LOW LEVEL RADIOSONDE DATA REPRESENTATIVENESS
AS A FUNCTION OF SAMPLING INTERVAL
FOR QUARTERLY PERIODS
22 March 1977
-------
CDM/LIMNETICS ENVIRONMENTAL CONSULTANTS
6132 West Fond du Lac Avenue
Milwaukee, Wl 53218 ¦ 4141461 -9500
9025 East Kenyon Avenue
Denver, CO 80237 ¦ 303/770-8252
3650 Westbank Expressway
Harvey, LA 70058 ¦ 504/340-3501
11485 West 48th Avenue
Wheat Ridge, CO 80033 ¦ 303/423-2766
-------
4
Lt>u3
K
z
[low level radiosonde data representativeness
AS A FUNCTION OF SAMPLING INTERVAL
FOR QUARTERLY PERIODS
Prepared For:
Environmental Protection Agency, Region VIII
1860 Lincoln Street
Denver, Colorado 80203
Prepared By:
CDM Environmental Sciences Division
11455 West 48th Avenue
Wheat Ridge, Colorado 80033
U.S. EPA Region 8 Library
80C-L
999 18th Si., Suite 500
Denver, CO 80202-2466
0^
'dark G. Musg^ove, Ph.D.
Vice President
Atmospheric Scientist
22 March 1977
-------
TABLE OF CONTENTS
Section Title Page
1.0 INTRODUCTION 1
2.0 METEOROLOGICAL CONSIDERATIONS 3
2.1 Meteorological Parameter Selection 3
2.2 Methodology to Calculate the Meteorological
Parameters 5
3.0 STATISTICAL METHODOLOGY 9
3.1 Frequency Distributions 9
3.2 Comparison of the Means (t-Test) 10
3.3 Comparison of the Frequency Distributions (x2-Test) 11
4.0 RESULTS 14
5.0 SUMMARY CONCLUSIONS 24
APPENDIX A A-l
APPENDIX B B-l
i
-------
LIST OF TABLES
Table Title Page
2-1 Meteorological data transcribed from radiosonde
soundings and their symbol definition 6
4-1 Arithmetic mean (upper left corner) and standard
deviation (lower right corner) of each of the
meteorological parameters for each season 15
5-1 Minimum number of days of sampling period required
for 90 percent of the samples to pass the x2-test... 25
5-2 Minimum number of days of sampling period required
for 90 percent of the samples to pass the t-test.... 25
5-3 Subjective weighting factors assigned to each
analyzed parameter with respect to its contribu-
tion in dispersion modeling results 26
ii
-------
LIST OF FIGURES
Figure Title Page
4-1 Results of the statistical tests for the surface
to 800mb lapse rate and the surface-based
inversion height 17
4-2 Results of the statistical tests for the surface
to 750mb lapse rate and the surface to 700mb
lapse rate 19
4-3 Results of the statistical tests for the pressure
thickness from 800mb to 700mb and the precipitable
water content between 800mb and 700mb 21
4-4 Results of the statistical tests for the surface
wind speed and the 700mb level wind speed 22
i i i
-------
ABSTRACT
Explicit criteria by both state and federal agencies addressing sampling
duration or frequency of upper air data in support of baseline monitoring
programs do not exist. Eight upper level meteorological parameters indirectly
related to the results of dispersion calculations were selected from an annual
data base from a National Weather Service station in western Colorado. These
data were analyzed by quarters to determine the minimum sampling duration
required to pass difference in mean tests and difference in frequency distri-
bution tests. Recommendations concerning results and the adequacy of the
statistical tests performed are made.
iv
-------
1.0 INTRODUCTION
This report was prepared for the United States Environmental Protection
Agency, Region VIII in order to provide quantitative justification of the
requirements for upper air sampling for predevelopment baseline monitoring
programs. Although not intended to represent a universal guide to determine
minimum sampling requirements, this study allows several recommendations
concerning the upper air monitoring period length to be made.
Explicit criteria by both state and federal agencies, addressing the sampling
duration of frequency of collection of upper air data in support of baseline
monitoring programs, do not exist. Cost considerations generally do not
allow continuous daily sampling over the entire duration of the baseline
period. Because of this, there is a need to know the minimum sampling period
which adequately represents the period of interest. For purposes of dispersion
modeling, it is appropriate to break down the analyses by seasonal quarters
because the dispersion characteristics follow seasonal patterns. The minimum
sampling period in each quarter is, therefore, of interest. The average
inversion height over this period could be used in numerical computations as
the average inversion height for the seasonal quarter. In addition, the
frequency of occurrence of surface-based inversions during the shorter interval
would probably be used to estimate their frequency during the quarter.
Although not as commonly used, the upper level winds are sometimes used to
estimate those which would be expected at the effective plume heights for the
entire course of the quarter. For short-term, worst-case estimates of pollut-
ant concentrations, the worst-case measurements (most intense inversion,
lowest mixing heights, etc.) taken during the shorter sampling period may be
used to approximate the worst-case measurements which would have been made
during the entire quarter.
These practices typically ignore the question of how accurately the data taken
during the shorter interval describe the conditions to be expected throughout
the seasonal quarter. The purpose of this report is to investigate the
adequacy of shorter intervals of data to describe the longer seasonal distri-
butions. Quantitative recommendations of the minimum sampling periods are
1
-------
presented along with interpretation of the types of parameters which provide
the most meaningful information.
For the purpose of this study, it was decided to limit the analyses to a
single year of data with statistical tests on the representativeness of the
data, independently performed for each seasonal quarter. The meteorological
parameters which were chosen for study are described in Section 2.0, along
with an explanation of their measurement and derivation. A brief justifica-
tion of the selection of each parameter, due to the difficulty in interpre-
tation of its contribution to modeling estimates, is also included.
Primarily due to the proximity of the federal lease oil shale tracts and
northwestern Colorado coal reserves, data from the National Weather Service
station at Grand Junction, Colorado were chosen for this analysis (elevation:
1,472 meters above Mean Sea Level). The 1200 Greenwich Mean Time (0500 local
time) radiosonde soundings over the period from 1 September 1974 through
31 August 1975 were obtained from the National Climatic Center and divided
into the four seasonal quarters. It was determined, on the basis of a brief
investigation, that the afternoon soundings (1700 local time) were quite
consistent from day to day and that extensive sampling during the afternoon
would not be required to adequately define the distribution by quarter. The
major emphasis, therefore, was placed on investigation of the morning
soundings.
Section 3.0 of this report describes the methodology used to perform the
statistical analyses, while Section 4.0 presents the results of the tests,
and Section 5.0 makes specific recommendations concerning minimum sampling
periods by season.
2
-------
2.0 METEOROLOGICAL CONSIDERATIONS
2.1 Meteorological Parameter Selection
Because the meteorological parameters which provide the largest impact on
dispersion modeling predictions are the temperature and wind structures with
height of the atmosphere, primary emphasis was placed on measurements related
to these values. Due to the complexity of profile analysis techniques, it
was determined that mean temperature structures within atmospheric layers
would adequately document the representativeness of shorter sampling intervals
to a seasonal data base.
The Grand Junction National Weather Service observing station is located at a
height of 1,472 meters above Mean Sea Level (MSL). Its pressure altitude
fluctuates about 850mb. The lowest mandatory level (pressure altitude at
which data is always reported) which is consistently given in the radiosonde
soundings is therefore at 800mb. The next two lowest mandatory levels are
at 750mb and 700mb. The reported temperatures at these levels were used to
determine the mean lapse rate from the surface to each respective pressure
altitude.
A common feature of most southwestern areas is the formation of a relatively
shallow but intense surface-based inversion due to radiative cooling each
night. In the area of interest, these low-level inversions form on about 75
percent of the nights in a year. Almost without exception, they are destroyed
by solar heating or good mixing by the noon hours of the day. Occasions upon
which surface inversions do not form are either associated with frontal
system passage and the accompanying cloud cover, or moderate to high winds
tending to mix the atmosphere vertically. Although not explicitly related to
concentration predictions, the height of these morning inversions dictates
whether or not a plume released in the stable layer will penetrate through the
top or remain and disperse in the stable layer. It was felt that analysis of
the representativeness of a shorter sampling period to that of the seasonal
data of surface-based inversion heights was appropriate.
Because the formation of surface-based inversions is substantially inhibited
3
-------
by cloud cover, a means of estimating the cloud cover was also desirable. The
meteorological parameter probably most closely related to cloud formation is
the precipitable water content of the atmosphere. Time constraints limited
the analysis of this meteorological parameter to only certain layers.
Winds also play an important role in the dispersion modeling predictions. For
this reason, wind speeds at the surface and at the 700-mb level were analyzed
in the same manner as the other parameters. The wind direction data were not
included in the analyses for the following reason: the air movements near
the surface at Grand Junction are primarily governed by local conditions and
are often completely different than those expected due to synoptic patterns.
The reporting station lies just to the north of the center of a large, broad
valley oriented in an east-west direction. High terrain surrounds the valley
on all but the west side. During the night, the greater radiative heat loss
from the higher elevations cools the air immediately adjacent to the surface
and this sinks into the valley. Downward movements of cold air, set in motion
this way, form downslope winds (easterly) which predominate into the daylight
hours. Soundings made at 0500 local time generally show this pattern.
Another meteorological parameter which was thought might indicate the passage
of frontal systems or possibly provide information on the persistence of
high pressure cells is the thickness of the layer between the 700-mb and
800-mb levels. With a low-pressure region in the area, the depth of the layer
between these two pressure levels would be shallower than normal.
It should be pointed out that the list of chosen parameters for the representa-
tiveness investigation is by no means inclusive. Other possible parameters
which could be studied include actual cloud cover, net daily solar insolation,
precipitation and visibility; all of which might provide more meaningful
estimates of the consequences of using data from short sampling periods in
place of larger data bases. The goal was to select parameters which would
adequately represent data obtained from soundings. Because of the relative
ease of obtaining the parameters which were selected, and their variety, it
was felt that they would provide an adequate basis for an initial analysis.
A listing, by month, of all of the data analyzed is presented for reference
in Appendix A. Each page lists, by day of the respective month, the eight
4
-------
meteorological parameters described above. Note that, for ease of analysis,
each seasonal quarter was made exactly 90 days long by dropping the appropriate
number of days in the last month of the season. The following section
describes the computational; techniques used to derive the above eight parameters
.from the data itemized in TableJM.
2.2 Methodology to Calculate the Meteorological Parameters
With the meteorological parameters described in Section 2.1 in mind, the data
listed in Table 2-1 was transcribed from the 1200 GMT radiosonde data for each
day during the period 1 September 1974 through 31 August 1975. The symbols
defined in Table 2-1 are used in the equations which follow.
Mean Lapse Rate
The mean lapse rate from the surface to 800mb was calculated according to:
r80o = Hg°o - {472 x ^00 (degrees Celsius/Kilometer) (1)
The mean lapse rate from the surface to 750mb was calculated according to:
r75o = n75Q ~ I^79 x 1000 (degrees Celsius/Kilometer) (2)
rl 7 5 0 ~ I £r/t
The mean lapse rate from the surface to 700mb was calculated according to:
r7oo = jj^°° _ I472 x 1000 (degrees Celsius/Kilometers) (3)
Note that the mean lapse rate from the surface to 700mb tends to disguise the
fine structure of the temperature behavior more than does the mean lapse rate
from the surface to 750mb or 800mb. The typical height of the 800-mb level
was about 2,000 meters above MSL, while those at the 750-mb and 700-mb levels
were typically about 2,600 and 3,100 meters above MSL, respectively. Thus,
r80o gives the mean lapse rate within the first 500 meters of the atmosphere,
and r75o and r7oo give the mean lapse rate within the first 1,100 and 1,600
meters, respectively.
5
-------
Table 2-1. Meteorological data transcribed from radiosonde soundings and
their symbol definition.
1.
H800
- Height above MSL at which the 800mb level occurs
(meters)
2.
H7 50
- Height above MSL at which the 750mb level occurs
(meters)
3.
H700
- Height above MSL at which the 700mb level occurs
(meters)
4.
T
s
- Temperature at the surface (degrees Celsius)
5.
T800
- Temperature at the 800mb level (degrees Celsius)
6.
T750
- Temperature at the 750mb level (degrees Celsius)
7.
T700
- Temperature at the 700mb level (degrees Celsius)
8.
R800
- Relative humidity at the 800mb level (percent)
9.
R7 50
- Relative humidity at the 750mb level (percent)
10.
R700
- Relative humidity at the 700mb level (percent)
11.
Ss
- Wind speed at the surface (meters/second)
12.
S700
- Wind speed at,the 700mb level (meters/second)
13.
L
- Surface-based inversion height
6
-------
Pressure Thickness
The thickness of the layer between the 800-mb and 700-mb levels was calculated
by:
Note that this parameter is an implicit function of temperature. During the
summer, the warm air is less dense and the pressure thickness is relatively
large. During the winter, the cold, dense air makes the typical pressure
thickness relatively small. There is thus an annual cycle associated with
this parameter which tends to make the statistical analyses described in
Section 3.0 less appropriate than for other parameters. The pressure thick-
ness ranged from a low of 990 meters in mid-January to a high of 1,153 meters
at the beginning of July. To enhance the appropriateness of the statistical
analyses, it may be possible to remove this cyclic variation by using the
mean daily temperature. This was not performed for this investigation.
Precipitable Water
Due to the time required to integrate the precipitable water content through
the height of the atmosphere, it was judged adequate to determine the content
between the 800-mb and 700-mb levels. This was done using both the reported
temperature and relative humidity at the 800-mb, 750-mb and 700-mb levels.
The total weight of water in a column of unit cross-section between heights
Zi and z2 can be expressed as:
.. /v
J Zj
where pw is the water vapor density or absolute humidity. Substituting from
the hydrostatic equation, dp = - pgdz, where p is the total air density:
where q is the specific humidity and pi and pz are the pressures at heights
Zi and z2. Approximating the specific humidity, which is the ratio of the
water vapor density to the total air density by the mixing ratio, which is
Pressure Thickness = H700 - Heoo (meters)
(4)
7
-------
the ratio of the water vapor density to the dry air density:
W = "9" J wdp = + ^50jx (800-750) + ^75° + 0)7 0 0 j x (750-700)
where o>eoo> W750. and cj7oo are the mixing ratios at the three levels. Due to
the unit density of water, the equivalent height of condensed water in a
column from the 800-mb to the 700-mb level was taken as:
W = (w8oo + 2u>750 + U700) x 25.51 (centimeters) (5)
The mixing ratio at each level was calculated from the temperature and relativ
humidity at that level using:
100
where R is the relative humidity in percent and ws is the saturation mixing
ratio .which is a function of temperature:
"S - 0.622 -|f
P eS
where p is the pressure at that level (millibars) and eg is the saturation
vapor pressure:
\
es = 6.11 exp 5418 x
273
(millibars)
/
The total precipitable water also has an annual cycle, as does pressure thick-
ness. The water-carrying capacity of the atmosphere depends upon the mean
temperature. During the summer months, the air can hold more water vapor
than during the winter months. Because of this, the statistical analyses
described in Section 3.0 are less appropriate for the precipitable water (as
with pressure thickness) than for other parameters.
Winds and Surface-Based Inversion Heights
The surface winds, 700-mb winds and surface-based inversion heights were
used in their reported form directly from the sounding data. Note that for
cases in which a surface-based inversion did not form the data was treated
as missing in the analysis.
8
-------
3.0 STATISTICAL METHODOLOGY
Because most of the meteorological parameters which lend themselves to
relatively easy analysis are only indirectly related to dispersion modeling
results, general statistical tests were used on each data set as opposed to
evaluating modeling results based on shorter periods. Primary emphasis was
placed on two analysis techniques: 1) testing the difference between the
means of the seasonal quarter data base and the shorter interval (t-test);
and 2) testing the difference between frequency distributions of the seasonal
quarter data base and the shorter interval (x2-test).
For these tests, computer programs were developed to analyze each quarter
separately. To determine the representativeness of a period of length n
days, a "sliding window" concept was employed for each of the above two tests.
Data for the short interval (days 1 through n) were used to calculate the
mean and develop the distribution. These were then compared to the mean and
distribution of the data from the entire quarter using the t-test and x2~test,
respectively. Data for days 2 through n+1, and so on, were analyzed similarly.
The proportion of sliding windows of length n which were not significantly
different from the data set for the entire quarter at a given confidence
level was thus calculated. The next step consisted of varying the sliding
window length from 5 days to 50 days. Thus, each meteorological variable was
analyzed by calculating the percent of time a data set adequately represented
the seasonal quarter as a function of the length of the data set (function of
sampling period).
3.1 Frequency Distributions
In order to clarify the results of the statistical tests, histograms of each
of the meteorological parameters selected for each quarter were developed.
The histograms show the relative number of occurrences of data within given
intervals. Note that once categorization of the data into given intervals
has been accomplished, all information relating to the time sequence of the
data is lost. An example which points out some restrictions on the usefulness
of histograms can be illustrated with data which linearly increases with time
over the entire season (e.g., precipitable water content in spring). If the
9
-------
data were truly increasing with time, then the corresponding histogram would
appear as a uniform distribution, with all values from the seasonal minimum
to the seasonal maximum occurring with equal probability.
Histograms of each of the selected meteorological parameters, by season, are
presented in Appendix B for reference purposes. Note that the mean and
standard deviation of each parameter are given as well.
3.2 Comparison of the Means (t-Test)
Certain meteorological parameters are used in dispersion calculations which
require the mean value of that parameter over a long period. The representa-
tiveness of the mean derived from a shorter time period (fewer samples) to
the actual mean of the entire quarterly period can be determined by using the
estimated standard error of the difference between two means. In effect,
this allows the determination of the probability that sampling error is
responsible for the difference between a sample mean and the true population
parameter.
The question to be answered is: What is the probability that the difference
between the mean value of the population parameter and the sample mean from
the short interval occurred by chance? If the difference between the means
is highly improbable (less than 5%), the hypothesis that sampling error alone
is responsible can be rejected. This permits the acceptance of the alternate
hypothesis that the difference between the means was the result of sampling
from different populations.
To determine if a given window length is representative of the quarter, the
t-statistic which tests the difference between hypothesized and sample
differences is formed according to:
¦j. _ Xi - X?
1 + t . (6)
ni + n2 -2 \ Hi )
where the Xi values come from the given window data and the X2 values make up
the entire quarter. The value of ni is the length of the window and n2 is
always the length of the quarter (90 days). The normal use of this statistic
10
-------
is to compare the difference between means of two samples drawn from a universal
population, although here we are using it to compare the difference between
means of a single sample and the universal population from which it was drawn.
Note that the denominator is the estimate for the standard error of the mean
difference and that the two samples have come from the same data base so
that the estimate can be formed from pooling the sums of squares and degrees
of freedom from each sample. The t-statistic arrived at in this fashion
always has ni + n2 - 2 degrees of freedom.
For the analysis of each of the meteorological parameters, n2 was always taken
as 90, while ni was varied from 5 to 50. The t-statistic was then used in a
two-tailed test at the 95 percent confidence level:
HQ : Mi = M2 for -1. 9 7 5 < t < t.975
The number of sliding windows of each length which were not significantly
different from the population parameter were then tabulated as a function of
the sliding window length. This allowed an interpretation of the probability
that a sample of given length,was not significantly different from the true
situation. It should be pointed out here that the passage of a given sample
at this level is only indicative of its representativeness to the seasonal
quarter data. All that can be rigorously stated is that, of the number of
sample sets of a given length which do not pass, there is a 5 percent or less
chance that sampling errors actually were the cause for the rejections.
The use of the t-statistic presumes that the two samples are drawn from the
same universal population so that both samples have the same frequency distri-
bution. In the manner in which the t-statistic is used for these analyses,
the interpretation of the results of each test must be veiwed in light of the
results of the tests for the difference between distributions. The distribu-
tions of the shorter interval and the quarterly period must be the same in
order to confidently apply the t-test to the difference between means.
3.3 Comparison of the Frequency Distributions (x2-Test)
Because the majority of the meteorological parameters which were analyzed are
11
-------
only indirectly related to the results of dispersion calculations, a x2~
statistic was used to compare the frequency distributions of each set of
sliding windows with that of the parent (seasonal quarter) distribution. The
X2-statistic is based on the discrepancy between frequencies in a sample and
frequencies expected according to some hypothesis. For each meteorological
parameter, the expected frequencies were derived from the quarterly data
base.
For a given sliding window length n, the observed frequencies of occurrence
of values lying in K class invervals were determined: Oi, o2, . . . o^.
According to the assumption that each window approximates the quarterly distri-
bution, the expected frequencies of occurrence are those of the quarterly
distribution: ei, e2, . . . e^. A measure of the discrepancy between observed
and expected frequencies is supplied by the x2-statistic given by:
2 = Y (Oj - ej)2
X 2, e (7)
j=l J
where Eo. = E e. = n and x2 has K-l degrees of freedom.
J J
If X2 = 0, the sliding window.and quarterly frequency distribution agree
exactly; while, if x2 is significantly greater than 0, they do not agree. The
expected frequencies were normalized to the condition that E e- = n. This
J
was done by multiplying the frequencies of the quarterly data set by the
ratio of the sliding window length to the number of observations in a quarter.
For the analysis of each of the meteorological values, the expected frequencies
within a maximum of 25 class intervals were tabulated by categorizing the 90
daily values into the 25 equally-spaced intervals between the minimum and
maximum values. Each of these expected frequencies was then normalized to the
sliding window length in each given analysis. For small sample sizes (e.g.,
sliding window length of 10 days), not all of the 25 classes of expected
frequencies had values associated with them. Indeed, most classes (due to
the normalization) had associated frequencies less than 1.0.
In order to enhance the power of the x2_test, it was necessary to consolidate
some intervals together in order that the expected frequency (normalized to
the sliding window length) of each consolidated class was greater than 1.0.
This was done by grouping the data in an interval with a frequency less than
12
-------
one into the next higher interval. If this resultant frequency was still
less than one, the two were grouped into the next higher class until the
resultant frequency was greater than one. This technique obviously resulted
in unequal interval sizes and fewer intervals than 25. The frequencies of
each of the sliding window data sets within the intervals defined by the
expected frequencies were then tabulated and the x2-statistic of equation (7)
calculated. The number of degrees of freedom used was the number of resultant
intervals minus one. This x2_statistic was then used in a one-tailed test
at the 95 percent confidence level for x2< X2-95> with the appropriate number
of degrees of freedom.
The number of sliding windows of each length which passed at this level of
confidence were then tabulated as a function of the sliding window length.
This allowed an interpretation of the probability of a sample of given length
passing at the 95 percent level for each quarter.
13
-------
4.0 RESULTS
For the interpretation of some of the results which follow, an understanding
of the seasonal variation of the various meteorological parameters is help-
ful. Table 4-1 shows the arithmetic mean along with the standard deviation
of each of the eight meteorological parameters for each season.
Note that the surface to 800-mb lapse rate shows a temperature increase with
height for all seasons with the exception of spring, where the mean tempera-
ture structure in this layer is nearly isothermal. The most intense
positive lapse rates (inversions) occur in winter. The mean lapse rate from
the surface to 750mb and 700mb is always negative for all seasons. Note that
the standard deviation of all three mean lapse rates is significantly greater
for the winter season than for any other quarter, indicating more variation
of the data in this period. In addition, the standard deviation of the mean
lapse rate from the surface to 700mb is less than that for the interval from
surface to 750mb which, in turn, is less than that for the interval from
surface to 800mb.
The pressure thickness data show a definite annual cycle, with a minimum
during winter and maximum during summer. Note that the standard deviations
of the spring and fall seasons are much higher than those for winter and
summer. This is primarily due to the fact that the data changes from one
extreme to another during these seasons, while the data are relatively con-
stant at their extremes during winter and summer. The precipitable water
data show similar behavior, with an annual cycle reaching a maximum in summer
and a minimum in winter. An important difference, however, is apparent in
the seasonal variation of the standard deviations of these parameters. Here
the standard deviation also reaches a minimum in winter and a maximum in
summer. In other words, the variation in the precipitable water follows the
actual precipitable water content as a function of season.
Little information can be determined from the surface and 700-mb wind data,
both from their means and standard deviations as well as from the distribu-
tions in Appendix B. In general, 700-mb winds are slightly higher in the
14
-------
Table 4-1. Arithmetic mean (upper left corner) and standard deviation
(lower right corner) of each of the meteorological parameters
for each season.
FALL
(Sept. , Oct. , Nov.)
WINTER
(Dec., Jan., Feb.)
SPRING
(Mar., Apr., May)
SUMMER
(Jun., Jul., Aug.)
Surface to 800mb Lapse Rate (°C/KM)
3.29/'
/5.18
4.08
-0.35/
./>.32
2.40/"
>/^24
Surface to 750mb Lapse Rate (°C/KM)
-2.80/"
25
-1.74
>/2.30
-3.78/
/""^3.99
-2.96/
/1.00
Surface to 700mb Lapse Rate (°C/KM)
-2.37 /^
/*0- 65
-1.63/
31
-2.6/'
/"^0.57
-2.67/^
./.51
Pressure Thickness (M)
imk.b/
/25.3
L045
/it,. 5
1071.0/
/23.8
L128.8/
/16.1
Precipitable Water (CM)
0.360
127
3.192 .X
./.078
D. 275
109
D. 496
Surface Winds (M/S)
2.67
^xf.83
2.54
80
3.16
yi. 19
3.28
yi.bb
700mb Winds (M/S)
5.lb
/J - 68
8.06/
8.09/"
/k.W
5.44/^
Surface-based Inversion Heights (M)
253
Saw
312
yik 8
171 /^
yvbi
198
y\\b
Key
x /
y/ S
15
-------
winter and spring, which is probably indicative of more frequent frontal
system passage during these seasons. In contrast to this, the surface winds
are higher in the spring and summer. This is probably due to the greater
vertical mixing during these periods, with the surface layers more frequently
coupling with the faster-moving upper level flows.
The surface-based inversion height data show that the mean depth of these
stable layers, when they occur, are greatest during the winter months. Fall
surface-based inversions are slightly less thick, with those in the spring
season tending to be the shallowest. Note also that the standard deviation
of the data, which indicates the degree of variation, tends to follow the
same seasonal trend as the mean. The variability within the data is related
to the mean as a function of season.
With these concepts in mind, we turn to the results of the t-tests and x2-
tests for the difference in means and difference in distributions as a func-
tion of the sampling period. The percent of sampling periods (sliding
windows) passing each of the tests outlined in Sections 3.2 and 3.3 for the
surface to 800-mb lapse rate and the surface-based inversion heights are
shown in Figure 4-1. These two meteorological parameters probably influence
dispersion modeling calculations more strongly than any of the others.
In order to more fully understand these curves, specific examples in the use
of each of the graphs is appropriate. For a quarterly period of 90 days and
a sliding window length of 21 days, there are 90 - 21 + 1 = 70 windows.
Refer to the results of the t-test for the surface to 800-mb lapse rate
parameter in the following discussion. For the summer season, 87 percent of
the 21-day samples passed the t-test. This is equivalent to saying that 65
of 70 windows satisfied the test outlined in Section 3.2, or that 87 percent
of the 21-day windows of surface to 800-mb lapse rate means are representa-
tive of the summer mean value. Note that only 9 days of sampling are
required for the same data set for 90 percent of the samples to pass the
X2-test.
It is interesting to note the behavior of the various curves in Figure 4-1.
Preliminary physical intuition would indicate that the percent of samples
passing would monotonically increase with increasing sample size. The
16
-------
Surface to 800mb Lapse Rate
Surface-Based Inversion Height
l o o
DO
c
H
X 9 0
W
cd
w
)0 50
10 2 0
3 0
hO 5 0
Sample Period (Days)
t-TEST
Sample Period (Days)
COM INC.
Figure 4-1. Results of the statistical tests for the surface to 800mb lapse
rate and the surface-based inversion height.
17
-------
curves, however, show various relative minimums and relative maximums
throughout the sample size range. This can be explained on the basis of
periodicities within the data sequence itself.
Consider a 90-day sinusoidal data set with a period of 30 days. Any sample
of 30-day duration would then produce both a mean value and frequency distri-
bution identical to that of the 90-day quarterly data base, and all sliding
windows of 30-day duration would pass the aforementioned tests. In other
words, 100 percent of the samples would pass for a 30-day sliding window.
For window lengths slightly less than 30 days, some samples would not pass,
depending upon the phase relationship to the periodic quarterly data.
Similarly, for window lengths slightly longer than 30 days, the percent passing
would reach a relative minimum at window lengths of 15 days and 45 days and
would again reach a relative maximum (100% passing) for a window length of
60 days.
This phenomenon is responsible for the relative minimums and relative
maximums in the plots of Figure 4-1. Here, periodicities of several different
lengths are present in the data, resulting in relative minimums and relative
maximums throughout the sample size range. As an example, refer to the x2_
test for the surface-based inversion height for fall. The relative maximum
for a sampling period of 10 days and the relative minimum for a sampling
period of 15 days indicates that there is a cyclical component within the
data with a period of about 10 days.
One further important consideration which should be noted is the correspon-
dence between results of the two different types of statistical tests. In
general, there is relative agreement for each season. For example, the spring
season showed the smallest sample size required for a given degree of
representativeness for the surface to 800-mb mean lapse rate when testing for
both the difference in distributions and the difference in means. Similarly,
the winter season showed the largest sample size required for a given degree
of representativeness.
The results of the statistical tests for the mean lapse rates between the
surface and 750mb and the mean lapse rates between the surface and 700mb are
shown in Figure 4-2. The sample size required for a given degree of
18
-------
Surface to 750mb Lapse Rate
Surface to 700mb Lapse Rate
loo
60
e
iH
W
W
cd
Ph
CO
i-H
a
E
td
CO
M-l
o
a
aj
o
M
a)
PL,
95
90
85
8 0
7 5
\/ //I
~ry i
V
1 0 0
CJD
I-)
tn
CO
n)
PL,
in
-------
representativeness for these two meteorological parameters is smaller than
that required for the lapse rate between surface and 800mb for all seasons,
with the exception of summer. In fact, the sample size actually increases
for this season with an increase in height. In other words, for a given
degree of representativeness for summer, a larger sample size (longer sampling
period) is required to adequately determine the lapse rate between the
surface and 700mb than needed to determine the lapse rate between the surface
and 750mb despite the decrease in the standard deviation from the latter to
the former. A satisfactory explanation for this behavior has not been
derived.
Similar plots of the results of the statistical tests for the pressure thick-
ness between 800mb and 700mb and for the precipi table water content between
the same levels are shown in Figure 4-3. Note that the percent of samples
passing the t-test and x2-test for a given sample size is generally much
lower than for the previously discussed parameters. This is primarily due to
the annual cycle associated with these two parameters. With the comments of
Section 2.2 pertaining to the pressure thickness in mind, note that the fall
and spring seasons (transition seasons) require larger samples for a given
degree of representativeness than do the summer and winter seasons. Although
the precipi table water content follows the same annual cycle, there is no
apparent seasonal pattern like the pressure thickness. Again, it should be
pointed out that, due to the overriding annual variation of these parameters,
the results of the statistical tests are difficult to interpret.
The analyses for the final two parameters are presented in Figure 4-4. For
both the surface winds as well as the winds at the 700-mb level, there is
relatively good representativeness, even for short sampling periods. Winds
at the 700-mb level during the summer tend to require longer sampling
periods for a given degree of representativeness than other seasons, while
there is no apparent similar pattern for.the surface wind speeds. Note the
general agreement between the results of the t-test (difference in sample
means) and the x2-test (difference in sample distribution). Because the
applicability of the t-test depends upon drawing samples with the same
distributions, the interpretation of the results of the t-test must be
viewed in light of the results of the x2-test. For example, the winter
surface wind speed data show that 90 percent of the 11-day samples adequately
20
-------
Pressure Thickness (800-7OOmb)
Precipitable Water (800-7OOmb)
to
c
H
CO
CO
CTJ
Cl,
CO
0)
rH
a,
e
cd
CO
U-l
o
G
t 0
2 0
1 0 2 0 3 0 4 0
Sample Period (Days)
5 o
50
iJESL
o 10 20 30 40 50
Sample Period (Days)
COM INC.
Figure 4-3. Results of the statistical tests for the pressure thickness from
800mb to 700mb and the precipitable water content between 800mb
and 700mb.
21
-------
Surface Wind Speed
700mb Wind Speed
l o o
to
c
H
CO
CO
CO
Ph
CO
Q)
iI
D-
g
CO
C
0)
G
U
d)
P-.
95
9 0
8 5
8 0
7 5
60
d
H
CO
co
co
P*
CO
0)
rH
a
e
CO
CO
U-l
o
C
0)
-------
represent the quarterly distribution and that all samples of any length
(greater than five) adequately represent the mean surface wind speed for the
quarter. This is misleading due to the inadequacy of the t-test if the
distributions are not the same. For this example, all that should be stated
about the results of the t-test is that 11-day samples are required in order
for the t-test to be appropriate.
23
-------
5.0
SUMMARY CONCLUSIONS
The two most important meteorological parameters from a dispersion modeling
viewpoint are surface-based inversion intensities and heights. These two
parameters show a definite seasonal variation for the minimum sample length
required to obtain an adequate representation of the quarter, with the winter
season requiring the longest. Mean lapse rates through layers higher than
900mb tend to require smaller and smaller samples sizes for a given degree
of representativeness. The annual behavior of pressure thickness and precipi-
table water tend to make the analysis techniques used less appropriate than
for other parameters. Surface wind speed samples of even short length provide
representative samples of the quarter as opposed to the upper level winds
(700mb) which are more variable.
The summary Table 5-1 shows the minimum number of days in a sample period
which are required in order to have 90 percent of the samples pass the x2-test
for differences in the distributions for each of the eight selected meteorolo-
gical parameters and four seasons. A similar table for a 90 percent pass of
the t-test for differences in the means is given in Table 5-2. Note the
general agreement between results of each of the tests.
Although it is not possible to quantitatively judge the relative roles each
of the parameters studied plays in dispersion modeling results, an effort
was made to subjectively weight the minimum sampling periods presented in
Tables 5-1 and 5-2. This extremely subjective judgement then allowed a
specific comparison of the minimum sampling period by quarter to acquire data
which would adequately produce modeling results had all data been used in the
evaluation. Table 5-3 presents these subjective weighting factors which
were chosen. The results of applying these weighting factors to the data in
Tables 5-1 and 5-2 are:
Season
Fal1 Winter Spring Summer
Weighted minimum sampling
duration (days)
22 21 7 15
24
-------
Table 5-1. Minimum number of days of sampling period required for 90 percent
of the samples to pass the x2~test.
FALL
WINTER
SPRING
SUMMER
Surface to 800-mb mean lapse rate
18
27
5
9
Surface-based inversion heights
<5
21
5
6
Surface to 750-mb mean lapse rate
9
14
<5
16
Surface to 700-mb mean lapse rate
18
10
<5
34
Pressure thickness (700mb - 800mb)
>50
16
23
29
Precipitable water (700mb - 800mb)
24
<5
27
29
Surface wind speed
28
11
<5
11
700-mb level wind speed
9
9
6
20
Table 5-2. Minimum number of days of sampling period required for 90 percent
of the samples to pass the t-test.
FALL
WINTER
SPRING
SUMMER
Surface to 800-mb mean lapse rate
38
43
5*
.30
Surface-based inversion heights
12
25
5*
6*
Surface to 750-mb mean lapse rate
9*
39
15
20
Surface to 700-mb mean lapse rate
18*
20
<5
35
Pressure thickness (700mb - 800mb)
>50
43
>50
>50
Precipitable water (700mb - 800mb)
40
6
>50
>50
Surface wind speed
28*
11*
<5
11*
700-mb level wind speed
9*
9*
6
>50
* Results of t-test showed sampling periods less than those required by x2~test.
Values shown are those based on x2-test.
25
-------
Table 5-3. Subjective weighting factors assigned to each analyzed
parameter with respect to its contribution in dispersion
modeling results.
Parameter
X2-test
Weight
t-test
Weight
Total
Weight
Surface to 800mb Mean Lapse Rate
0.2
0.1
0.3
Surface-based Inversion Heights
0. 1
0.1
0.2
Surface to 750mb Mean Lapse Rate
0.05
0.02
0.07
Surfact to 700mb Mean Lapse Rate
0.02
0.01
0.03
Pressure Thickness
0.04
0.01
0.05
Precipitable Water
0.015
0.01
0.025
Surface Wind Speed
0.2
0.1
0.3
700mb Wind Speed
0.015
0.01
0.025
Total Weight
0.64
0.36
1.0
26
-------
It should be pointed out that the results of these analyses are limited by
the fact that they were performed for a single monitoring station for a
single year. Future work should concentrate on the spatial variability of
these results, as well as how they vary from one year to the next. Further
consideration should also be given to the interpretation of "representative-
ness". Simple dispersion calculations using data from varying sample size
could also be performed to investigate the results of using shorter sampling
intervals.
27
-------
APPENDIX A
MONTHLY LISTINGS OF RADIOSONDE DATA TAKEN OVER
GRAND JUNCTION, COLORADO
September, 1974 through August, 1975
Notes:
1. In order to make each seasonal quarter exactly 90 days in
length, the months of November, May and August were each
shortened 1 day, 2 days and 2 days, respectively.
.2. Missing data is indicated by asterisks in the respective
column.
3. Asterisks in the Inversion Height column indicate that no
surface based inversion existed on that day.
4. Meteorological parameters are defined in Section 2.2
A-l
-------
pRtii~ 7arspe hrth rv.'EP i-pprqi uiHf"nnM rnLOpRpn
'EPTFNFP 1974
RT IPOn GHT
MERH LRPIE PhTE
HEbPEE.: C¦ V- M
iin MF
I 0
II
4
- M
- 1
1
1 1
4
»
I 0
+ +
- Il
-1
-4
-1
-1
-4
-4
7 n ni'ii-'
+ +
ppe: :upe
thu: p rj v:
netep-t
+h
t
1 1.4
1 irf.
1 |-'l7
14ll
IP"
1 ?7
14 n
14 7
141
14?
0 75
1 Hi
1 OP
I" I 9 I"!
U99
1 07
1 09
1PP
09 P
1 19
IIP
1 IP
1PP
1PP
l 3 0
1 18
07?
1 i"iP
1 I'M
PPECIP
MPTEP
C M
j. ?P1
-4
? 4 r:
=' 7 0
?: 1'?
? -t -
=, ? i
4 4r-"i
? P
P 1 4
K P
P IV'
P l.l'-'
44?
5 P J,
4i-.'4
4 fi 7
54P
5PP
457
4 ¦=- ,P
4 -:9
c' 111 I
¦' i''tL' :pefo:
m :
;iippr root'if:
-t- H -I h
,? 11
4
1 0
] P
)
Q
I
1 I
iriVEP.I 1DH
HEIGHT
METER"
+ +
P44
~ ~
P 0 fi
3 P 1
"5? 4
?17
CT
.1 O
1 31
54?
C C. C'
1 ?5
+ + *¦*
4 09
151
P 4 7
1 5P
7* *£
1 3 P
1 P?
1 17
P 4 1
1 P'3
P45
pns
A-2
-------
-'HT'jnir.Hl'b. I'HTh THt- FN D"FP FFpNp J'.INi'TIDH CHL DPPI'D
~rTOPER r?74
RT 1£On ihiIT
DflV
MEbl
LRP: E
P H T K
Pc-FTT.llpE
pppr
] F
MINI'
" F EED:
~ F
HE
PPFE" «
.."V M
th3cirrr
i.iptff
11
""
MONTH
9 i riMf
7"?''" IIP
7 0 Mi'i F-l
'1FTEP-:
r |.i
ZUPFf
7 0 OMF:
+ - +
+ ¦+
+ +
+ +
+ +
A
- +
+ - ₯
f- +-
1
1 1 .
P
-£. 1
1 " a 1
1117
n.
£ £ 1
"l
C"
t
Q
~
- ij. T'
-£ .
1 1 £4
i'i ,
41-,
a
C.
-.
1 0.
1
- £, 4
- ,£. 7
1 1 £"5
0.
i:":'1:
P
:£
4
C ¦
Q
- 4 _ ^
-£. :£
1 Li? 7
n.
t ¦£ 1
1
1 1
c
. J
!
-.8
- . P
lllli
|j.
4 04
14
I-;,
-1.
iT.
-4. £
- 0 . £
1 i'i ~J £
ij,
44 r
j
r;> B
¦I;
'" I' ¦ fl
-1 . 9
-,£.4
£ . b
1 1 f £
1 1 I u
1 ! .
M ,
£tf "f.
1
c
y
.1.
8
-£. 1
inn
1 1
4 £-
"
'U
1 0
CT
4
Z ¦ i
-£. 9
1 1 1 £
\ 1 .
4
T
1 ]
4.
»7
1
- £, 1
1 ] 1 "11;
l"i _
4 ^
,£
1 £
c. ¦
1
-£. n
- M. 5
] iT^li
1 1
¦^tp
1
1
4.
fl
tr
U » 1
1 U9?
II ,
c;4£
1
L.
1 4
~4
"!¦
- :i.4
C ¦ f
1 i"im
'j .
£P 1
1
1 £
1 s
rr
_i ,
Ij
-1 . 1
-£.4
1 i j 4
Ij .
.£ 11 £
(T
CT
16
11 .
8
-1 . £
~c ¦
1 1 m£
I'l .
.£45
i~
ir
17
:£.
C
w 1
-1 . £
-1 6
111 s
0.
fJ 0
f_
IS
1 ¦
-j . '=¦
-1.4
1119
II.
Mb
z>
1?
.
-1 . £
-£. £
1 1 £ 1
I'l,
~r C." ~i
ci
£ 0
"j"1
-1.9
C ¦
1114
Ij.
£ £ £
¦:
4
21
1 .
1
iz. ¦ r
*; it
1117
I'l.
I ?
1
* *
c c
- ?,.
4
-4. 0
B ,=i
1 n??
Ij.
4 p
c
_~
-1
c
h*
-4. 1
__ I« c-
1 0?5
Ij.
5£b
14
£4
£
-£. £
1 09S
fl.
I
4
-| c
C.
1.
8
-£. t.
-1.9
1 Li 99
Ij.
ih,4
1
¦r' 8
fl -
'¦U
****+¦
1 i" ,£
+ ~
7«
c r
M .
C_
~i c:
" n .1
~ C ¦ i~
1 |j :£
II
IT "rC.'
_'|T.
i~
1
a o
-7 .
S
- 9. 9
-£. £
1 L'££
ft ,
4 £
1
4
p-^
-1 .
~i
1 Li P 9
4
1 4
£ Li
0 .
-4r £
-£. 1
1 i j "=.
Ij.
"-¦4 ='
1
9
£ 1
t .
-4 . P
-£. 0
1
fl.
ii
£
c-
HE TbHT
metfp:
+ +
4 fi 4
d54
499
31 £
~i * CT
C, J
84
4I j
1 £4
f
K X
-i c
C L. J
08
477;
11?
24t-
-:48
~ * + ~
8 8
+ + ~ +
11,5
!"' 'f" V
158
+ + ~ +
1 0 m
1 8 k
A-3
-------
PHlf T ~: DHTIF PPTR Th> EM IT"'EP RF'HRF JUH':T]nH f DL DP H [«~
m'-'EMT-iFP 1974
Hi 1c!Oil
DRV
~ F
MOUTH
+ - +
1
MURfl I.PPTE PRTF
LFFPFF" f ¦ (¦ M
PFF"rI.IPE pF F'L I F'
THto-tr: I.IRTFF
Q
I o
II
i a
i :¦
i j
15
if
1 7
i r
i r
R ij
£1
R4
R n ril'iF
+ +
~ r.'. 4
4
1R
i'i
1 M
-4
- II
++
-4
4
-1
-1
- 1
-1
-4
-t
- fi
-1
-4
-]
-4
- i'i
r fifiHF
+ 1-
-1
-l
-i
-t
-l
I'lFTFP ¦
+ -- +
1 I" ? M
Ob 9
on
ij =. ^
057
0 5 Q
iV.4
i"i R
mF
Ob"1
05 R
07 Ij
M:-: f:
n 7 if.
URI
07 4
07 R
Li Re'
i>:-R
Ut 4
i.i 7F,
OR.a
l.lif, l"l
O'r'iR
ij7F
n57
Hi-". R
05:f
ri7
l r-1
+ +
4 4R
41 1
4 '5R
-:4 F
RRl
R 4R
R .n4
4 5R
J." s
45
IbR
-"-i
117
1 J?
c'R 0
c. c
1
1 fJR
1 8°
1 ? n
o j
1b
i.i i r-rj tpefd:
i'i :
"J.IPFL 7 01JI '1 p
+-+ +-(-
*¦ *-
l
i
IR
to
i'i
1
1 1
1 ' '
cr
i
1 ij
4
4
invep:ion
HFIRHT
hftfp:
b R
*¦
+ ***
1 05
5 0
1 3t"
RR4
~ ~ * ~
*¦ ** +
R ij 1
1 7:7
4 R 7
¦f' 4 b
5 0 1
Rl
1 1 1
R4R
5RR
* *¦ *¦ +
R'"1 "!
5 s
b 0
R 0 R
51
A-4
-------
F'hD IP". DUPE HflTfl THEM n'-'EF bPP^P .'HHn'TDM L GL OPhPD
PFfFMLEP 197 4
Hi lE'On hMT
Hv
'¦iepm
I HF
:e
C'H IE
F'PE " 7 l.'F'F
FPFC1P
i-l I HP '7
PEED 7
]hvep:I
~ F
DEb
PEE
\"
¦ KM
T H 117 h N7 7
i'l M T iff F'
M
"
HEIGHT
HTH
3 fi nl-i
P 7
sfi|"i
?
7 ft Ml'1
P
METEP7
r |'i
7 IJPFC
7 t"1 i"'M B
METEP7
- +
+
+ +
*r
4-
V
f +
+ V
H 1-
+ -+-
V +.
1
.-i
. 4
- 'J
. 4
0
M
1 nc'?
. 1 fij
3 7'4
lZ.
j
1
. o
- l"l
. 3
1 LI if, ]
0. 1 ? 4
cr
C. C' ~
"i
1 1
. 1
i
. 0
-1
¦ iZ
1 I'ltr.-f.
M.ISU
,?
h,
!P t'
4
is
t i~
ij
. 1
I.".
. 1
1 I'PO
n. 1 9f-
1
Ii
7 S3
cr
C.
Q
-1
<~l
-3
. :f:
1 117 3
m .47":
I
1S7
- I'l
7-
-j;
. -1
1 MS 3
m. f: .7 9
c
137
1
-1
. l
- f:
rr
_ i
, K
1 M S i'l
m. 1-: 'J 7
7.
r;
1 :-:5
O
-.f:
'U
4
~i
I'l
1 ii."-::-
ij . ! f' 7'
1
L
1 n
1 0 0
Q
1
9 Q
1
- n
M
1 fir. 1
0. 1 '"14
A
7
."*¦ 1
1 n
c>
. 4
1
l"l
-1
. 1
1 Ui-0
0 . 0 '"-1 '7
-j
? 3 3
1 3
O
L.
. 1
- n
B t-j,
1 Of. ?
i"i. 1 11
a
cr
_!
L" L~
13
O
. 1
!=¦
t'
-1
'-4
1 114 'f:
>"'. ,f:,:"7.
1
cr
3 6 Li
1 3
. t-
-4
. ?:
""" d.
LT
1 HF4
I'l. 1 f - 'j
Q
1 0
f:44
14
o
C.
-4
i>' j
il
¦ 1
1 03*
M . 1 = f:
L
11
1 3 7.
1 -
-1
. n
-4
~<
r 1
n. 1 'f. 7.
4
13
1 if
-1J
¦ l"
i~i
-3
1
1 114"'
0 . .f'.f: i1
1
1 T
f'43
17
'^1
~l
-1
, i")
~ C
. n
1 i'l-", 0
n. f m °
4
14 n
13
V,
f.
- f:
.
-,z
, y
1 n
M. 1 -.7
£
c 1 if:
1 9
J
-J?
in
c.
- il
t i"i4t.
l"l . 1 ° 1
1
1 ~
4 3
3 0
. 1
- f:
¦
- 3
c
. j
1 '141
11. 1 "¦ 7:
4
'7:
3 3 0
,f:l
-1
. 3
-1
cr
]
~
C. i
~"i
-4
. 8
-l
cr
1 044
m .
1 7:
»-*¦ + *
C
_ -
a r
-4
- 3
1 H1 7
0. m 9
1
1-,
£4
-:f:
C"
- 3
"i
-1
¦ C
¦j
m . '"i 9 T
P
~¦ + + *
-I C"
r. j
i'l
l_-
-1
J
IT
J
1 Ml 1
ij. ri:?7
a
1
1 17
d b
1
¦ c
j
.
-1
¦ 0
i 04 7
n. i-i'r-'7
\
3
C. i
¦f:
- 0
t t-
n
. *7.<
1 M" ,f
n. 137
1
r~
1S9
c. z>
o
,
1
p
-
. 4
1 H44
IJ. J if. -1
I
M
"1 -1 cr
C C. _
1' M
f:
.
-1
B il.
¦ C
1 O'-'O
ri. -: 1 :f.
i
i
T"
rJU*
? 0
i"~
. if:
i~
. 4
-1
Q
1 i'l "'.f
'"1. ,f: 1
4
31
4
. 4
u.
tr
-1
cr
1
Ij. ] :f-i
1
1
A-5
-------
pHniD~r]Hi'F iihth Tnf ph gvrp r-phi'D ji.'ij|"tititJ ¦: rjLnpHPn
JHI MJPPY 1 '-T j
HT IP I'lfi 8KT
IlH'l'
HFPH
1 .HP "
E PPTE
feF.171 IPP
PFEC TP
t.) r m
"ppf d."
imvep:1
nf7
DE
?pfe:
r i- m
t H 11; \ !J" :
MPTFP
r1
¦
HP T 8HT
ppmth
n m M p
75 0MF
r iiriMB
mftfp:
PP
: ijpFr
" 0 01'13
f'lFTEP "
f-+
+ j.
-i +.
V +
-1 1-
+ +
A (.
+ - +
+ +
]
-t
- ri.
p o. m
1 op 7
M. 1 3'-'
4
4
18 0
-8.
1
-i.
P ~ 1 . '3
1 085
0. 8 11'^
1
+ ~ + +
->
8.
Z<
-8. n
i n3
'"i. 8 "i*
4
1 0 3
4
1 a
4
8.
8 - 0 . 8
1 0 38
0. 1 MS
c
Q
8 q 4
5
c'
-0.
'ri - l:' . 8
l M 8 3
0 . 1 55
,r.
847
K
-1 .
4
-0.
8 - 0 . 5
1 ri48
0 . C tT'l*
1.
i
+ ¦*¦ + <~
-P .
~
- o.
7 - L . 7
1 04 3
0 . 3 3 1
1
8
*¦** »
8
4.
1
-i .
! -i. r
1 ir='5
0. 887
''-i
1 1
84 0
Q
cr
8
- 3 .
_ I1 C'
1 1141"'
0. 8 r -
8
» + * *
1 II
(-¦ .
fi
_ cr
1 n m 7
0. 1 ,38
3,
ir
1 1
1 a
Q
_ 1
'-A p m
liH5
0. 1 ?.'?
1
1 1
5Q8
18
8.
c.
- 3.
4 - 1 . =¦
8'^ 0
0, 0 3-.
c
.»
1 3
1 37
1 3
11.
3.
1 1 . ii
10
11. ] 08
1-
O
1 84
14
18.
r
4.
5 - n. 7
j o^
0 . 1 8 7
1
L
18
14 04
1 <=;
17.
¦j
o
1 0. 1
1 088
0. 14P
1
1 3
1 4 3 P
16
81 .
i
3.
3 -1.4
1 0^1
0 , 1 3 3
I
Q
q58
i
X t
15.
i
11.
ir -> cr
' L-
1 058
0. 1 °8
8
7 43
18
8 i"1.
8
1 088
+ * *- + *>
iZ_
8
16
18
8 o.
i
- 3.
5 -8. i'
1 051
1 j . 8 t; ¦ I'l
I
0
'< \ i.~
8 0
1 5.
f
-1 .
'3 '"i. P
1 0 f :l
0 . 1 '8:
1
Q
-" -1 cr
1 t
81
fr
1 .
P - 8 .
1 osr
'"1 . 8 UP
1
*¦ *¦
1 j «I_j
c, c
.
t,
-1 .
3 - 0 . 8
1 0 38
ii. 143
J
-r
1 38
C Z'
4
1 .
M 1 . 5
1 'V 1
0.1 3
v
-1
5 0 7
4
O
c
,
* + * +
* * * *
-I
* +
8 7P
-i cr
C J
1 4.
II
1 .
p -8. n
1 0 78
n . 3511
1
8 'j
84 ^
£v,
IT.
-1 .
8 -1.8
1 08 1
0. 1 37
4
It?
c-
_r |
I« j1
1 1 .
6
-1 .
t 3 . ' ~l
1 0 7 4
0 . P 1 1
485
«r_ '~r
8
-a ,
' _ cr
i _ 1 0 . 1
11175
0. 3?-'
,3 L
1 08
88
n.
t.
"""" ll ¦
8 -1.4
1 0 35
0.11 P
~i
:3
¦1 j
3 0
i_,
-8.
8 -1.7
1 050
it.: p 1
1 '1
*¦
31
! I .
c
-4.
3 -8.8
1 04 5
0 . .3 i p
1
41
A-6
-------
pHTiin-[ltU!E PhTh T fti- F t-< D''FP GPhUl1 j I j >" T T ~ t J i:nLDFPnn
FEPF' UPP'i 11-1 ¦
HT 1 ft I
hMT
I1 H' i'
~ F
MDHTH
+ - +
1
1 ij
1 1
Id
1
1 4
15
1ft
19
ft lj
ftl
ft n i'1'lp
4 . 1
a,
4
i
1
4
C"
_ !
4
G
4
_ i
4
-1
- 4
4.4
7'
LHP
:e
P H 1" E
PPE7 II 'PE
P PE
IIP
I-' T l-Jp 7
FEEP"
PFF
¦ i"
^ M
T H J C I- r i 7 7
14H TFF'
I'i
~
SfiM P'
7" I'M 1 h I
f'lETFF' 7
i
1
" Upftf
T" 0 ijl'T
¦H
4- +
+ +¦
A-
1-
+ -1-
+ +-
-,z
cr
m i
- ~ .
7:
1 1 j 4 ?
IJ
. 7; 5 4
1 1
-1
L~ ¦
4
1 055
ft
. 1=7
c
1
^1
""" l~
- 7:.
ft
1 IjlT' i"1
Ij
, ! 4ft
4
- 7:
- 7'.
0
i 054
IJ
. 1 7:7:
cz
tr
-
. 1
-ft,
cr
_)
1 045
Ij
. 7 i.11"
4
c
-3
. n
-,7 .
i
1 ij ,7.' 7:
1 1
. 1 ft 4
?.
4
-1
. c'
- n.
ft
1 f 4 7'
I'I
->
-i
1 0
-1
"i
- 0.
ft
1 05'-'
II
. ft d 9
1
1 1
*¦ * *
j in, r
¦*
i !
ft
-4
,
4
1 i'i ft f-.
I'I
P
1 4
- 7:
. 1
- 7:.
3
1 i j 4 5
I'I
. l?'^
1
-ft
. 6
-1 .
] i'i 5 ft
Ij
. 171
'7:
-t
Q
-1 .
ft
1 h74
I'I
. 14'?
1
i
*¦ + +
₯ *¦ <>+ *¦
1 nc7
*
* +
nr
-4
. M
ij
1 i'i-";?
lj
7,4 4
1
4
-4
fl
-1 .
¦j"1
1 114 ft
1 1
. ,7' 7''
i
4
- 3
.
- ,7'..
1 04 5
0
. 3 »¦(¦=.
1
*¦*
- 7'
i,
- ?.,
1 i.1]
Ij
. 1 T 4
1
4
- 7:
C
5 03 i
1!
. i r?
4
"
- 7'
'I1
- -'.
lj
] r'5n
f:
. ft 4 4
1 1
-4
- 7:.
,7:
1 nft ft
ri
. I'I ? 1
4
¦4
1
Q
-ft .
4
1 0 1 ft
ij
. Ij'^l
ft
- 7:
0 ,
4
1 II,-: 7
ij
. 1 3ft
1
1,7.
M
ir-
lj ,
j
1 1
ij
. ft Oft
L.
Q
-ft
, II
-1 .
7:
1 Oft 4
ri
. 1 ft !
4
"" i"l
. 4
- 7-'.
ft
1 05?
ij
. 1 5 c-
¦4
1
- ?>
, 4
*¦ + +¦¦*¦
*¦ + **
+ + *- + ¦*¦
t,
* f
-ft.
ft
**¦+ +
4
tmvff-: i arn
HEIGHT
|'1etep:
+ -i-
J* c. o
19ft
71?
h. !-
JL£*
* + ₯ +
?1 ?
i c!
cr
C i J
cr cr
_t i j
441.
~ + + *
~ ~~~
4 7;
~
ft 011
17?
7: ft i'i
A-7
-------
pHi'inrnnpE phth tfn
MP^'CH 1P7j
HT
1 d 0 Li
FI-1T
H'i"
NFHlJ
LhF
"E
F'HTF
ppf :
" 1 iPF
PP'Ff IP
I-! I I-4H
i^EEn:
] riVEF" 1
1F
DE
rPFE
7 f
¦ M
THTC
m::
I'.'H T FP
I'I
'HFIFFIT
JTH
'? IJ UM
P
r jiim
T.,
r M r: I F
(¦IF T
fp:
M
" ijp^r
7fl ij |'1 p
MPTEF 7
- +
+
+
4.
+
i +
+ -
-+
+
I
I
I
4
-1 U
+ h
+¦ 4-
1
11
. 4
~ *
~ +
ij r i
~ *~
z>
1 P4
C.
Q
-
. 4
- P . >1'
i"i p
0. jl.M
~rf
I 1
414
o
. 0
-4
A
c
'j7
f. PP-J
-i
t I
ISP
4
4
. ft
- 7-:
c.
1 1
OF. 7
0 . c' Ijp'
C"
F
T Of'
C
I
i
C."
m _j
- >:
T. ¦
OP I1
U . ]
ij
4t.b
f£.
. r~
. t£-
cr
. Ij
-1.9
0 ~"P
ii.
4
17
1
. 8
- 1:
- 8
- ? . F
ht.P
0. Pi- ?
?
1 I
£
¦j-1
-4
lL
r. . F,
i >hS
n. V=lli
c.
o
'*-i
-4
CT
-c'
- t . 7
0
n. 4P4
l",
Ij
71
. C
-1:
LT
-P. 4
n TF
'J . ? 7 P
,7:
~
i
U
. Ij
-4
. i
-p. P
IJ-.4
ii. JJP
-<
1 FT
ii
.
'"J
-4
. !"i
- i . F
OT 1
0. Pc,cr.
i
4
** + *
-«
-4
¦ i~
- 7-:
r' _ '4
ns n
I ' ¦ C i l".
i
" 1
4
, c
-4
1
. '-t
"j cr
i1 ,:r; >¦}
0. P PP:
1
cr
C7
J
-]
cr
~ ¦ ¦
Ot-r'
'"i . P44
Ii
+ + +¦+¦
h-
1
>
- P
* * + + *
<>» +
I",
9
1
. 1
- P
s
~ C '¦ O
047
0 . j: 7
4
+ * + +
r£
. 4
- V
. Km>
~rl. >
OT' ?
n. j s ij
c
-1
54
cr
Q
-
. ij
-c'. b
0 7 p
11
4
1 C1
u
i~
_ i
c
-4
cr
'. 1
OP i1
Ij.
14
I ,_14
.'1
.Tj
. i~
-4
~i
cr
0 7 p.
ii. :-;ij4
* +-
1 :?
c:
. 4
cr
. 4
I' * C.
OP 1
0. P ijP'
1 3
-*~ + -*-
: ?
ij
cr
r _!
-4
cr
- ?. 4
fi
ij . ] P t
i
Ii
! C
;4
-1
. 0
- J
. a
Ij Cj
I"!4 11
u. i P7
i
r
17-,
C
C
.
-
c.-
-P. 4
0 7 E'
0. P i i ,i.'
'?
*¦
V.
0
. M
- 4
. l
-P. 1
ij F '-j
¦j. ? ? i
-
X h,
i ~T
-7
. 1
'"~l
cr
TL I
II 1 Q
11. i
'~i
ir
¦ !
-4
, ;!
-P. 4
i j i j P'
0. lit-., 1
4
r.<
-
-4
U
-P. 1
fi i ¦ ¦ j
Ii. 0F7
i
\
Ad
! 0
-1
Q
-P
C ¦
i'i
'j. 1 P7
4
* * +
:l
I
cr
- 7;
j
- d. 8
1 j b 5
0 . P 1 P
1 1
A-8
-------
PHTiininmiF JihTh Tfll^ Fr-! DVEP L-FHii f
HPPIL. 1975
!mr t I nn r_~ l CF p DD
RT l£fin !?M"
I'ltHH LPF'E F-RTE
hegpee." c t i'i
iOmMB 7SriMp 7 0 n!-tT,"
4- 1- H 4-
₯
FT"
J
1
- fi
t 9
-1
¦4
-4
f-.
-4
-1
_ 1
M
1
-4
- I"!
-4
^
crj
-4
-4
-4
-4
-4
-4
-4
-J
PPE-: "I IF'E"
THU'Ml" :
I-IFTEP-
+
1 OS-
i l',:4
1 04 9
1 OyF
J 07 y
i
1 fi5o
J 07
1 04- 3
1 fSf
1 0>£ 1
I 07'?
1 0^ 4
1 07E'
1 09 "
0*1
07 I:
OS i'i
MS'",
OA ,=i
0:1:1
fl'llO
rr'jS
09 4
1 i'i 4
L' 7 ?
077
1 u=.'?
1 0 4 ij
1 I'I Si-",
PPECIP
UPTEP
i". n
?
0':J7
J u' IJ
1
159
1 -.4
"¦ i-'
1 F. i"1
.1 "
1 ?
lV 17
19F.
-:Q 1
9 4 1
z ' 1 '
I <
i X
1 T7
1 0
i7 O
r 1
1 S]
iiiIr-T; :r-'EEP'"-
i-i :
: i ir-'pT 7 o oi'1?
+-+¦ +-+
,1' 1 7:
4
I
4
4
r-.
1
1 9
16
I 1
"I
4
It.
1 4
1 4
19
r?
4
1 lj
] MVEP-: TDM
HE I9HT
metep:
H I-
+ ~ + ~
It.'?
17 0
47?
9 ij
+ + + +
+ +¦+¦ +
*--*»*
9 1 5
bj
4 ? i j
i
1 07
1 1 9
] ?S
95
?l
* ** +
*¦ f + *
94
A-9
-------
phpfrrnrim-: pptr t^lf/h n'/rp hpmuj 'iiiittidm rglijpRi.>n
PRY 1. ?7 j
HT i=nn 5MT
dp't
DF
M ~ r -t T H Si'MMt
4- - + + 1-
1
4
1 ii
] 1
1,3
I
14
15
i i-;,
£1
i_ J
t-.
HERN I. HP7E PPTF
HEGPEL-' r M'i
?: n r*i t ¦
i
-4
-1
-4. £
- 4. c'
-3.4
-4 . 4
-4 . P:
-4. 1
-1 . I
-,3. 4
-,-J. 1
-4 . 4
-4 , 3
-4 ]
-=¦1 . 4
-3. 1
- 4 . 0
-4. ;¦
-4. 1
-4 . 5
-4 . P
7 i"111|'1 p
+ +
-7.1
-1
4
*¦*-* P- *¦
-,?,4
-1
-]
;'Pf::upe pffcip
-unrj.r^: mptep
'¦1F TEF'I CM
J. + f 4-
0.171
07,3 n.,r"?7
1 y?
p 1 a
4 1 P.
Pi i.i
Ml HP ".PEEP'
!'1 /
1 04
Lit c'
07 1
0P1
1 1 0
I 0 0
OP 7
1 i ,E'
1P-1
1 o 7
J 0 3
1 o,?.
1 1 o
i it. 5
071
07 1
1 1 7:
09 1
1 f
os 3
1 0 7 P
p o .1
4 >
4 4
~ -*¦ 4
" 14 7
47 1
h , P
4 ,37
: >¦" T.
] 71
7 4
C T 1
+ - f
1
4
1 !
11 ip P
4
1
IE
14
Zl
] 1
1 1
4
c
*¦ *
1
-f
* *
1 ::
14
1 1
& 6
4:-
irv./EF": inn
HEIGHT
f-'ETEPl
+ +
¦51
P. j
9 0
lie'
*¦* + ¦*
1 41=4
1 07
7
'3 47
>_T
1 07
7 P
1 1 P
PE5
3 9 4
147
PI
4,3.3
*- + + +
A-10
-------
H
~
n
I
1
1
1
]
1
1
1
I
l
iT.
il
iT
i."
C.
PR['io.:nniT I'htr tpi em h'-'fp i~ppia1
JUMP 3 97^
JUMC T 1 DM CDLOPfHliD
pr i c
GMT
MEPM LPP"E PhTE
iifgfee: c ri-i
¦| MMP
+
; riMr
--1
-1
-4
1
Q
1
-4 . 0
- ^
1
- II
i'Ot'h F;
h h
ppfT " - | J p P"
TPTi> M :
mftep:
J- H
1 1 04
3 L?4
+ * + ¦*
in?
1 IPS
] 1 ?.T.
l n??
i
] nr,~j
1 f'PP
mi
*¦~*¦+
1 1 ?4
1 1 P4
1 1 3P
j 1P 7
1 IP 4
1 OPP
iii?
1 i"i ° "ri
1 O'r'4
1 1 1 7
1 1 37
1 IP*
¦I I | ii
i llr.r
1 1 n
1 1 -II
> 1 -I
i 141
pppc]p
MPTEP
C M
+ +
fl . S f! V
II. "If;. I j
* +
* ₯ 4> * *¦
* *~+ fr
r
PP4
4 4S
4 114
];4 "¦
i-iimd "peep: imvepiidm
3 4 p
3P4
j 1 4
4 |"M I
31
P p 3
:: '-i
¦;4
'I.IPPf
' 0 '";MJ;
-ii-
ir
4
+ ~
*¦
I
i
l n
3 1
11
HEIGHT
METER::
+ +
IP 4
lL »
3 4P
IP'?
P7n
+ *~+
* **+
¦f-1 7
144
p 14
pn3
r?£
» ~+¦*
PPO
346
3P u
4 3
3 1 3
'^7
¦-I
¦: '¦q
i ii
A-ll
-------
PRUTiTCTNtiF dhth t pi- fn n'¦ ¦'r-:F1 frani"'. n ¦!Jr t inr-i r.ninpfit'n
9T lc 0 0 GMT
R'r
1EHM
lhp:e
PHTE
p'pE:
:upe
PPECIP
i-iihii :
PEED 7
I H'.'r P 7 I
OF
DEG
PEE" 1
H
THIC
m::
MflTFP
1-1
height
t4 T H
s 0 npl
ui' 1
7 M!'1p
7 n OMR
meter:
f M
7 mpp r
7 0 0|'!T"
I1FTEP7
+¦
+
+ +
+
+¦ x
-t- -
-+
f 4.
j1-
1-- +
₯ +
3
>-
-1-.'. 1
¦ C
1
I T( '
0. :.?4
7
1 lL 1
C.
cr
j
cr
¦
-9. 9
~ i~'
¦j c:
11.4 p. j
¦"j
51
- U
j
C . b
- 9. 4
1
L
1"'. T-P4
4
i~i
149
4
. Ij
c.' 4
<7 ¦ C.
1
1
II. f
1
~l ~r cr
L_ L. - 1
cr
1
C.
-1.9
- 9.
1
1 cr
0. P if 9
L~
ir
49
f£.
-p
. II
-4 . 0
-1. 7
1
1 ^
ri, 7 1?
,f
1
~ + ~
-3
. 4
1
-9. 0
1
j
14 ?:
0. 7 7,9
c
1
31
'j'
,_i
. 4
""" C. » O
4
1
3 4"
0. 744
IZ
~i
3 99
J
-p,
'Zi
-3.4
~d , 4
1
1 "
0 . ?'- 3
'}
4
1 0
-3
. H
C_ 1
-1.7:
1
L " 1
'"1. f 1"11"1
t7
949
1 1
r.
- 3.
\Z- > c.
1
1 '
1 j . P ^ ¦ 1
-.1
c.*
cr -1
C.
lc'
-. 4
7'
} j'.h1.
0.5 -.7
179
H -i
L Z'
- [1
-1
- ? , "r1
' ¦
1
1
Ij . P.-1
c'
c'^c
14
1
cr
t _ i
- 9. 6
rrr
1
19 -1.
'1. 7 74
C"
4
9 0
15
71
-7'. 0
~9. 0
1
1 1
0 .
~i
i
1 6
-9
. 4
- 9. 9
j
1 3 -1
0 . J11
¦4
f-,
₯ * **
1 1
4. 1
- ij
. 4
1
1 ii'i'.'
* A -*¦ 4*-
4
r 1
9 14
1 I1!
*£
-4. 3
L_ a 1
1
] ! i",
|j . 7. r"
::
1" 1
?¦ 1 0
] "<
h
. ri
-9. n
~ c
i_
1 4 5
ij. 4 ':n
L*.
7
5 P'
9 i"1
4
. n
. if
-9.4
1
! 4?
Ij c 4 -¦
C
fi
74
91
~l
0 *-f
-2. 7:
L. O
1
1 4 r,
11.1 i''t>
j
4
199
C iT.
J.
. ri
~ ¦ l1
- 1 . '?
1
1
n, 74 p
1
c!
j 1 j q
C. Z'
- 7'
*z>
"1 cr
C. _1
_ p
t
1 4-'
0.4 7';"
c
94
-3
S
- 7:. 4
0
1
143
11, 7 Q
-1
,r"
9 5 P.
rz*
rT J
- I!
Q
,f_'. ?
n
1
1 4 11
II. 0 7:7
J'
z> z> z>
- 0
cr
" 1 . 9
-u-'. u
1
1 ."'P
Ij. 5 1 If
4
1
199
i~ i
1
. c:
~9. ri
C. . C.
1
1 - cr
1 " -1
n, 7 n 11
ff
f-,
(zr
c! f:
- Ij
cr
- ij.
L. ¦ t
1
140
11. j 7
c
J
4
t" C. 1..
lT.' 3
-1
cr
-4.il
t~ ¦ 1
1
1 4 Ij
0. !- 17:
4
97 0
z' 0
-1
71
-4. 0
? . M
1
3 3 9
'i.77 0
1
:-:l
1
. 1
9 . 9
~~ j * 1
i
1 ? L1
0. .7.77
1 cr
Z. 1 J
A-12
-------
p&riio: nnDE phta Fr-i dvep gphud t ii; GLD^hTin
Hlli_-'jlT 197
HT 1 iz i'i11 'rUT
DPV
CF
MDHTH
₯- +
1
¦J
l"l
1
[¦1RhN LRP-'E PhTF
"degree : c km
o fi|'i7T'i.''f,1P 7 0 IMP
+ + + f
PPF--;ilPF PPETIF
THTOrCl' MhTE P
H
4
4
4
L
1
-4.
-4
- 0
4
-1
-1J
- i'i
C.
-4
11
-1
-4
-4
I'IF TEP
48
rr
¦?]
47
4C".
r;:
1 8
C M
+
.-:4>r
~:48
_> ¦ i
52*
f.nr
48?
T'T'O
874
7?]
8": 4
4 91"
4 n-
4Sy
!-:81
4^4
h-, '-4 t I
r?4
5 ! ¦-
'.94
i rr c
1: 11
m] r-sr, :peet>:
n :
; ijfpt 7 i'i i'i I'I v.
4
4
4
4
4
1 P
1 n
i-
1 ?
1 n
INVEPCIDH
HEIGHT
MFTEPl
f +
8 8
88 0
3 0 8
1*4
8 0 8
4T.8
884
I Sx
1 98
8 0
] 97
18 3
*¦ ~ +~
cr
Q'*!
3 78
858
¦*r CT -|
11. J
154
""i
1 0 1
A-13
-------
APPENDIX B
SEASONAL HISTOGRAMS OF THE METEOROLOGICAL PARAMETERS
DESCRIBED IN SECTION 2.2
Each histogram shows the number of occurrences of data values (ordinate)
which lie within the indicated intervals along the abscissa. Note that
each interval is defined by the label immediately adjacent to the left
of each vertical bar and the label immediately beneath each vertical
bar.
B-l
-------
CO
c+
QJ
=3 3>
CL -S
Qj
""S r+
Cl rr
3
O fD
CD rt-
<
-j. o
DJ
<"+ 3
(V
O QJ
3 13
»» ro
CO vo
ro co
00
c
-5
-+>
CD
O
0)
c+
o
CO
o
o
a>
~a
in
(X)
D>
c+
rD
NUMBER OF GC C UP EPC E.
CD
-pa
LT>
o
7C
J H
J 1
1 ?¦
1 4
15
1 f,
: i +
!". I -
l'l
- + J- T
-------
CD
I
CO
CO
r+
QJ
Q- -S
OJ
-S c+
Q_ IT"
3
O fD
fD ri-
< »¦
-> O
OJ
r+ 12
-« fD
O a>
3 13
II II
^1
O O
CT> CO
j c_n
oo
c
"5
-+>
&
o
fD
e+
O
CO
o
o
3
cr
QJ
TD
fD
^5
OJ
c+
- + + +1 1- + + +1+ + + -1-I+-H-+J
mi ii'! F'Ep hp nc nj°t h>:
J +
] o,
-------
03
I
GO
ri-
Oi
=3
Q- -s
a> <
-s <-+
CL IT"
3
o CD
CP rt
< -J-
-» r>
a>
r+ 12
-»
O QJ
=5 =3
II U
CO OJ
r\3 -ps»
4^ CO
go
c
-5
-h
CU
o
n>
r+
O
CO
o
o
en
~o
in
CD
-S
a>
c+
CD
to
a
"5
UD
MijMf Ep DF ~CCUFENCE'
: o
- 1
11
CD -
73
m 4
m ^
GO -r¦
O
7^
'-A
I 0
II
12
1 4
I O
I!
1
- -l. 4. T .
I + i- + + I
+ ~ * # ~ ~
I +
i
+
+
4-
+
-+1 ++f->-1
-------
CD
I
en
co
c+
Qj
CL -s
0> -j.
-s c~t*
Q_ 3"
3
o
fD ci"
< -J.
-» O
0>
c+ 12
05
O QJ
3 3
II II
-P» no
no u>
oo cr>
oo
a
-s
-+»
cu
o
CD
c+
o
CO
o
o
3
cr
DJ
"O
to
fO
Cli
r+
rt>
CO
c:
0)
-5
-4
o
m
CD
73
t/>
7"s
-I
I 0
J i
1 c'
1 I:
14
1
- -4" T -4- + -
rnMt'EF- n1'- orcuPFMrET
I ++++1 4- -
1 u 4- + '
-+lM.+ +.j.r4. + + +1+».-(.+ T t-4- + 4-J.
¦ j; '"1 pep gf nr c ij^'EHC e "
-------
CO
I
cr>
go
c+
Cu
=3 &>
Q-
OJ -J.
-5 c+
Q. ^y
3
O <"D
CD c~t-
<
O
Qj
r+ S
-i. fD
O OJ
^ 13
tl II
I
i» ro
po
-P=»
"--4 kO
GO
c
-5
-+>
QJ
O
(D
cn
o
QJ
"O
CO
ro
-s
Qj
c+
fD
h
OJ
O
PO
oo
O
I++V+-I
NUMPEP Grrijpprir
¦!+ + -»- + T + -«-4-+T+-r-(-+J
- + I + + + -- I
+-
+
+ + -<~»
T J- + -+--
l'l
J ¦+
1 4 .
-------
tn
Ct-
3 "3>
Q. -S
QJ -J.
r+
Q.
3
o rD
("D c+
< ->
-¦ O
CD
r+ 3
-«¦ ro
O cw
3 13
II II
I
oo
c
-s
h
QJ
O
05
cn
o
3
cr
cu
~a
CO
05
-5
Q>
M'Jf'i E'EF' DF DCCUFEMCE:
- 1
ro
o
m
- i. n
- fi. s
o
s:
11 . f!
__i.
rn
m
o. s
c+
ro
o
J . 0
"S
i ^ ¦=;
- o
ro
OJ
vo <_n
-------
UJ
1
00
00
c+
Q)
a. ~s
CD -J.
-5 r+
Q_ zr
3
O CD
cd <¦+
< ->
' o
Cj
c+ ZH
-j. a>
o oj
3 3
ii ii
o Co
O CO
ro an
oo
c
-s
-+>
QJ
o
fD
0 0
c+
' cr
o
"S
1
"-"-J
5 ft
-a
-4
cr
oo
\
* tr
CD
'4
JT. J
-5
-¦4
0 U
a>
-1
rt
I'
. J
ct>
a _
rr. Ci
CD
TO ~ :
^ '
oo
T3
m
m
0 0
-5
oo _
* ir
o _
^ 1")
to
""--s. ¦ -
75 _
- s
-
-1
lit;
r + 4- + -
NUMBEP OF HlC IJPENCEl
¦I+ + + +I+ +¦«-+!
-i- H
+
+
I -"-- + +!
Ntjf-iBEP HF nCCUFENTF:
I+ + + + J
-------
UJ
I
OJ
o
ro
r+
O
¦^J
cn
o
3
cr
a>
"O
t/>
ro
-5
&
r+
(D
to
C
3
3
fD
""S
- - . M 'J
IJ
CD - t,
73
m
CO
c~> - :«
7s
cr
t- _ i
- i >
'ill
Hllpip-EF' OF DCCUPFUCE:
¦T+ + ¦+¦+T-
. J t .
I f+ [
tVJ^BF.P HF ~ICUFEh'.TE"
¦1 +
1 ij.
-------
CD
I
CO
c+
QJ
=3 ;>
CL
CD
-5 r+
Q_ 3"
3
O fD
CD r+
< -i.
-j. o
CJ
r-t- 12
-<¦ fD
O Q>
3 3
II II
I
o ro
en CO
-P^ ^
ro
go
c
^5
-h
O)
O
fD
c+
O
o
o
3
cr
oj
"O
(/>
Oj
c+
fD
CD
PO
CO
o
7^
-l
_ 1
¦1
1
1
I I
I I-
! +
r4i.»MF.FC' DP OffUPEMlE
¦1 ++-
t
i + ~ *
± *¦ + «> * *
+ + ~
+ + + T + + + -
| -+- + 4- I 4* T + J -l- "h 1 + + 4- + [ 4- + +- -e- T
! -i11; 1FJFF n- ijC r UPEMC E 7
+
+
¦H
f
+
+
-t-
I f
P ''I ,
-------
CO
c
-s
-+>
O"
o
(D
r+
O
o
o
-4
C3
I
QJ
TJ
CO
CD
^5
a>
c+
fD
13
c+
ro
^5
O
rn
o
TO
CO
O
7s
- {
- 1
- I
CO
<-+
a»
3 D>
Q-
CU -J.
-s Ct
Q_ =3"
3
O
c+ 3
-i. CO
o CD
Z5 3
II II
I
4
4
-4
oj cr>
f4 CO
MLiMBEP Dc nrr IJPENCE'f
-------
CO
1
oo
r+
QJ
3 >
O- -5
Qj
"5 c+
CL
3
O (D
rD r+
<
-J. o
CU
<"+ 3
-< ro
O CU
o ro
U1 CD
cpi ro
-£»
oo
c
"5
h
Qj
O
ro
o
o
DJ
"O
in
ro
Of
c+
ft)
{/)
"O
^5
«_i,
Z3
tQ
MJM^EP DP nCCUPEtiCE:
o
m
go
o
7^
i ! i i 1 I
J + + + -r I
T+ + T+I'
. cl.1
5 6 0
4 0 0
£40
r' ij
'?.? ij
7
1
c 1": C1
T + *
- 1.
44 M
Z O » i
¦J- -»
- i
1 c1 i'i
+
0
0
M
- i 1
3 0
+
-11
-I- 4;"!
X
- 1"'
4,-11
-f- * *
- 0
il
- ft
1 o fl
- f!
0 01 '
¦h
1 F> 0
I + +
-+ + -<- + y -r 4- -t- | + + -*-+T++-t-r [
hun^ff np nc cu-'Epc e t
+
+
4-
+¦
+
+
+
+
+¦
T +
-------
00
I
GO
rt
QJ
3 n=»
CL -5
CD -j.
c+
Q_ ZT
O
(D
<
0)
0)
O QJ
13 3
II \\
o ro
cjn
i» cn
(» oo
CO
c
-h
CD
o
fD
c+
O
'I
O
o
QJ
"O
CO
(T>
-*>
&»
c+
fD
rD
-5
M U M "F c P CP CL L UPEMLE:-
oo
o
7^:
-4
o _ r
o
33
- 1
~ 1
-i
_ i
~ i
-1
- ri
-11
- M
~ ' J
- tl
I"!
,=: 4
II4
4 IJ
='4
-14
r 4
L tj.
0 0
+ +!
4- -*¦*¦* )
+
1-
+
+
-f-
I + + -*- + I
14,
-------
UD
I
I>
GO
r+
QJ
3
:>
Q-
a>
>.
-s
<-h
Q_
3T
3
O
0)
CD
ci-
<
_i.
i.
n
Cu
C~t"
_j.
ro
O
QJ
Z2
II
II
l»
O
ro
cn
-F*
ro
CT>
00
ro
ro
fNJ
"O
-"5
CO
to
cn
c
-5
rD
o
7*r
3
ro
CO
00
Wl.r-iBFP riF ~ f iJFEHf E I
po
CO
L ".i1:
i ? i
1 0?"
1 '"i a
1 i'4,:i
1 !"l f f-l
1 0 o L
I !j ~ i
' UT ?
1 >'?
1 -?5
¦|? 1
l"! '7-< 7'
i 0
1 fi9
111?
1 ] -J 1
1 i2r
ii?:
11":?
1 1 l1
1 15 1
1. 1 57
] 1 -
1 1 r 'r4
1 175
1 1]
-------
CD
I
CO
C~t"
QJ
Q_
QJ
1.
-5
ci-
Q.
3"
3
O
0)
o
c+
<
j.
1.
O
a»
<-t-
CD
O
CD
3
=5
11
II
H-»
o
t»
-pi
CO
(J1
TJ
CD
(/>
00
c
o
7T
Z3
ro
cn
(/>
r+
a>
-s
NUMBER DP" DCCUPEMCE:
CO
'-1
O
? 3
-3
1 0
1 f«
i"' 1
I-! CT
1 ! _t
l r>
"'l 'H
1 0
1 ?
i o
X 1
1 fl
L ' 1
c'l
I I
i»:
II
1 t;
1'
i t;
i \
i f
11*
i r
i ¦;
11"
i »;
¦4 1
JQ
?1
1 4- ~r + 4-
-P*
CTi PO
-P»
-------
hlmbep ~ p nccijpemce:
DJ
I
on
r+
QJ
3
D_
-s
DJ
1.
-"5
r+
a.
rr
3
o
fD
fD
r+
<
_j .
j.
O
QJ
i.
a>
O
a>
d
II
ii
i'
o
ro
OJ
(1
CO
o
-P»
o
4^
o
~0
"5
(X>
(/)
CO
c
ro
o
7T
ft)
t/)
on
"O
-5
to
l r
i1;
i i"
TO
GO
t t
1 :
I i
i i
x
| '¦
i*:
i i'
i i
II
1 i
U
rtc
1_ r
1T
4i.
4 *"
"=¦'.1
1 * (
i r
i L1:
i w
i ?f
' T + + ^ T J
1 + + + + T
¦ -*¦ ¦* -* *-*¦
t + -
I +<
i 4-
+
4-
T+ + -<-+I+ + + -J-l +¦
1 0.
-------
UJ
I
CO
r+
o>
!£>
Q_
-5
QJ
i.
"5
r+
CL
=T
3
O
Q
rr>
r+
<
i.
O
QJ
c+
IS
__i,
0)
O
CLI
3
Z3
II
II
h'
I>
h-»
ro
CX>
CO
CO
ro
r\3
CTt
ro
TD
-s
ro
CO
-s
(D
O
13
ro
LO
CO
CO
c
fl)
number cf dccupeucf.:
TO
CO
1
1. t'
i8 0
) ?4
t'^r'
1 1 M fl
i i 04
1 1 0'?:
U!.I
1 1 1 *
1 1 d U
1 1 ?¦*
i ir-v
1 J ?
1 ] ?.-
1 1 4 i i
i 1-4
i-L'--
i i ^
i 1
U - 0
1 1 -,ul
1 + + +-
1 +
~ ~ + ~
¦ *¦ *
I+'
4- i - -i- -M -t- + + + [+ + +4-T
rHu:,-,p,^p np rjr r i.ipt MC E'
l T
1.-.
-------
CO
I1
CO
TJ
-5
a>
o
c+
G>
cr
0>
(T>
go
ri-
oi
Q_ -5
fl)
-5 r+
Q_ rr
3
O CD
fD <-+
< -»
-J. O
a>
r+
_». fD
O QJ
13 3
II II
O O
»4 CO
ro
^4 O
nun pep of accuFFhCE:-
-------
CD
I
00
c+
Q)
Q_ -5
QJ -J.
-"5 c+
cl zr
3
O CD
fD c+
< -j-
-« o
Q)
rt 3
rD
O CD
13 Z3
II II
o o
O 1
oo no
"O
-5
CD
O
*0
r+
Qj
cr
rD
S
CD
r+
rD
-5
s:
(D
-s
"UMBFP GF DCCUPFIIC-FZ
o
1 , 'J U
04
ij .£*
0 ¦?
1 H
1
1 4
1 r-
\
a .j.
4
1 +
+
+
+
+
1 +
:> I+ + -
fl
EP nF ccc IJF FHL FI _
+
+
1 4 ,
-------
DO
I
r\)
o
LT)
c+
qj
=3
Q_ -S
Qj -J.
r+
Q_ 13"
3
o fD
CD c~t-
<
-> o
QJ
c+ IS
O QJ
II H
° P
i» ro
o
i£> en
~a
-5
ro
o
QJ
O"
s:
QJ
ct*
CD
CO
-a
Z3
to
NUHFER np nCCUPEriCE
I. II
rlc'4 +¦
114'-: +¦
11 fr
1 £.' 0
1 44
11£.£'
1 ?.='
1
4 0
44
4 :: -r *¦¦*¦*¦» *¦
4"-,: + * + -*»»<
4 0
~ri4
~ 4-
+ ~ ¦*
.'.II' +
¦ £4 I
r -f- -i* -f- r -+¦ ^
M'jr^b'FP nc HC C U^'FNf r ;
+
+
f
4-
+
¦1 +
1 i">.
-------
CD
ro
co
r+
CU
a. -s
OI '
-5 c-t-
d- zt
3
O CD
fD ri*
<
' O
D>
r+ 3
0)
O Cu
13 3
II II
O O
H-1 -P»
cn ud
-P» en
~o
-s
n>
o
QJ
cr
ft>
QJ
CD
"5
(/)
C
3
3
a>
n. -'Hi'
o
LT^r-M!
?¦ ^ 0
3 -I M
z>
-------
GO
C
-s
o>
o
a>
3
o.
(/>
QJ
CD
I
no
ro
GO
c+
DJ
3 :e»
Q- -5
Qj _i.
""5 c-t-
Q_ 3"
3
O fD
fD r+
<
' o
QJ
c+ 12
fD
O CD
3 3
II II
>» ro
CO CTi
GO
GO
I
T
j ¦
~h
+
+
- j -i- -k J + Hh J 4- -4- + + -f
4-
' +
-------
CO
c
-*5
h
Q>
O
0>
3
Q.
to
M'JM E'EP HF OC C UPEMl E '
D3
I
no
CO
CO
c+
O)
3
Q_ -5
Qj
-S r-t-
Q_ 3"
3
O CD
("D c~t-
< -<
i. r>
a>
n- 3
a>
o CU
3 3
N II
fD
i L
i I
i J^ + 4--uT + +^-+t
! J- «~ 4 h
t I-
f I
Ni ^pfp gp .n'jy.iPF.r^CE:
Co en
O 4^
-------
CO
C
"5
-h
CX>
o
(D
CL
in
iHiU'itFF nf nc cupehi-^:
to
i
ro
-P*
CO
c+
QJ
3 3=»
CL -5
Qj
-5 rt
Q_ 3"
3
o ro
("D c+
<
-j. n
QJ
ri- 3
-u fD
O Cu
3 3
II II
r\3 co
CO CTi
CO ~»
l/>
¦a
-s
13
lQ
00
'A
I r\
II
u l.l
I I -h + H
! -4- * * ~ H
t ¦+¦
+ + + + [- + -+!>-^+1+-.+
rJ-lji'it'r.p Gr rjr"luPEr-r
1 + + + +I-
+
¦ 0.
-------
go
c:
-5
CL»
o
fD
CL
00
"UjM REc GF DC CUPEf JC e.;
CO
I
ro
en
GO
c+
o>
13
Q-
Q) _i.
-s c+
CL IT
3
O fD
fD r+
< -j.
_i. n
0>
c+ 2
ro
o CD
a>
1 i:
1 1
. + + +J-1 ¦
¦ I I ¦
¦1 +
r 4 i J i -i PP F Or DC^UrEfiiE:
- ] +
d 0 ,
-t> rv)
OD
CO UD
-------
*-0
o
o
3
Q-
oo
h
a>
4.
DO
ro
CTi
GO
r+
QJ
=3 It-
CL -5
QJ ->
n-
Cl zr
3
o fD
fD c+
< -«
O
a>
c-+ 3
a>
o a»
3 :3
ii ti
co en
en
CO cn
-P* CO
-I
1 !l
! 1
] p
1 ?:
1 4
1 5
1'
r
i,
1 M
M!.1''1 PEP D- DC CL'PEHC E 7
] + + + +!+-- + ¦+-1 + + J- + I ¦
¦I +
+
+
₯
t-
+
+
4-
+
4-
-(-+] ₯+-
+
+
-V+l
¦1-4--+ T +-4h + T
-------
CO
I
ro
j
o
o
3
a*
3
Q_
l/>
15
CD
"5
MHMBF* GF ~rrUKEf^'F--:-
1 . I!
1 1
1 £
1 ?
14
1 -
CO
i r. s
c+
QJ
i ¦?.1
3
3=*
CL
~5
CD
i.
-5
r+
- *1J
CX
3T
..
3
,:1 i . '
o
_ .
a>
c~i~
~ > 1
<
i.
- _
i.
O
, :
OJ
c+
2
.4- . i
<
ro
O
a*
j_' ^ . I
3
i_ i
II
ii
-P*
00
o
o
no
cn
ro
UD
» ¦*¦ 4» * +
+¦
+
+
+
+
-------
CD
I
ro
co
CO
c+
Cu
=3 3=»
Q. -S
Cu -»
"5 r+
a. rr
B
O fD
fD c+
< -«
n
a>
r+ 2
-j. a>
O Oi
3 =5
II II
-P* 03
^4 O
<*o
*sD CO
O
o
3
cr
Q_
(/)
co
"U
13
CO
mumpep OF Grri.iPENr.E"
I 0
II
i v
3 ^ t;
00 i4.
IT'
1
1_ 7
1 9
i ci
c r!
f~ L
\ 1
I+ + - + I
- + + L-
»^ ~
~ *¦ *--*-*¦ +
-+ 1 + -
+
+
+
+
+
+
-------
DO
I
r>o
vo
co
c+
QJ
Q- -5
CD -J.
*"5 c+
Q_ 3"
3
o ro
CD c+
< -*
-j. o
Q;
c+ ^
-j. fD
O Q>
3 3
II II
00 On
CT> -P*
» -P*
O CO
O
o
3
O"
£
j.
3
Q.
CO
CO
c
CD
-s
!.l lit.:
0 0 0
u '~i 0
i?1' 0
1! 11 fl
0 U fi
0 0 l"i
fl fl fl
NUHEEF nrCUPEHi-E:
] + + -L--t-'
M
1"' rl ! »
0 t 1 i"l
+
1 u
0 i"i 0
+
11
fi 0
¦+
12
0 n 0
4-
4-
I 7:
m i"' Ij
4-
14
f! U IJ
4-
-1-
.1. 5
¦J 1"! I"i
+
4-
1 c.
fl fl "i
+
4-
1 7
\_. fi
4-
4-
1?
i ,1 11 l"!
4-
-L.
1 ?
11 ,"i
4-
+
.V 0
i ' f> fi
4
4-
: 1 t t
+
-t-
£2
ft M
-i.
'"l fi
4. -*¦ -* -»
4-
->L
-J 'J
+
T- ^
'' i"; i,
-r
-H
¦ T ~t 1 1 T ¦
M'J"i
-+ + t .
: pr,r c
-------
CD
I
CO
o
00
c
"5
b
QJ
n
fD
I
cr
QJ
t/i
Q_
3
<
fD
CO
fD
tQ
mu^bep of nrcuPEr^CF;
i n u
1 £'<~
1 ?. o
175
=* n i"i
m
_» .*
4 !.: i"i
4 .7" ?'
CO
- -
r+
4"
i"; I.1
o»
r3
^ t"i P
M 1 l 1 ,
Q_
"5
QJ
j.
"t cr
! ! i ! f;
-5
r+
" --
Q.
Z3-
I'M C\
3
o
fD
'-"l r-i
l": i i !
fD
c+
"
<
i.
t '' v
i;. \< u
1.
O
QJ
lT Z'
I 1 1 ! 1
c+
3
_ 1_ _
_j.
fD
!~l
'J C ' 1
O
QJ
3
3
1 f
m ri ri
II
II
\»
ro
U1
cn
^£>
CO
--vj
ro
cn
CO
ro
CO
-------
CD
I
CO
CO
c+
OJ
3 >
O. -S
CD _i.
-s ct-
CL 3"
3
O fD
fD c+
< -»
o
o>
r+ 3
-J. fD
O o>
II II
no co
Co ro
cn co
00 cr>
no ko
CO
c
-5
-+i
QJ
O
CD
I
CT
O)
cn
ro
Q_
3
<
ro
-s
CO
O
13
(D
_j .
tQ
c+
fD
-s
.3 I""1
0 u
t 0 Ct
'J {1
u '"1
! 4 0
0 n
1 x i"i
ft
3 d.I_s
fl IJ
2 b n
0 0
3 0 U
!~1 Ij
34 M
f| i i
\1.'
4 d. '"i
!_} ! 1
!"¦ (i
0 0
5 f! 1":
n ¦ ¦
IETERS
54 U
8 t"i
"£> el 0
"£ - t"!
'"i: i
0 ':
ft '"i
0 'j
71 f 0
1": 11
T4i",
0 T1
iMi
» ! '
'7' ?'
A
!_ i 1
; i f i
0*s
r411 ¦"*
'"1 Ij
-4 o
.! ft
¦ i
0 0
l r. J o
f: f!
1 f i 1
0 i".
numbep np dcoupeuce:
~++
+ -a-
-m- 9-
¦T-t- + -r+J-
T+ + -
+ y ++-i-f j + t + ] ¦
¦1 +
+
+
+
+
+
+
¦1 +
I 0.
-------
MUMPER DF DC C UF EMC E 7
m
co
i
CO
ro
CO
c+
QJ
=3 :>
Q_ "*5
Qj -J.
CL ZT
3
O fD
ro <~+
< -j-
o
O)
r+ 3
n>
O CD
13 3
CO '1
CTt I»
WO CO
cn wo
i> cn
0J
r>
fD
1
cr
o>
c/>
CD
Q_
<
fD
-5
C/>
i.
O
3
(D
_1.
CO
3"
r+
(/i
cn
"O
~5
3
U2>
1 0 >'
1.2:
1 j C
TO
C/>
4 o r
4cl~
45
47"
5 i"1
j111
i \ n
!_¦' fi
0 M
fi M
"i "!
.Mtfl j
T-H++-+ f+ + 4-+I +
nrCoPeni'e;
*-
+
¦ T 4- 4- 4- y 4-
3 n,
-------
MUI'IPER C1F DCCUF'Ehf F~
D3
I
oo
OJ
GO
c+
OJ
Cl
a> -j.
-5 <t-
cl =r
3
O CD
CD rt
< '¦
n
&
c+ 3:
-1. (V
O DJ
3 3
II II
DJ
O
CD
l
D"
Oj
CO
fD
CL
3
<
CD
-5
00
o'
=3
a>
UD
Z3"
c+
oo
fD
-s
pa
C/0
^ 11
cr
i
i i"i o
15 0
175
n i i
Z> -Z' cr,
-i I 1
: i v
4-im
4 E S
4 ?. > ¦
a rs
+
+
+
+
+
+
+
¦h
±.
+
+
+
¦ J"
~ F Di
!JF FMC E
I4 kO
CT> 00
ro ro
v^D »1
cr> cn
------- |