1. Introduction
This is a study of the difference in
detection of RWIS vs. AWOS/ASOS. The
data is concentrated in an area in Iowa where it has been noticed the
temperatures and dewpoints are on average higher than in other parts of
Iowa. This has become known as the
southwest Iowa heat pool. There are no
publications officially on this subject, but hopefully this study will spark
interest in the area. The point of the
study will be to compare the dewpoints and temperatures of two cities in the
area of heating and compare to see if both systems pick up the heat.
The base of the study is on the Iowa
Environmental Mesonet in development here at Iowa State University. This is very similar to the Oklahoma
Mesonet, but the Iowa mesonet is being constructed among a group of cooperative
systems. Iowa DOT, NWS, and Iowa State
University stations are some of the partners in the mesonet.
There two cites in the study are Creston
and Red Oak. There are two reasons for
picking these two cities. One they are
in southwest Iowa, and they both have relatively close RWIS and AWOS
stations. Creston contains the AWOS
station KCSQ, and the RWIS station RCRE.
Red Oak contains the stations: AWOS-KRDK & RWIS-RRED. The Red Oak Stations are only about 13 miles
from each other; the Creston stations are only 1-2 miles apart.
The AWOS stations, KCSQ & KRDK, are
certified AWOS-III stations, and FAA certified. Theses stations take dewpoints in the same way; A wetbulb
thermometer that has a fan to draw air over the bulb takes dewpoints. These instruments follow the FAA required
error of plus-minus 1.5 degrees.
The RWIS finds the dewpoint in a different
manner. It uses a relative humidity
sensor utilizing an error of plus-minus 5%.
This sensor works on the human hair technique. The thickness of the hair is the indication of relative
humidity. Then it plugs the thickness
into a function and arrives with the dewpoint.
The temperature gauge also has a relative error of .5 degrees
(plus-minus).
Included
are two maps of the Iowa Environmental Mesonet. On map 1.1 the Creston AWOS and Red Oak RWIS and AWOS are clearly
marked in the southwest portion of Iowa.
The two maps show an example of the suspected Iowa heat pool. Notice that in map 1.1 the southwest part of
Iowa is experiencing northwest flow at ten to fifteen knots. A northwest flow on average indicates cooler
temperatures, but the stations are reporting mid to upper 80’s. The area surrounding the southwest Iowa is
reporting mid to lower 70’s. Map 1.2
shows a similar plot, but winds have lessened to 5 knots on average. This study intends to find if the RWIS data
indicates the same heat as the AWOS.
2. Data Analysis
The data was collected from May 2001 till
October 2001. The output contained
dewpoints, temperatures, valid UTC, date, and station identifier. The data was then broke down. Each data set for the stations was separated
by day. The data had to be cleaned up
with some quality control. Days with
hourly reports missing from 18 (UTC) and 23 (UTC) are thrown out. That averaged about 3 days a month. Then the daily dewpoint average, daily
dewpoint max, daily temperature max, and daily temperature average were found
each day from 0 (UTC) to 23 (UTC) each day.
After that the data for the corresponding cities were combined, omitting
days that RWIS had and AWOS didn’t and vise versa. Then the data was sorted by one of the fields for
continuity. Finally the data was
interpreted on a city by city bases first of which was Creston.
First analysis was on Creston Temperature
trends for both RWIS and AWOS (figure 2.1).
The data showed for trend lines of all fields the Temperature max for
CSQ was the highest. The slope of the
trend line for Temperature max CSQ was comparable to the slope of the trend
line for RCRE temperature max. The
temperature averages for CSQ and RCRE showed the same trends as max but with
slightly smaller slopes. On the
dewpoints (figure 2.2) the averages for each station had comparable slopes, but
the maxes had slopes .01 apart. This was the largest slope difference so
far. Next the RWIS-AWOS difference was
found for each parameter (figure 2.3).
The temperature average difference trend line showed that the temperature
average had the largest difference, which was also in the negative. The next largest difference was in the
temperature max difference. The other
two, dewpoint average and dewpoint max, displayed positive differences.
If Creston is looked at a confidence
interval view-point (figures 2.4 & 2.5), the biggest confidence interval is
owned by the temperature max of RCRE.
The AWOS station for temperature averages and max had smaller confidence
intervals for this sample. This was
reversed on the confidence interval for dewpoints in Creston. The RWIS had the smaller confidence intervals
then.
Next the data was cut down to maximum
temperatures above 80. The comparison
was made again for the fields (figure 2.6).
The trend lines all had slopes within plus or minus .02. The relative patterns remained the same for
the placement of the trend lines RWIS vs. AWOS. The confidence intervals (figure 2.7) switched for dewpoint max
and temperature average. Also the
temperature max for both stations equalized.
The next step was to break the data down to above 90, but the data for
both stations became to small for reliable conclusion
For Red Oak the same analysis was made for
each field. The trend lines for the temperature comparison (figure 2.8) showed
some interesting features. Slopes of
RWIS trend lines for temperature were .02 and .03 difference. So as the
temperature increases the RWIS changes faster.
This was true for the averages and maximums too.
This was the exact opposite for the
dewpoint comparison (figure 2.9). The
AWOS stations trend line slopes were higher than the RWIS, and they are
increasing with temperature.
This makes the field differences very
interesting (figure 2.10). Unlike the
Creston dewpoint differences, the Red Oak dewpoint differences are negative,
and increasing negative with temperature increase. The temperature differences are sloping positive with temperature
increase, but they too are on the negative side.
Now for the confidence intervals for Red
Oak. The temperature confidence
interval (figure 2.11) the RWIS stations are higher for both max and average,
both are even higher than the other two fields, AWOS max and average,
combined. On average the confidence
values for temperature where .2 higher than Creston intervals.
Just like for the dewpoints, the
confidence intervals (figure 2.12) for Red Oak are topped by the AWOS
stations. The confidence intervals for
dewpoints are also higher. The
intervals are .3 to .4 higher than those for Creston.
With the increasing tendices for Red Oak,
the above 80 chart should be interesting.
Just like for Creston the trend lines tend to get closer. Once again the field comparison (figure
2.13) showed that the max temperatures for RWIS eventually surpasses the AWOS
trend line.
These discrepancies also show up in the
confidence intervals (figure 2.14). The
confidence intervals do decrease for all fields by about .4, but the difference
between the individual fields increase.
Also noticed was the RWIS confidence level for each field is higher than
the interval for the AWOS. Once again
though to take the temperatures above 90, the data became to small for reliable
conclusion.
3. Results
For Creston, the temperature comparison
(figure 2.1) display that on average the RWIS data for both temperature max and
temperature average is colder than AWOS data.
The Creston dewpoint comparison (figure 2.2) shows the opposite. On average the AWOS data is colder than the
RWIS data, but they are very close in difference. The Creston difference (figure 2.3) showed that the dewpoints are
very close and positive, and the temperature differences are negative which
follows that RWIS is colder for temperatures.
For temperature intervals, the RWIS data
showed a larger interval for temperature fields (figure 2.4). This means that on average the RWIS data is
less accurate by about .1 degrees. Now
for the dewpoint confidence interval (figure 2.5), the AWOS data has a larger
confidence interval than RWIS for each field, but the variance is very small
with each field difference around .05.
So all field are equally confident.
The field trends for temperatures above 80
show differences in the max fields and average fields, but the dewpoint fields
are still very close. The trend line
differences are now more linear than before so above 90 would look a lot like
above 80. That’s another reason for not
going above 90 for analysis. The
confidence intervals (figure 2.7) are less than the intervals for each field
before the above 80 analysis, but the confidence intervals are larger in their
individual difference. The differences
are not large though so the fields are relatively the same confidence for RWIS
and AWOS for matching fields. Over all
Creston did not show a significant difference in higher temperature values.
There are differences in the Red Oak
data. First the temperature comparison
(figure 2.8) on average RWIS is colder than AWOS, but at higher temps the
opposite is true. For the temperature
averages though RWIS is constantly colder than AWOS even at higher
temperatures. The dewpoints show that
the RWIS is colder than AWOS on average.
This is opposite of Creston.
Also the max and averages are getting farther apart as the temperature
climb. The differences comparison
(figure 2.10) displays the same characteristics as observed before. The RWIS data is colder on average except
for the temperature max at higher temperatures. The temperature confidence fields for Red Oak show that the
confidence intervals for the RWIS data is less accurate than AWOS, and the
fields differ by about .1. The dewpoint
confidence intervals (figure 2.12) show the same that the RWIS data is less
accurate than AWOS, but interesting the dewpoint max for RWIS is .1-.15 less
accurate than AWOS. This is the largest
inaccuracy.
Now for the above 80 intervals. In temperature and dewpoint comparison
(figure 2.13), the data shows the same trends, as did the first
comparison. RWIS surpassed AWOS at
higher temperatures for max analysis.
The other fields are leveling out when it comes to slopes. The confidence levels (figure 2.14)
decreased for all fields, but the differences once again got larger. The temperature max for each site is the
lowest, but the RWIS is .1 higher than AWOS.
This is the same as before the above 80 data. So accuracy stays the same above 80, but the RWIS wants to trend
higher than AWOS for temperature maxes.
The data was too small for an above 90 analysis, which would probably
show that RWIS is warmer in Red Oak than AWOS.
4. Summary
The
data showed that RWIS temperatures are on average colder than AWOS. The only exception was RWIS in Red Oak at
high temperature. Hopefully with more data this can be discounted or found
true. The dewpoints showed different
results in each city. Creston RWIS was
very close to AWOS if the temperatures were a little warmer, and in Red Oak RWIS
was colder. The confidence of the data
showed that RWIS was less accurate in Creston, and AWOS was less accurate for
dewpoints. At temperatures above 80 the
RWIS was less accurate for dewpoints, and the temperatures ended up pretty
close on accuracy. Red Oak RWIS was
less accurate on temperatures, and dewpoints are less accurate with AWOS. The intervals were very close though. Then AWOS confidence showed that above 80
RWIS was less accurate for all fields, but once again they were very close at
times. Also Red Oak showed that there
was a tendency for RWIS to get warmer at temperature maximums.
This study shows there is a difference in
RWIS and AWOS detection for the heating effect in southwest Iowa. The difference in detection is conditional
though. Hopefully this study will spark
interest in the heat pool, and more data being archived can lead to more
accurate analysis of the detection of the heat pool effect.
REFERENCES
Humphrey, M. D., J. D. Istok,
J. Y. Lee, J. A. Hevesi, and A. L. Flint, 1997: A New Method for Automated
Dynamic Calibration of Tipping-Bucket Rain Gauge. J. Atmos. Oceanic
Technol., 16, 1513-1519.
Iowa State University. Iowa Enviromental Mesonet.
http://www.mesonet.agron.iastate.edu/archive/, September 20, 2001.
Iowa State University. Iowa Enviromental Mesonet.
http://www.mesonet.agron.iastate.edu/archive/, September 21, 2001.
Mandel, Eric, 1975: An Early
Look at the Development of an Unmanned Automated Surface Aviation Weather
Observation System. Bull. Amer.
Meteor. Soc., 56, 979-983.
McGinn, S.M., and H.D.J.
McLean, 1995: Monitoring Free-Water Evaporation at Automated Weather Stations. J.
Atmos. Oceanic Technol., 12, 1363-1366.
Richardson, Scott J., 1995:
Automated Temperature and Relative Huimidity Calibrations for the Oklahoma
Mesonetwork. J. Atmos. Oceanic Technol., 14, 951-959.
Snyder, R.L., P.W. Brown,
K.G. Hubbard, and S.J. Meyer, 1996: A Guide to Automated Weather Station Networks in North America.
Advances in Bioclimatology 4, Springer,
1-62.
