rate of fall The Universal weighing gage
incorporates a chart drum that is made
to rotate by either an 8-day spring-
wound clock or a battery-powered clock
Recent developments include a unit
with a quartz crystal mechanism with
gear shafts for a wide range of rotation
periods from half a day to one month
The weighing gage is sometimes
identified by the name of its designer
(Fergusson) and comes with one of two
recording mechanisms In the single
traverse unit, the pen moves from the
base of the drum to the top, typically a
water equivalent of 6 inches In a dual
traverse unit, the pen moves up and
then down for a total of 12 inches of
precipitation A variation of the weighing
gage, a "high capacity "design with dual
traverse, will collect as much as 760
mm or 30 inches To minimize the
oscillations incurred by strong winds on
the balance mechanism, weighing
gages are fitted with a damper immersed
in silicone fluid. By incorporating a
potentiometer in the mechanism, the
gage becomes a remote transmitting
unit, capable of providing a resistance
or, as another refinement, a voltage
proportional to the amount of precip-
itation collected. Linearity of response
is usually a factory adjustment involving
the use of calibrated weights to simulate
precipitation amounts In spite of
manufacturer's specifications, it is
doubtful that the gage can resolve 0.01
inches, especially when the bucket is
nearly empty.
In the tipping bucket gage, the
balance of the buckets and the leveling
of the bucket frame are critical. Low
voltage at the gage is imperative for
reasons of safety Power is typically 6 V
d c The signal is provided by a switch
closure each time the bucket assembly
tips (0.01 inches of rainfall per bucket)
Ram rales are calculated from an event
recorder with pens energized sequen-
tially to improve resolution. The tipping
bucket (a mechanical device) takes time
to tilt from one position to the next.
When the rate of fall is high, there is
spillage and the unmeasured precipita-
tion falls into the reservoir. Where there
is a need for greater accuracy, the
collected water is measured manually,
and excess amounts are allocated
proportionately in the record The
accuracy of the gage is given as 1
percent for rainfall rates of 1 in/hr or
less; 4 percent for rates of 3 m/hr, and 6
percent for rates up to 6 in/hr
2.7.1.2 Windshields and Heaters -
Accuracy of measurement for all types
ot gages is influenced perhaps more by
exposure than by variations in design
Windshields represent an essential
accessory to improve the catch of
precipitation, especially snow m windy
conditions The improved Alter design,
made of 32 free-swinging but separated
leaves supported 1/2 inch above the
level of the gage collecting orifice, is an
effective way to improve the catch In a
comparison of shielded and unshielded
26
-------
8-inch gages, it has been shown that at
a wind speed of 5 mph, the efficiency of
the unshielded gage decreases by 25
percent, and at 10 mph, the efficiency of
the gage decreases by 40 percent
(Weiss, 1961).
In belowfreezmg conditions when the
catch in a gage is snow or some other
form of solid precipitation, it is necessary
to remove the collector/ funnel of
nonrecordmg gages and the funnel in
recording gages Some instruments are
available with built-in heater elements
that are thermostatically controlled An
effective heater for conditions that are
not too severe is an incandescent lamp
installed in the housing of the gage
Caution should be exercised, however,
as too great a heat will result in
evaporative loss
2.7.1 3 Precipitation Data Require-
ments-ln research studies, especially
those related to acid rain, the instru-
ment used most frequently is the
Automatic Precipitation Collector with
one or two collecting buckets and a
cover to prevent evaporation In opera-
tional activities, the choice is between
the weighing gage and the tipping
bucket gage For climatological surveys,
the choice might include one of the
above gages as well as a nonrecordmg
type gage The use of a windshield is
recommended to minimize the errors
that result from windy conditions if the
application requires maximum accuracy
The precipitation measurement
made in air quality monitoring stations
is frequently used for descriptive
purposes or for episodal analysis If
the effort required to achieve the level
of accuracy specified by most manu-
facturers of electrical recording gages is
more than the application of the data
can justify, a tolerance of 10 percent
may be adequate
2.7.2 Procurement
In purchasing a suitable precipitation
measuring sytem, follow the general
guidelines set forth in Section 2.2 1,
"Instrument Procurement " A variety of
gages are available commercially In
general, the standards established by
NWS specifications result in the fewest
problems For example, there are
numerous 8-mch gages available, but
those following NWS specifications are
made only of brass and copper, are more
durable, and are reported to rupture less
frequently under extended freezing
conditions than those made of galva-
nized steel
The procurement of a weighing-type
gage should include a tripod mounting
base as well as a set of calibration
weights For locations that may not be
readily accessible, or for locations with
heavy precipitation, the bucket of the
weighing gage should have an overflow
tube If the resolution of time is not too
important, recording ram gages of the
drum type can be obtained with monthly
rather than weekly mechanisms Unless
the tipping bucket gage is equipped with
a heater, it is of no use for frozen
precipitation
2.7.3 Acceptance Testing
Follow the general guidelines advanced
in Section 222, "Acceptance Testing."
Except for visual inspection, nonrecord-
mg gages do not require acceptance
testing. The weighing gages should be
assembled and given a quick "bench-
top" calibration check using standard
weights or a measured volume of water
In addition, the clock mechanism
supplied with the gage should be
checked for at least a couple of days,
preferably a week The tipping bucket
gage should also be bench tested,
primarily to be certain that the bucket
mechanism assembly is balanced and
that the switch is operational
2.7.4 Calibration
Bench calibrations should follow the
recommendation of the manufacturer
The electrical output gage or the drum
recording gage measures weight,
whether total weight in the case of the
"weighing gage" or increments of
weight in the case of the tipping bucket
gage Density of water is assumed so
the weight can be expressed in units of
volume or depth assuming the area of
the collector opening Calibrations of
the measurement apparatus can be
based upon the introduction of known
volumes of water. For rate-sensitive
systems such as the tipping bucket, the
rate of simulated precipitation should be
kept less than one inch per hour
Calibrations require properly leveled
weighing systems (gages) whether on
the bench or in the field
2.7.5 Installation
Follow the general guidelines set
forth in Section 226, "Preparation for
Field Installation," and siting guidance
described in Section 3.
The support, or base, of any gage
must be firmly anchored, preferably on a
level surface so that the sides of the
gage are vertical and the collector is
horizontal The collector can be checked
with a carpenter's level placed at two
intersecting positions. The level of the
bucket assembly on the tipping bucket
gage is also critical and should also be
checked along its length and width
Once the weighing gage is installed,
the silicone fluid should be poured into
the damping cylinder as required The
pen of the drum recording type is inked
to less than capacity because the ink
used is hygroscopic and expands with
increasing humidity, easily spilling over
on the chart. The final calibration check
with standard weights or suitable
substitute should be made at this time.
To check the operation of the tipping
bucket, the best approach is to put a
known quantity of water in a can with a
small hole so that the slow flow can be
timed. It may be necessary to adjust the
set screws, which act as limits to the
travel of the tilting buckets The average
of a minimum of ten tips should be used.
Adjustment is required if a 10 percent or
greater error is found or if greater
accuracy is needed.
2.7.6 Field Operation of a
Precipitation Measurement System
Calibration checks for weighing and
tipping bucket gages using the techni-
ques described above are recommended
at 6-month intervals Nonrecordmg
gages, whether alone or in a network,
should be read daily at a standard time
Although the weighing gage is used
for liquid and frozen precipitation, it
requires some special attention for
winter operations. First, the funnel
must be removed when snow is expected
Second, the bucket must be charged
with an antifreeze, 24 oz of ethylene
glycol mixed with 8 oz of oil. The weight
of this mixture represents the baseline
from which precipitation a mounts a re to
be noted. The bucket should be emptied
and recharged when necessary, at
about 5 inches in the Universal gage,
and at about 10 inches in the punched
tape gage All operational activities
should be recorded in the station log
2.7.7 Preventive Maintenance
Possible leaks m the measuring tube
or theoverflowcontainerofthe gage are
easily checked The receptacles are
partially filled with water colored with
red ink and placed over a piece of
newspaper This procedure isespecially
applicable to the clear plastic 4-mch
gage which is more easily damaged
Repairs are performed by soldering the
8-mch gage and by applying a solventto
the plastic.
A number of pens, some with greater
capacity than others, can be used with
the Universal gage All require occa-
sional cleaning, including a good
soaking and wiping in a mixture of water
and detergent After inking problems,
the next source of trouble is the chart
drive, but these problems can sometimes
be avoided by having the clock drive
lubricated for the environmental condi-
tions expected It is a good practice to
have spare clocks in stock
Routine visual checks of the perform-
ance of weighing type gages should
27
-------
be made every time there is a chart
change. The time and date of change,
and site location should be documented.
Routine maintenance should include
inking the pen and winding the clock.
Battery-powered chart drives will
require periodic replacement of batteries
based on either experience or manu-
facturer's recommendations. All PM
activities should be logged.
2.7.8 Audit Procedures
Refer to Section 2.2.4, "Audits," for
general guidelines. Audits on precipita-
tion measuring systems need be no
more frequent than every 6 months. The
irregular occurrence of precipitation
makes the use of a CTS impractical The
performance audit should depend upon
the challenging of the gage with
amounts of water known to an accuracy
of at least 1 percent of the total to be
used This method will provide an
accuracy of the measurement system
but not the collection efficiency of the
gage in natural precipitation. Fortipping
bucket gages use a rate of less than one
inch per hour and an amount which will
cause a minimum of ten tips.
For weighing gages, it is more
convenient to use calibration weights to
challenge the weighing mechanism
rather than using the gallons of water
necessary for full scale testing
28
-------
Section 3.0
Methods for Judging Suitability of Sensor Siting
3.1 Introduction
Although good instrumentation is a
necessity, proper site selection is
critical to obtain good meteorological
data. It is, from an absolute error point of
view, much more important than proper
placement of any other kind of air
monitoring equipment. Poor placement
can and has caused errors of 180° in
wind direction, and can cause major
errors in any other meteorological
variable, including wind speed, tem-
perature, humidity, and solar radiation.
The purpose of this section is to offer
guidance in assessing the suitability of
meteorological monitoring sites The
guidance given is based principally on
standards set by the World Meteoro-
logical Organization (WMO, 1971) and
Federal agencies involved m meteoro-
logical data collection (U.S. National
Weather Service [NWS] and the Ten-
nessee Valley Authority [TVA, 1977]).
Proper siting is part of the total quality
control program. Of course, as m many
other monitoring activities, the ideal
may not be attainable and, in many
urban areas where air quality studies
are traditionally done, it will be impossible
to find sites that meet all of the siting
criteria. In those cases, compromises
must be made. The important thing to
realize is that the data will be compro-
mised, but not necessarily in a random
way It is incumbent upon the agency
gathering the data to describe carefully
the deficiencies m the site and, if
possible, quantify or at least evaluate
the probable consequences to the data
3.2 Instrument Siting
The primary objective of instrument
siting is to place the instrument in a
location where it can make precise
measurements that are representative
of the general state of the atmosphere in
that area. Because most atmospheric
properties change dramatically with
height and surroundings, certain some-
what arbitrary conventions must be
observed so that measurements can be
compared. In this section, conventions
published by the World Meteorological
Organization (WMO, 1971) have been
adopted wherever possible
Secondary considerations such as
accessibility and security must be taken
into account, but should not be allowed
to compromise data quality.
3.2.1 Wind Speed and Direction
"The standard exposure of wind
instruments over level, open terrain is
10 m above the ground" (WMO, 1971).
Open terrain is defined as an area
where the distance between the instru-
ment and any obstruction is at least ten
(10) times the height of that obstruction.
An obstruction may be man-made (such
as a building) or natural (such as a tree)
(Figure 32.1).
The wind instrument should be
securely mounted on a mast that will
not twist, rotate, or sway.
If it is necessary to mount the wind
instrument on a roof of a building, it
should be mounted high enough to be
out of the area m which the air flow is
disturbed by the building. This is usually
1 5 times the height of the building. If it
is impossible to obtain the proper
unobstructed exposure through hori-
zontal relocation of the measurement
site, it may be necessary to raise the
height of the instrument so that it is out
of the wake of the obstruction. This is
not a good practice, however, and
should only be resorted to when
absolutely necessary. Sensor height
and its height above obstructions, as
well as the character of nearby obstruc-
tions, should be documented.
3.2.2 Temperature and Humidity
Temperature and humidity sensors
should be mounted over a plot of open
level ground at least 9 meters in
diameter. The ground surface should be
covered with short grass or, m areas
where grass does not grow, natural
earth. The surface must not be concrete
or asphalt. The standard height for
climatological purposes is 1 25 to 2 m,
but different heights may frequently be
required in air quality studies
The sensors should not be closer to
obstructions such as trees and/or
buildings than a distance equal to four
times their height. They should be at
least 30 m from large paved areas and
not close to steep slopes, ridges, or
hollows. Areas of standing water should
also be avoided.
Louvered instrument shelters should
be oriented with the door opening
toward true north.
3.2.3 Radiation
Solar and whole sky radiation mea-
surements should be taken in a location
free from any obstruction to the mea-
10m
-10H
10T
Figure 3.2.1. Siting wind instruments; a 10-mtower. located at least Wtimes
the height of obstructions away from those obstructions (not to
scale).
29
-------
surements. This means there should be
nothing above the horizontal plane of
the sensing element that would cast a
shadow on it. Neither should the
instrument be near light colored walls
or artificial sources of radiation. Usually
a tall platform or roof of a building is the
most suitable location.
3.2.4 Precipitation
A ram gage should be mounted on
level ground so that the mouth or
opening is horizontal The gage should
be shielded from the wind but not put in
an area where there will be excessive
turbulence caused by the shield. For
example, a good location would be an
opening in an orchard or grove of trees
where the wind speed near the ground
is reduced due to the canopy effect, but
a location that is mostly open except for
one or two trees would not be good
because of the strong eddies that could
be set up by the trees. This admittedly
requires a good deal of subjective
judgment but it cannot be avoided
Obstructions to the wind should not be
closer than two to four times the
obstruction height from the instrument.
In open areas, a wind shield such as
that used by the U S National Weather
Service should be used.
The ground surface around the rain
gage may be natural vegetation or
gravel It should not be paved, as this
may cause splashing into the gage. The
gage should be mounted a minimum of
30 cm above the ground and should be
high enough so that it will not be
covered by snow.
3.2.5 Meteorological Towers It is
frequently necessary to measure some
meteorological variables at more than
one height. For continuous measure-
ments or where the height requirement
is not too restrictive, towers may offer
the most advantageous measurement
platform
Towers should be located in an open
level area (see Table 321) representa-
tive of the area under study. In terrain
with significant topographic features,
different levels of the tower may be
under the influence of different mete-
orological regimes at the same time
Such regimes should be well docu-
mented
Towers should be of the open grid
type of construction, such as istypicalof
most television and radio broadcast
towers. Enclosed towers, stacks, water
storage tanks, grain elevators, cooling
towers, and similar structures should
not be used (Mollo-Chnstensen, 1979)
Towers must be rugged enough so that
they may be safely climbedto install and
service the instruments Folding or
collapsible towers that require the
instruments to be serviced or calibrated
at the ground are not acceptable
because they are usually not rigid
enough to ensure that the instruments
will be in the proper orientation.
Instruments should be mounted on
booms projecting horizontally out from
the tower. Wind instruments should be
mounted on a boom that will hold it a
distance equal to twice the maximum
diameter or diagonal of the tower away
from the nearest point on the tower. The
boom should project into the prevailing
wind (e.g., toward the north if the
prevailing wind is from the north) if this
can be done ambiguously. One may
wish to consider having two sets of
instruments at each level, located on
opposite sides of the tower. A simple
automatic switch can choose which set
of data to use (NASA, 1968). Docu-
mentation of the tower should include
the orientation of the booms.
Temperature sensors must be mounted
on booms to hold them away from the
tower, but a boom length equal to the
diameter of the tower is sufficient.
Temperature sensors should have
downward facing aspirated shields.
The booms must be strong enough so
that they will not sway or vibrate
excessively in strong winds The best
vertical location on the tower for the
sensors is at a point with a minimum
number of diagonal cross members, and
away from major horizontal cross
members Even with these precautions,
data obtained while the wind blows
from the sector transected by the tower
may not be free from error
These instrument siting suggestions
may seem to preclude the use of many
air monitoring sites that otherwise
would be desirable, but this need not be
the case. In siting air quality monitors
that are to be used for long-term trend
analysis or large geographic area
coverage, it may be perfectly acceptable
to have some or all of the meteorological
equipment at a different location that
better meets the large-scale require-
ments of the study As long as both sites
are in the same area of interest and
meet their respective siting criteria, this
should present no problems. When the
air quality data are to be used for short-
term diffusion model validation or
studies or short-term levels from
specific sources, however, a meteoro-
logical station should be located in the
vicinity of the air quality sensor.
3.3 Station Siting
Besides care in selecting the local
environment of a meteorological sensor,
it is important that care be taken in
selecting station location with respect
to major man-made and topographic
features such as cities, mountains,
large bodies of water, etc. Meteorological
variables are obviously affected by the
large-scale surrounding features. The
effect of cities has been studied exten-
sively (Ito, 1972; Vukovich, 1971; U S.
PHS, 1961). Documented effects include
a decrease in an average wind speed,
decrease in atmospheric stability,
increase in turbulence, increase in
temperature, and changes in precipita-
tion patterns. These changes will
obviously have an effect on the evalua-
tion and interpretation of meteorological
and air quality data taken in an urban
area.
Even more pronounced are the effects
of large natural features (Slade, 1968)
Besides their obvious effecton humidity,
oceans and large bodies of water are
usually at a different temperature than
the nearby land. This generates the well
known land- and sea-breezes which, in
many coastal areas, dominate the wind
patterns. There are also often simulta-
neous differences in cloud cover between
the oceans and nearby land surfaces.
This difference in thermal lag, insolation,
and changes in surface roughness and
vertical temperature structure can have
a profound effect on atmospheric
stability (SethuRaman, 1974)
The effects of mountains and valleys
on meteorological variables and atmos-
pheric dispersion continue to be
studied Two of the more interesting
recent papers are by Kao (1974) and
Hunt (1978). Well-known effects include
the channeling of flow up or down a
valley, the creation of drainage flow, the
Table 3.2.1. Limits on Terrain and Obstacles Near
Towers
Distance
from tower
fm)
0-15
15-30
30-100
100-300
Slope
(no greater than)
<%)
±2
±3
±7
±11
Max. obstruction or
vegetation height
(m)
0.3
0.5-1 0 (most veg. <0 3)
3.0
10 xht. must be less than
distance to obstruction
Source. TVA, 1977
30
-------
establishment of lee-waves, and an
increase in mechanically generated
turbulence. All of these effects and
others can play a major role in the
transport and dispersion of pollutants.
The important point is that almost any
physical object has an effect upon
atmospheric motion. In fact, it is
probably impossible to find a site that is
completely free from obstruction. This
being the case, it is the responsibility of
the person choosing a monitoring site to
have in mind the various forces at work
and to choose a site that will be most
representative of the area of interest. If
the area is a valley or a sea coast, then
the meteorological instruments should
be in that valley or near the coast; not on
a nearby hilltop or inland 30 km at a
more convenient airport site. Of course
one must also keep in mind the vertical
structure of the atmosphere. Winds
measured at the bottom of a 100 m
valley will not represent the winds at the
top of a 200 m stack whose base
happens to be in that valley
The choice of a station for meteoro-
logical data collection must be made
with a complete understanding of the
large-scale geographic area, the sources
being investigated, and the potential
uses of the data. Then rational, informed
choices can be made
Once they are made, the site should
be completely documented This should
include both small- and large-scale site
descriptions, local and topographic
maps (1.24,000 scale), photographs of
the site (if possible), and a written
description of the area that is adequately
represented by this site. This last point
is most important for it will allowa more
rational interpretation of the data. It
might state, for example, that a site
adequately represents a certain section
of a particular valley, the suburban part
of a given city, or several rural counties.
Whatever it is, the nature of the site
should be clearly described in a way that
will be clear to those who will be using
the data in the future.
31
-------
Section 4.0
Meteorological Data Validation
4.1 Methods
Once data are collected, they should
be reviewed to screen out possible
incorrect data points before they are put
into accessible storage or passed on to
the user. While the purpose of a QA
program is to avoid the generation of
bad data, it is impossible to do so
completely. Even in the best planned
and best conducted programs, unde-
tected errors can be generated by faulty
equipment, noisy data transmission
lines, faulty key punching, and a myriad
of other sources Filippov (1968) offers a
detailed and thorough discussion of the
various possible sources of error
In both automatic (ADP) and manual
data screening the most obvious checks
should be performed first These include
such things as being sure that the data
exist and are properly identified, the
forms or files are filled out properly, that
numbers are in the blocks where they
should be, letters are where they should
be, and blanks exist where nothing
should be This sort of data editing is a
subject unto itself and will not be
pursued here
Methods of editing or screening
meteorological data usually involve
comparison of the measured value with
some expected value or range of values
Techniques for checking the measured
value usually fall into one or more of the
following categories
1 Comparison with upper and/or lower
limit on the allowed range of the data,
2 Comparison with known statistical
distribution of the data,
3 Comparison with spatial and/or
temporal data fields; and
4 Comparison based upon known
physical relationships
A choice must also be made of what to
do with the datum that does not pass a
validation procedure. Basically there
are two choices, eliminate the ques-
tionable data from the file, or flag it for
further examination. Automatically
discarding data may be a viable, cost-
effective option if the screening proce-
dure is carefully designed and each
datum is not of high value Records
must be kept of discarded data so the
reason for the fault can be found and
corrected Flagged data are examined
and a decision madeontheiracceptabil-
ity. If unacceptable, it may be possible to
correct them or substitute a more
reasonable value (Reynolds, 1979)
Corrected or substituted values should
be so indicated in the data file, with an
explanation of the substitution available
to the user Alternatively, data of
questionable value may be kept in the
data file under a flagged status, with a
notation of why they are questionable,
so that the user can make a decision as
to their usefulness This procedure is of
questionable value to most users
because the collecting agency is fre-
quently in the best position to make a
decision on the data
The range test is the most common
and simplest test Data are checked to
see if they fall within specified limits
The limits are set ahead of time based
usually upon historical data or physically
impossible values Some examples of
reasonable range tests are rainfall rate
greater than 10 in /hour or wind
direction not between 1° and 360° In
setting the limits, one must take into
consideration whether or not the
system will select only outrageous,
extreme (i e , impossible) values usually
caused by data handling errors (such as
wind speeds greater than 100 m/s or
less than zero) or just unusually high
(i.e , possible) values, which should be
examined further This may require a
further decision on just how extreme a
value should be flagged This decision
should be based on the real impact of
using extreme values should they be in
error Considerations of the cost of
incorrect data, the possibility of correc-
tion or substitution, or replacement by
obtaining new data should be made
Unfortunately, the decision may also
frequently be made on the available
resources of those who examine the
flagged data
Comparison with known statistical
distributions may involve comparison of
means, standard deviations, means of
extremes, or higher order statistics For
example, Lee and Stokes (1978) report
thattheirdata base usually had kurtosis
of approximately 3 with zero skewness
Any of their instruments, then, that
showed a marked departure from these
values were considered to be in need of
further verification (Additional research
is needed to determine whether these
or similar criteria could be used in other
areas )
Lockhart (1979) suggests compressing
data into a densely packed graph where
long-term (week, month, or seasonal)
patterns can be easily seen. Major
departures from these subjectively seen
patterns can be noted and the data
checked Although this method of data
verification is usually used to check a
particular data set against a longer term
climatology, it can also be used to check
individual values For example, one
might compare a temperature reading
with the monthly average maximum or
minimum plus or minus (respectively)
two or three standard deviations This
technique obviously depends on a
reliable history or representative mea-
surements being available from the site
and is ineffective for noting significant
long-term changes in the instrument
Screening data by comparison with
fields- of similar or related data is
commonly done when large amounts of
data are taken and when assumptions
of spatial continuity of the meteorological
variable are physically reasonable The
most easily visualized example of this is
a field of atmospheric pressure mea-
surements Any value can be compared
with those in a large area around it,
either visually, or by numerical inter-
polation Major deviations from the
oommant pattern (a low pressure
reading in the middle of a high pressure
area) are not to be expected Of course,
allowance must be made for meso- and
micro-scale phenomena such as a
shortwave or pressure jump area ahead
of a convective storm
Not all meteorological fields can be
expected to have the needed continuity
Rainfall is a notorious example of
discontinuity or microscale variations
Wind speed and direction can exhibit
continuity on some spatial scales, but
care must be taken to account for the
many effects, such as topography, that
can confuse the issue (See Section 3.)
Interrelated fields can also be used to
screen data Rainfall, lor example, is
unusual without clouds and high
humidity while wind direction and
speed, especially above the surface
layer, are related to pressure gradients
Fields of data in time, rather than
space, are also used to check datum
points These checks are usually made
on rates of change of the data Checks
are made both on rates of change that
are too high and not high enough For
example, atmospheric stability is not
expected to change by several classes
within an hour A wind direction
reading, however, that does not change
at all for several hours may indicate that
32
-------
the vane is stuck (assuming the wind
speed is not zero) or that there is some
other problem with the system.
Screening checks can also be made to
assure that physically improbable
situations are not reported in the data
This kind of check is not commonly used
because of the wide variety of conditions
that can occur in the atmosphere under
extreme conditions. These unusual
events would frequently be noted first
by some of the statistical or range
checks noted above
Some screening points of this type
that are used include assuring that the
dew point is not greater than the
temperature, and that the lapse rate is
not greater than the autoconvective
lapse rate. Checks on stability class
versus time (not allowing "strongly
unstable" at night or "stable" during
the day) may also be considered in this
category.
Table 4.1.1 gives examples of some of
the data editing criteria used by three
Federal agencies: the National Climatic
Center, Klmt (1979); the Nuclear
Regulatory Commission, Fairobent
(1979); and the Tennessee Valley
Authority, Reynolds (1978) Examination
of the table shows some interesting
differences that can be ascribed to the
differing missions of the agencies
Because of their global concerns, the
NCC must allow a far wider range of
limits on fields such as temperature and
humidity than does an agency with
only local interest, such as TVA. On the
other hand, the NCC has the data
available to do spatial checks over a
wider area than would be possible for
many local study situations Differences
can also be noted depending on the type
of data collection (spot readings once
per hour or three hours versus con-
tinuous recordings) and major interests
(synoptic weather patterns versus
stability) Fihppov (1968) gives an
exhaustive review of checks used by
weather services of many other coun-
tries
4.2 Recommendations
The following is a data validation
system recommended for EMSL/RTPto
replace the present sytem for screening
meteorological data. It could be used to
screen data gathered by EMSL, contrac-
tors, or state and local agencies. The
system takes into account the variable
nature of EMSL's field activities. It does
not depend on, or have the advantages
of, long-term multistation network
design, nor is it labor intensive The
basic goal is that of rapid identification
of field problems, with low value
assigned to individual data points, thus
allowing the discard of questionable
values. Flexibility is available, however,
if an individual project's meteorological
data are judged to warrant a more
criticial approach.
The flow of the system is shown in
Figure 4.2.1. All data will go first
through a hard copy auditing-procedure
designed to find data entry and keypunch
errors. In the hard copy audit, a
percentage of data points will be
randomly selected for audit. A second,
independent file of these values, as well
as the hour just before and after the
hour, will be created from the original
hard copy This file will be compared
with the master file and discrepancies
noted. If there are only a few random
discrepancies, these points will be
eliminated from the system. If there are
several, or there seems to be a systematic
pattern of errors, the project office (the
office responsible for gathering and
reducing the data) will be notified so
that they can correct and re-enter the
data and correct the data entry system.
The data are next passed through a
screening program, which is designed
to note and flag questionable values.
Flagged data will go to the laboratory
meteorological office for review There
they will either be accepted, discarded,
or returned to the project officer if there
is a large amount of questionable data.
That officer may accept, discard, or
correct the data. The screening values
are given in Table 4.2 1 They offer a
combination of range, rate of change,
and physical impossibility checks that
are chosen to be reasonably restrictive.
It is anticipated that some good data will
be flagged, but that most data handling
and gross instrument failure problems
will be caught.
Data that pass the screening program
will go through a comparison program
This program will randomly select
certain values for manual comparison
with information collected by the
National Weather Service In the
selection process, one day and one hour
will be chosen on which data from all
stations in the network will be audited
One day in every 20 will be randomly
chosen, and on that day, one hour
between 5 and 9am (EST), will be
randomly chosen. Data from that hour
and day for all stations will be printed
out by the audit program for the manual
checks. The program will also make
compressed time scale plot (20 days of
hourly values on one line) for each
parameter for the use of the validators.
The data generated by the audit
programs will first be compared with
National Weather Service data to see if
they fit reasonably well with synoptic
conditions prevalent in that area The
meteorologist will choose the stations
to be used in the verification, and train
the data clerks in the subjective
comparison procedure. All questionable
data will be given to the meteorologist
for review as above. The variables to be
checked in this way will include wind
speed and direction, temperature, dew
point, pressure, and occurrence of
precipitation.
Naturally, if the audit checks show a
problem with one or more instruments,
an attempt will be made to identify the
time range of that problem so that all
questionable data can be found. Logs of
bad data will also be kept and used to
identify troublesome instruments and
other problems.
This system is suggested principally
for EMSL/RTP, but may prove a useful
starting place for state and local air
pollution agencies wishing to develop a
meteorological data validation procedure.
The suggested system is very complete
and will be evaluated over a period of
time Changes to the system may have
to be made, depending upon the needs
and resources of the users
33
-------
Table 4.1.1. Examples of Data Editing Criteria
Data are edited or challenged for further review in various systems if the editing criteria are met or exceeded.
Wind speed
>25 m/s (NRC)
>50 kts (NCC)
>20 kts and doubles at next 3-hour observation (NCC)
First 5 hourly values within ±0.2 m/hour of next 4 (TVA)
Wind direction
Any recorded with calm wind speed (NCC)
Same sector for more than 18 hours (NRC)
First 5 within ±0.2° of next 4 (TVA)
Temperature
9°F > mean daily maximum for the month (TVA)
9°F < mean daily minimum for the month (TVA)
>10°F change in 1 hour at a site (TVA)
First 5 hours within ±0.5°F of next 4 (TVA)
>125°F (NCC)
<-60°F (NCC)
> 70°f change in 1 hour or 20°F change m 3 hours (NCC)
Same temperature 12 or more hours (NRC)
Dew point
Dew point > temperature (TVA, NRC)
Dew point change > 7°F m 1 hour (TVA)
First 5 hours within ±0.5°F of next 4 (TVA)
>90°F (NCC)
<-60°F (NCC)
Temperature - dew point > 5°F during precipitation (NRC)
Temperature = dew point more than 12 consecutive hours (NRC)
Pressure
>1060 mb (sea level) (NCC)
<940 mb (sea level) (NCC)
Station pressure (inches Hg) + elevation (feet) x 10 ~3 <27.75 or >31.3 inches Hg (NCC)
Change of 6 mb or 0.2 inch Hg in 3 hours (NCC)
Vertical temperature gradient and stability
A77Az > 7°CY 700 m between 10 am and 5 pm (TVA)
ATVAz < -1°C/100 m between 6 pm and 5 am (TVA)
A7VAz > 15°C/100 m (TVA)
A7~/Az < -3.4°C/100 m (autoconvective lapse rate) (TVA. NRC)
ATVAz changes sign twice in 3 hours (TVA)
A, B, F, or G stability during precipitation (NRC)
F or G stability during the day (NRC)
A, B, or C stability during the night (NRC)
Change in stability of more than 3 classes between 2 consecutive hours (NRC)
Same stability for>12 hours (NRC)
NCC - National Climatic Center (Klint, 1979)
NRC = Nuclear Regulatory Commission (Fairobent, 1979)
TVA - Tennessee Valley Authority (Reynolds, 1978)
34
-------
Start
t*
Edit/Merge
Program
<
1
Correct Data
Acquisition/
Processing
Procedures
/ Unvalidated
\ File
Comparision of
Selected Data
with Data Obtained
Independently from
Original Source
Are Bad Data
Indicative of
Recurring
Problem?
Project Office
Determine if Suspect
Data are to be:
Approved.
Corrected.
or Deleted.
Screening Program
Flagged
Data
Comparison of
Subset of Data
with N W.S. Data
Meteorological
Office
Determine if Data
Approved
Corrected.
or Deleted.
Questionable
Data
Unresolved
Data
Data Corrections,
Approvals, or
Deletions.
Figure 4.2.1. Schematic flow of decisions
in EMSL data validation
scheme.
Table 4.2.1. Suggested EMSL Screening Criteria
Data should be flagged if they meet or exceed the following criteria
Wind speed
>20 m/s (1 hour average)
Unchanged for 12 or more consecutive hours (within ±0.5 m/s)
Wind direction
Any reported with calm or no wind speed
Same 10° sector for 18 or more consecutive hours
Temperature
>45°C
<-35°C
>+5° or <-5°C change/1 hour
Unchanged for 12 or more consecutive hours (within ± 0.5°C)
Dew point
> temperature at that hour (same location)
>+3°C or <-3°C change/1 hour
Unchanged for 12 or more consecutive hours (within 0.5°C)
= temperature for 12 or more consecutive hours (within 0.1°C)
Pressure
>1060 mb (sea level)
<940 mb (sea level)
>+6 or <-6 mb change/3 hours
Rainfall
>15 cm/24 hours
<5 cm/3 months
35
-------
Section 5.0
Bibliography
Acheson, D T , 1963. Some limitations
and errors inherent in the use of dew-
cell for measurement of atmospheric
dew points. Monthly Weather Review.
American National Standards Institute,
1979: Standard for obtaining meteorol-
ogical information at nuclear power
sites ANS-2.5, N 179, (draft)
Arizona Department of Health, 1977
Quality Control Procedures State of
Arizona
Bahm, Raymond J., 1977 Instrument
errors in National Weather Service solar
radiation data Proceedings of the
Annual Meeting: International Solar
Energy Society, pp 1417-1421
Bauer, D. A., and G. A. Cresswell, 1 975
Quality assurance for meteorological
and air quality studies in support of the
Rio Blanco oil shale project. Quarterly of
the Colorado School of Mines, 70(4),
187-198
Bergman, Kenneth H, 1978. Role of
observational errors in optimum inter-
polation analysis Bulletin of the Ameri-
can Meteorological Society, 59 (12),
1603-1611
Bryan, R. J et al , 1975. Guidelines for
enforcement and surveillance of supple-
mentary control systems EPA-340/1-
75-008, U S Environmental Protection
Agency.
Carter, E A etal., 1977 Catalog of solar
radiation measuring equipment ERDA/
ORO/5362-1, U.S Energy and Deve-
lopment Administration
Champ, D H , and R C. Bourke, 1978:
The role of instrument calibration in
data quality assurance Presented at the
4th Symposium on Meteorological
Observations and Instrumentation,
American Meteorological Society, Den-
ver
Conover, J H, 1950 Tests and adapta-
tion of the Foxboro dew-point recorder
for weather observatory use Bulletin of
American Meteorological Society, 31(1),
13-22
Craig, R, and W David Zittel, 1 974 The
NSSL/WKY-TV Tower Data Collection
Program- April-July 1972 NOAA,
National Severe Storms Laboratory
Crutcher, Harold L, 1970: Centralized
quality control and evaluation programs
Meteorological Monographs, 11(33),
137-140.
Davey, F., 1965 Hair humidity elements.
Humidity and Moisture, edited by A
Wexler, Remhold Publishing Co
Deacon, E L, 1968 The leveling error
in Reynolds stress measurement Bulle-
tin of the American Meteorological
Society, 49(8), 836
Eddy, Amos, 1970- The statistical
evaluation of observational data Mete-
orological Monographs, 11(33), 110-
120
Eppley Laboratory Instrumentation for
the Measurement of the Components of
Solar and Terrestrial Radiation Newport,
R I.
Essenwanger, Oskar M., 1970 Analyt-
ical procedures for the quality control of
meteorological data Meteorological
Monographs, 11(33), 141-147
Fairobent, James E, 1979 (personal
communication) Nuclear Regulatory
Commission.
Fmkelstem, Peter L, 1981 Measuring
the dynamic performance of wind
vanes Journal of Applied Meteorology,
20, pp 588-594
Francisco and Beaubien, 1965 An
automatic dew point hygrometer with
thermoelectric cooling Humidity and
Moisture, edited by A Wexler, Remhold
Publishing Company
Filippov, V V, 1968 Quality control
procedures for meteorological data
World Weather Watch Planning Report
No. 26, World Meteorological Organiza-
tion
, 1969: Quality Control procedures
for meteorological data World Meteor-
ological Organization Technical Note
No. 100, pp 35-38
Frost, Walter, and Trevor H Moulden,
1 977. Handbook of Turbulence, Plenum
Press
Gandm, L S., 1969 Statistical methods
for automatic check of meteorological
information World Meteorological
Organization Technical Note No WO,
pp 49-51
Gates, David M , 1962. Energy Exchange
in the Biosphere, Harper and Row
Gill, Gerald C. and Paul L Hexter, 1 972
Some instrumentation definitions for
use by meteorologists and engineers
Bulletin of the American Meteorological
Society, 53(9), pp. 846-851.
et al , 1967. Accuracy of wind
measurements on towers and stacks.
Bulletin of American Meteorological
Society, 48(9), pp 665-674
, 1 967: On the dynamic response
of meteorological sensors and recorders
Proceedings of the First Canadian
Conference on Micrometeorology, Part
I, Meteorological Service of Canada,
Toronto
, 1 973. The helicoid anemometer.
Atmosphere, ll(4).
, 1975 Development and use of
the Gill UVW anemometer Third Sym-
posium on Meteorological Observations
and Instrumentation, American Mete-
orological Society
Hamby, J W., 1979 Traceability in the
TVA Meteorological System Calibration
Program Presented at the Quality
Assurance in Air Pollution Measure-
ment Conference, Air Pollution Control
Association, New Orleans.
Hanna, S R., et al , 1977' Meeting
review AMS workshop on stability
classification schemes and sigma
curves—summary of recommendations
Bulletin of the American Meteorological
Society, 58(12), pp. 1305-1309
Hicks, W F , 1947- Humidity measure-
ment by a new system Refrigerating
Engineering. American Society of Refrig-
erating Engineering.
Hoehne, Walter E , 1973 Standardized
functional tests IEEE Transactions of
Geoscience Electronics, Vol G.E -1 1
Number 2, April
1977 Progress and results of
functional testing NOAA Technical
Memorandum, National Weather Service
T&EL-15
36
-------
Hofmann, Gustav, 1965: Hints on
measurement techniques used in mic-
rochmatologic and micrometeorological
investigations. The Climate Near the
Ground, Harvard University Press.
Humphrey, Paul, 1976: Guidelines for
Siting and Exposure of Meteorological
Instruments for Environmental Purposes.
Division of Meteorology, U S. Environ-
mental Protection Agency (draft).
Hunt, J C R., W. H Snyder and R. E
Lawson, Jr, 1978 Flow structure and
turbulent diffusion around a three
dimensional hill parti EPA-600/4-78-
041, U.S Environmental Protection
Agency
Huschke, R., ed , 1959: The Glossary of
Meteorology. American Meteorological
Society.
Ito, Masashietal., 1972 An examination
of local wind measurements in cities
Annual Report of the Tokyo Meteoro-
politan District Public Damage Research
Institute, 3(3), pp. 27-32 (APTIC 41974
TR 186-73).
Kaneshige, T M., and P West, 1969:
Deviational Analysis. U S. Air Force
Global Weather Central, Offutt, Nebraska.
Kao, S K, H N Lee and K I Smidy,
1974 A preliminary analysis of the
effect of mountain-valley terrains on
turbulence and diffusion Proceedings
from Symposium on Atmospheric Diffu-
sion and Air Pollution, Santa Barbara,
California, American Meteorological
Society, pp 59-63
Klmt, William E., 1979' Screening
checks used by the National Climatic
Center for meteorological data NCC,
Asheville, North Carolina, (unpublished)
Kraus, E B, 1968 Reply to Deacon
Bulletin of the American Meteorological
Society, 49(8), 836
Latimer, J Ronald, 1972- Radiation
measurement Technical Manual Series
No 2, International Field Year for the
Great Lakes, Canadian National Com-
mission for the Hydrological Decade
Laznow, Joseph, 1978 A centrally
located data acquisition and processing
system for a network of remote-sensing
meteorological towers at existing and
proposed nuclear power plant sites
Preprint 4th Symposium on Meteoro-
logical Observations and Instrumenta-
tion, American Meteorological Society,
pp 99-102
Lee, J T , and Judith Stokes, 1 978 Tall
tower and aircraft instrumentation
quality control procedures—develop-
ment and application Preprint 4th
Symposium on Meteorological Obser-
vations and Instrumentation, American
Meteorological Society, pp 19-24
List, R., 1971: Smithsonian Meteoro-
logical Tables. 6th rev ed., Smithsonian
Institute Press
Lockhart, T J , 1970: Bivanes and direct
turbulence sensors. MRI 70 Pa-928, to
EPA Institute for Air Pollution Training,
June.
1978: A field calibration strategy
for rotating anemometers and wind
vanes Presented at 4th Symposium on
Meteorological Observations and In-
strumentation, American Meteorology
Society, Denver.
, and M Gannon, 1 978'Accuracy
and precision of field calibration of
temperature difference systems Pre-
sented at National Conference on
Quality Assurance of Environmental
Measurement, Denver
,1979 Data graphics for assess-
ment of measurement quality Presented
at Quality Assurance in Air Pollution
Measurements Conference, Air Pollution
Control Association, New Orleans
1979 Quality assurance of
temperature and temperature gradient
data from monitoring systems Presented
at Quality Assurance in Air Pollution
Measurements Conference, Air Pollution
Control Association, New Orleans
MacCready, P B., 1965 Dynamic
response characteristics of meteoro-
logical sensors Bulletin of American
Meteorological Society, 46(9), pp 533-
538
, 1 966 Mean wind speed mea-
surements in turbulence. Journal of
Applied Meteorology, 5, pp 219-225
Mazzarella, D A., 1972 An inventory of
specifications for wind measuring
instruments Bulletin of American
Meteorological Society, 53(9), pp. 860-
871
1977 Meteorological instru-
ments their selection and use in air
pollution studies Proceedings of the
Meeting on Education and Training in
Meteorological Aspects of Atmospheric
Pollution and Related Environmental
Problems, World Meteorological Or-
ganization, No 493
1978 Meteorological instru-
ments for use near the ground their
selection and use m air pollution
studies Air Quality Meteorology and
Atmospheric Ozone, edited by Morris
and Barras, American Society for
Testing and Materials
McKay, D J , and J. D. McTaggert-
Cowan, 1977' An intercompanson of
radiation shields for auto stations
World Meteorological Organization
Publication No 480, pp. 208-213
McTaggart-Cowan, J D , and D J
McKay, 1976 Radiation shields - an
intercompanson Canadian Atmospheric
Environment Service Report.
Middleton, W E. K.andA F Spilhaus,
1953. Meteorological Instruments,
University of Toronto Press.
Mollo-Chnstensen, E, 1979 Upwind
distortion due to probe support in
boundary layer observations Journal of
Applied Meteorology, 18(3), pp 367-
370.
Monteith, J L, 1972 Survey of instru-
ments for micrometeorology Interna-
tional Biological Programs Handbook
No 22, Blackwell Scientific Publications,
Osney Mead, Oxford, England
, 1975 Vegetation and the
Atmosphere Academic Press, London
National Aeronautics and Space Ad-
ministration, 1968 Meteorological
Measuring and Recording Equipment
Description, Calibration and Mainte-
nance Procedures for NASA's 150
Meter Meteorological Tower Facility,
Kennedy Space Center, Florida, GP-465
Norris, D J, 1973 Calibration of
pyranometers Solar Energy, 14(2), pp
99-108, Oxford
Pokrouskaya, I A, 1967. Calibration
of Meteorological Instruments. Glavna-
ya Geofizicheskaya Observatonya,
Trudy, Leningrad.
Pond, S, 1968 Some effects of buoy
motion on measurements of wind
speed and stress. Journal of Geophys-
ical Research, 73(2), pp 507-512
Qumn, F C, 1963 Humidity - the
neglected parameter Testing Engineer-
ing. The Mattmgly Publishing Company,
Inc
Rockwell International, Inc, 1977
Meteorological quality control RAPS
Report No AMC7010 25 SFPM-F
Reynolds, G W, and D E Pittman,
1978 The TVA meteorological data
acceptance analysis program Proceed
ings 4th Symposium on Meteorological
Observations and Instrumentation.
American Meteorological Society, pp
3-6
37
-------
, 1979: Final acceptance review
for TVA meteorological data. Presented
at Quality Assurance in Air Pollution
Measurements Conference, Air Pollution
Control Association, New Orleans.
Rhodes, R. C., and S. Hochheiser, Ed.,
1977. Data validation conference
proceedings. EPA 600/9-79-042,
U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
Schiermeier, F A., 1979: Collection and
validation of upper air meteorological
data for the regional air pollution study
(RAPS) Presented at the Quality Assur-
ance in Air Pollution Monitoring Con-
ference, Air Pollution Control Associa-
tion, New Orleans, Louisiana.
Segum, W R , et al , 1977: U.S
National Processing Center for GATE:
B-Scale surface meteorological and
radiation system National Oceanic and
Atmospheric Administration Technical
Report EDS 22.
SethuRaman, S , R. M. Brown, and J.
Tichler, 1974 Spectra of atmospheric
turbulence over the sea during stably
stratified conditions. Symposium on
Atmospheric Diffusion and Air Pollution,
American Meteorological Society, Santa
Barbara, California, pp 71-76.
Slade, David H., 1968' Meteorology and
Atomic Energy. U.S. Atomic Energy
Commission, TID-24190.
Slob, W H , 1973 Temperature error of
a ventilated thermometer in dependence
of radiation and wind Scientific Dis-
cussions, CIMO VI, World Meteorological
Organization
Strimaitis, David, G Hoffnagle and A
Bass, 1981. On-site meteorological
instrumentation requirements to char-
acterize diffusion from point sources
EPA-600/9-81-020, C-1-C-5
Suntola and Antson, 1973 A thin film
humidity sensor. Scientific Discussions,
CIMO VI, World Meteorological Organ-
ization
Talvmen, Timo, 1970 Equipment main-
tenance as part of the quality control
program for observations. Meteorologi-
cal Monographs, 7(33), pp 148-152
Tennessee Valley Authority, 1977
Criteria for meteorological measure-
ment site acceptance and/or preparation
(unpublished)
Ueda, Fumio, 1976' On the quality
control of marine meteorological data
practices in Japan Meteorological
Agency The OceanographicalMagazine,
27(12), pp 43-56
U.S. Air Force, 1977: Production Unit
Quality Control AWS Reg. 178-1
Management Analysis Program, Chap-
ter 2, Air Weather Service.
U.S. Army, 1974: Part 1, basic environ-
mental concepts. Engineering Design
Handbook, Environmental Series, De-
partment of the Army, Material Com-
mand
1975 Part 2, natural environ-
mental factors Engineering Design
Handbook, Environmental Series, De-
partment of the Army, Material Com-
mand
U.S Department of Commerce, NOAA
Edit Procedures-Surface Observational
Data, National Climatic Center, (unpub-
lished)
, 1977: Wind calibration sheet
Engineering Handbook No 8, Section
2.2, Note No. 26, National Weather
Service
1970: Engineering quality
control inspections Engineering Hand-
book No. 12, National Weather Service.
, 1972 Sub-station observations
National Weather Service Handbook
No. 2
U S Department of Commerce, Defense,
and Transportation, 1979 Surface
observations Federal Meteorological
Handbook No 1
U S Environmental Protection Agency,
1976 Quality assurance handbook for
air pollution measurement systems,
Vol l-Prmciples EPA-600/9-76-005,
Office of Research and Development
1980 Ambient monitoring
guidelines for prevention of significant
deterioration (PSD) EPA-450/4-80-
012, Research Triangle Park, North
Carolina
, 1980 Interim guidelines and
specifications for preparing quality
assurance project plans (draft). Research
and Development, Office of Monitoring
Systems and Quality Assurance, QAMS-
005/80, Washington, D C , 20460
U S Government Services Administra-
tion Code of Federal Regulations
Office of Federal Registry, 19(50),
Appendix B
US Public Health Service, 196V Air
Over Cities A symposium, Cincinnati,
Ohio
Vukovich, F. M , 1971 Theoretical
analysis of the effect of mean wind and
stability on a heat island circulation
characteristic of an urban complex.
Monthly Weather Review, 99(12), pp.
919-926.
Wang, J. Y, 1975: Instruments for
Physical Environmental Measurements.
Millieu Information Service.
Weiss, Leonard L., 1961 Relative
catches of snow in shielded and
unshielded gages at different wind
speeds Monthly Weather Review, Vol
89.
Wiermga, J., 1967: Evaluation and
design of wind vanes. Journal of
Applied Meteorology, 6(6), pp 1114-
1122
Woodward, Keith, 1975: Sensitivity of
atmospheric diffusion estimates to
meteorological data recovery and accu-
racy. Nuclear Technology, 25, pp. 635-
639.
World Meteorological Organization
(WMO), 1971 Guide to meteorological
instrument and observing practices
World Meteorological Organization No
8TP3, 4th edition, Geneva, Switzerland
38
-------
5" 9
13 I^1
-
>
00
O V/l V^> 3
X W O a
•- o r~ •
o sr
> c« -c c|
ffl • O
o
«• f^
r-
>- >
f 3O
7?
& V>
O -4
O- •
O
rv
c-
o
O
(J*
O)
•o
CE
o
03 C
O 2
o ir
^ Q
QJ
C/l
3)
0)
------- |