United States
Environmental Protection
Agency
Atmospheric Research and ^
Exposure Assessment Laboratory -y
Research Triangle Park NC 27711 /
Research and Development
EPA/600/S3-88/043 Mar. 1989
&EPA Project Summary
Meteorological Processor for
Regulatory Models
(MPRM-1.1)-User's Guide
John S. Irwin, James O. Paumier, and Roger W. Brode
Version 1.1 of MPRM provides a
general purpose computer processor
for organizing available meteor-
ological data into a format suitable
for use by air quality dispersion
models. Specifically, the processor is
designed to accommodate those
dispersion models that have gained
EPA approval for use in regulatory
decision making. A unique feature of
the processor is the ability to employ
user collected meteorological
measurements as well as those
routinely collected by the National
Weather Service (NWS).
MPRM-1.1 will support the
following dispersion models listed in
the Guideline on Air Quality Models
(Revised) (EPA, 1986), as well as
three screening models:
• Those requiring RAMMET for-
matted data: BLP, RAM, ISCST,
MPTER, CRSTER, and COMPLEX1.
• Those requiring STAR formatted
data: COM (with either 16 or 36
wind direction sectors), ISCLT,
and VALLEY (long-term).
• Those requiring special formats:
CALINE-3 and RTOM (default).
MPRM can be envisioned as a
three-stage system. The first stage
retrieves the meteorological data
from computer tape or disk files and
processes the data through various
quality assessment checks. The
second stage collects all data
available for a 24-hour period (up-
per air observations, hourly surface
weather observations, and data col-
lected as part of an on-site
meteorological measurement pro-
gram) and stores these data in a
combined (merged) format The third
stage reads the merged meteor-
ological data and performs the
necessary processing to produce a
meteorological data file suitable for
use by the specified dispersion
model.
This Project Summary was devel-
oped by EPA's Atmospheric Research
and Exposure Assessment Laboratory,
Research Triangle Park, NC, to
announce key findings of the research
project that Is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
EPA has recently issued guidance on
the use of meteorological data, collected
via an on-site measurement program,
for regulatory modeling applications. The
meteorological processors currently
available from EPA do not have the
capability of processing user collected
on-site meteorological data as directed
by the guidance. Therefore, MPRM-1.1
has been designed to construct meteor-
ological data files of upper air, mixing
height, surface observations, and on-
site data for air pollution dispersion
models that are routinely used in
regulatory decision making by EPA.
Specifically, the processor is designed to
accommodate those dispersion models
recommended for use in the Guideline on
Air Quality Models (Revised in 1986).
-------
Why This Design?
Recognizing that the list of approved
dispersion models is apt to change, and
that those changes are likely to call for
use of various processing methods, the
new processor is designed to have a
highly modular structure. Additionally,
the new processor is designed to avoid
computer conflict problems. It was
further decided to develop an input
structure that would ultimately support
use of a menu data entry system. This
would allow construction of the input
through a computer-controlled question
and answer session, and possibly
facilitate the usability of the processor by
a variety of users.
What Does MPRM Do?
MPRM can be envisioned as a three-
stage processing system, as depicted in
Figure 1. During the first stage, the
processor extracts upper air, mixing
height, and surface data from the raw
data files delivered from NCDC and on-
site data from the raw data files
developed from the on-site meas-
urement program. The extracted data are
processed through a series of quality
assessment checks. As a result, reports
of missing and suspect values are
generated. During the second stage, the
processor combines the available data
for each midnight-to-midnight 24-
hour period (twice-daily upper air
soundings and mixing height data, hourly
surface weather observations, and hourly
on-site data) and stores these data in a
combined (merged) format. During the
third and final stage, the processor reads
the merged data and develops a
meteorological data file for the dispersion
model selected by the user.
Extraction and Quality
Assessment (Stage 1
Processing)
The goal of this first stage of pro-
cessing is to:
• Read the on-site and NWS meteor-
ological data files
• Find the data within the time period
specified by the user
• Store these data in American
Standard Code for Information Inter-
change (ASCII) data files
• Scan the stored values and report
occurrences of missing or suspect
values.
MPRM can currently process hourly
surface observations in CD-144 format,
upper air soundings in TD-5600 format,
and mixing height data in TD-9689 for-
mat. Persons experienced with RAMMET
have a working knowledge of hourly
surface weather (CD-144) data and
twice-daily mixing height (TD-9689)
data. NCDC can provide CD-144 and
TD-9689 data on computer magnetic
tape or on 5 1/4-inch diskettes in a
format suitable for use in IBM compatible
personal computers (PCs). MPRM in-
stalled on a mainframe computer can
process these data either from the
computer tape or from mass storage data
files. MPRM installed on a PC can
process these data either from data files
on the hard disk or from 5 1/4-inch
diskettes. We would at least suggest an
IBM-AT or equivalent with 640K random
access memory (RAM) and a hard disk.
As of 1984, NCDC had converted to
new data storage formats for surface
I Meteorological Processor
for
Regulatory Models
Stage 1
Processing
Extract and Quality
Assess Data
Stage 2
Processing
Combine (Merge)
Data Files
Stage 3
Processing
Create Data
File for Modeling
Figure 1. Overview of processing stages within MPFtM.
observations and upper air soundings
namely, elemental formats TD-3280 an<
TD-6201, respectively. NCDC nov
converts from these formats to servic«
requests for CD-144 and TD-560(
formatted data. MPRM is designet
internally for ready adaptation to thes<
new formats. One, if not the first, upgrade
will expand the data formats supportec
by MPRM to include TD-3290 and TD
6201 formats.
Because there is no standard forma
for storage of on-site meteorologica
data, MPRM is designed to process i
variety of on-site data formats by having
the user define the structure of the inpu
data. The two major restrictions t<
consider in processing the on-site dat:
are 1) the order in which the data value;
are presented and 2) the data file mus
be a standard ASCII data file. In principle
MPRM will be able to process the on
site data as long as the data values fo
each observation are ordered correcth
(date and time, then meteorologica
values) and the observation can be rea<
using a FORTRAN FORMAT statement.
An additional capability of this firs
stage is assessing the quality of the dat.
by checking for possible missing o
suspect values. Any occurrences c
missing or suspect data values an
reported before the upper air soundings
mixing height data, surface observations
and on-site data are combined.
The output files from this first stage c
processing can be edited using standan
text editors routinely available oi
computer systems. The only foreseeabd
problem in editing these data files is the
some text editors have a limitation on th
size of the file to be edited. This probler
is of most concern when the editing i
performed on a PC where the text editc
is typically limited to the available RAN/
For example, because of the RAJ
limitation, a file consisting of 700 Kb of
year of hourly surface observations i
CD-144 format could not be edited. .
possible solution to this problem is t
break the file into parts that can be edite
with software designed for this purpose
The modified larger file is recreated b
concatenating the smaller files.
Combining Data (Stage 2
Processing)
The goal of this second stage c
processing is to:
• Combine into one file the availabl
on-site and NWS meteorological dal
-------
files created during Stage 1 process-
ing
• Store the data in a more compact
format.
For the specification of the dispersive
state of the atmosphere, it is most
convenient to consider the physics of the
atmosphere on a daily basis, i.e., a 24-
hour period. Estimation of the depth of
convective mixing is in reality the
summation of effects starting with the
heating of the surface shortly after
sunrise. Thus, the merging of the
available data for each 24-hour period
is the next logical step in the processing
before developing the characterization of
the input meteorological data files for the
dispersion models.
The merged data are stored in un-
formatted form because this format
provides a more efficient usage of
storage than the formatted ASCII data file
storage used during the first stage of
processing. The ASCII files are
convenient for text editors but are no
longer needed once the quality
assessment and editing are completed.
Creating a Model Input File
(Stage 3 Processing)
The goal of this third stage of
processing is to:
• Create a meteorological data file for
use with a regulatory dispersion
model chosen by the user.
MPRM can generate any one of
several output formats to meet the input
requirements of the regulatory dispersion
model chosen by the user. The
RAMMET format has been selected as
the default output with default methods
for processing wind, temperature,
stability category and mixing heights.
These methods employ the NWS hourly
surface weather observations and NCDC
twice-daily mixing heights and dupli-
cate the processing performed by the
RAMMET meteorological processor. 4
Since no "modeling* 'has beten^
performed prior to the third stage of*
processing, it is anticipated* that future
changes to the modeling guidelines will
have the most impact on this stage of
meteorological processing. Acceptance
of new algorithms for mixing height
estimation, or methods for characterizing
the variation of wind speed and wind
direction with height, would require that
new computer algorithms be supplied
within MPRM for use at this stage of
processing. Acceptance of a new
dispersion model might require changes
to the output subroutine within MPRM in
order to provide the meteorological data
in the format required by the new ac-
cepted dispersion model. In consid-
eration of these possibilities, MPRM has
a highly modularized design. This allows
upgrading of specific parts of the
computer code without having to rede-
sign the processor.
Relationship of MPRM to EPA
Air Pollution Modeling
Guidance
The data processing methods incor-
porated into MPRM are intended to
implement the recommendations con-
tained in the EPA on-site meteor-
ological program guidance document.
These recommendations include the
determination of Pasquill stability cate-
gories from on-site measurements,
based on the recommendations in the
EPA Guideline on Air Quality Models
(Revised). As data processing recom-
mendations are modified, MPRM will be
upgraded to reflect the latest guidance.
Moreover, any discrepancies that might
exist between MPRM and current regu-
latory guidance should not be construed
as guidance, but as errors within the
MPRM system.
It is not the purpose of this user's
guide to provide a comprehensive sum-
mary of all relevant guidance on disper-
sion modeling for regulatory applications.
Other recommendations from the guid-
ance document for conduct of an on-
site meteorological measurement pro-
gram may be relevant to a particular
application. For example, the on-site
meteorological program guidance docu-
ment contains recommendations on in-
strument siting and quality assurance. An
important recommendation is that a
minimum of 90 percent valid data
•recovery exist for each variable before a
data set can be used for regulatory
modeling. The issue of handling missing
data values for dispersion modeling is
discussed below in more detail. Please
note that data substitution cannot be
.used to reach the 90 percent data
recovery rate required by regulatory
modeling guidance.
Missing Values, the Bane of All
Data Sets
When the meteorological conditions
are insufficiently declared, a dispersion
model will not be able to produce
concentration estimates. In the case of
dispersion models currently accepted for
regulatory applications, the situation is
aggravated by the fact that none of the
hourly dispersion models can continue
processing if the meteorological record is
not continuous. The only way to continue
processing is to present the dispersion
model with an unbroken meteorological
record having no missing values.
Substitutions based on other on-site
data, if available and deemed to be
representative, may provide the best
solution for providing an unbroken
meteorological record. If the situation is
such that only 1 hour's data is missing, it
might be practical to use linear inter-
polation between adjacent hours to
estimate the missing meteorological
conditions. As the time period of missing
values increases, the usefulness as well
as the reasonableness of linear inter-
polation to fill in missing values becomes
increasingly more dubious. Rational-
izations involving use of monthly mean
values from climatological records are
sometimes employed. The fact remains
that, given the right circumstances, any
technique employed for filling in missing
values can prove to be inadequate.
A clear consensus has yet to be
reached on how best to resolve the
dilemma created by a broken meteor-
ological record. A possible solution may
involve making substitutions for missing
meteorological values for isolated 1-
hour periods, and treating longer breaks
in the meteorological record within the
dispersion model. At the very least, the
dispersion models might be modified to
process available valid data and to skip
(or output a missing value indicator for
the concentration estimate) those hours
when processing could not continue due
to missing values in the meteorological
record.
While the guidance document for
conduct of on-site meteorological
measurement programs contains a
recommended hierarchy of data substi-
tution strategies for regulatory ap-
plications, the implementation of these
recommendations requires expert judg-
ment of "representativeness" of the data
substitutions. Yet to be developed is a
system of numerical rules having
sufficient expertise that we can con-
fidently recommend their use in a
universal sense for automatic processing
of missing data values.
Since MPRM is constrained to include
only processing methods that have been
accepted for use in regulatory
applications and since we have yet to
develop a set of numerical techniques for
universal use in handling missing values,
MPRM has no automatic method for
-------
correcting missing values. If and when
techniques are accepted for handling
missing values on an automatic basis,
they will be incorporated into MPRM,
unless of course the resolution is within
the dispersion model algorithm.
The EPA author, John S. Irwin, is with the Atmospheric Research and Exposure
Assessment Laboratory, Research Triangle Park, NC 27711; James 0. Paumier
is with the Computer Sciences Corporation, P.O. Box 12767, Research Triangle
Park, NC 27709; and Roger W. Brode is also an EPA author with the Office of
Air Quality Planning and Standards, Research Triangle Park, NC 27711
D. Bruce Turner is the EPA Project Officer (see below).
The complete report, entitled, "Meteorological Processor for Regulatory Models
(MPRM-1.1) User's Guide," (Order No. PB 89-127 5261 AS; Cost: $28.95,
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
States
Environmental Protection
Agency
Center for Environmental Resear
Information
Cincinnati OH 45268
Boi* Business -
Penalty for Private Use $300
EPA/600/S3-88/043
------- |