lowa department of transportation · 2009. 9. 21. · fog formation fog originates when the ambient...
TRANSCRIPT
Development and Implementation of an Expert System for Forecasting Fog on US Highway 30 in Cedar Rapids, lowa
Final Report
September 1996
IDOT Research Project HR-357
lowa Department of Transportation
Eugene S. Takle, Principal Investigator Professor of Atmospheric Science
301 0 Agronomy Hall lowa State University
Ames, IA 5001 1
Iowa Department of Transportation Research Project HR-357
Development and implementation of an expert system for forecasting fog on US Highway 30 in Cedar Rapids
Eugene S. TaMe, Principal Investigator Professor of Atmospheric Science
3010 Agronomy Hall Iowa State University
Ames, LA 5001 1 Tel: (515) 294-9871 Fax: (515) 294-3163
Final Report September 1996
Abstract:
Several accidents, some involving fatalities, have occurred on U. S. Highway 30 near the Archer Daniels Midland Company (ASM) Corn Sweeteners plant in Cedar Rapids, Iowa. A contributing factor to many of these accidents has been the large amounts of water (vapor and liquid) emitted from multiple sources at ADM's facility located along the south side of the highway. Weather and road-closure data acquired from D O T have been used to develop a database of meteorological conditions preceding and accompanying closure of Highway 30 in Cedar Rapids. An expert system and a FORTRAN program were developed as aids in decision-making with regard to closure of Highway 30 near the plant. The computer programs were delivered to Freese-Notis Associates in Des Moines, Mr. James Phinney, Residence Maintenance Engineer in Cedar Rapids, and Surface Systems, Inc, in St. Louis for testing, evaluation, and final deployment. Reports from D O T personnel and D O T contract meteorologists indicate the decision tools have been successfully implemented and were judged to be helpful in forecasting road closures and in reducing costs and personnel time in monitoring the roadway.
ADM Pknt Setting and Emissions
The ADM corn sweeteners plant is on the southern edge of Cedar Rapid and is located along the south side of US Highway 30, an elevated, divided, 4-lane roadway of width 35 m and elevation 9 m above the plant. The plant consists of several sets of cooling towers, grain-dryer stacks, and water-treatment ponds. More details on the plant and its operation can be found in Thomson (1995). The cooling-tower complex closest to the roadway (about 150 m south) is a 7-cell linear mechanical-draft crossflow-type alcohol cooling tower #4 that had been observed by IDOT personnel to contribute most to the reduced visibility on the roadway. When this study began in December 1992, this cooling tower complex was operated frequently, if not continuously, during the fall, winter, and spring when cool temperatures and high ambient humidities could potentially combine with tower effluent to produce copious amounts of fog. By December 1993, however, the 7-cell unit #4 was being used only sporadically during the cold season, reducing (but not eliminating) the need to close the road at this time of year. ADM personnel advised us that use of unit #4 would depend on demand and could be put back in full operation. Also, even though occurrences of fog on the roadway were reduced, IDOT personnel were obligated to continue a high volume of roadway monitoring during the winter season.
Fog Formation
Fog originates when the ambient temperature and dewpoint temperature become identical (or nearly so), provided that sufficient condensation nuclei are available. Cooling tower fogs occur when a moisture vlume from a cooling tower is advected to ground level. While natural fogs " - - generally require small dewpoint depressions (temperature minus dewpoint temperature), cooling-tower fogs can occur with relatively large dewpoint depressions of more than 15' F. ow ever, roadway visibility does not become abroblkm if ambient conditions allow the copious amounts of cooling tower fog to rise and dissipate. Although some cooling-tower plumes can be large at these larger dewpoint depressions, observations indicated that a very small dewpoint depression is required to cause the ground fogging along US Highway 30 near the ADM plant.
Data Collection
IDOT began recording conditions of potential low visibility on Highway 30 in Cedar Rapids during the winter of 1989-90. A Surface Systems, Inc. (SSI) weather station at the site provided temperature, relative humidity, wind speed, and wind direction data. Visual estimates were made of the position of the vapor plumes relative to the roadway and its impact on driving visibility. Roadway closurelre-opening times and current weather conditions also were recorded with the plume observations.
Hourly surface weather observations from Cedar Rapids, Waterloo, Des Moines, Fort Dodge, Omaha, Sioux City, Ottumwa, and Mason City were archived at ISU to provide supplementary data for determining general weather conditions throughout the state. The Cedar Rapids airport is located south of the ADM plant about 4.8 km (3 mi). Comparison of portions (300 h) of the
two primary data sets (Cedar Rapids airport and SSYIDOT) revealed no significant difference in the observed temperature, dewpoint, or winds except that the SSI temperature and dewpoint sensors tended to give values about 1 degree F warmer than the airport. Additional airport data such as values and trends of cloud ceiling and measured visibility eventually were determined to be good indicators of potential fog formation.
IDOT data from October 1989 through March 1994 contain 2153 hours of observations including 27 road-closure events. Of the 27 closure events only 25 events were used due to questionable and missing IDOT data surrounding two events. A summary of the closure and monitoring information is shown in Table 1. Due to missing data, roughly 30% of the monitoring periods after November 1991 are without surface weather observations. All the available data that corresponded to periods of monitoring or road closure by IDOT personnel were used to further define the atmospheric conditions at the ADM site.
Table 1. Closure and monitoring events recorded by IDOT.
1989-1990 1990-1991 1991-1992 1992-1993 1993-1994 Totals
Road Closure 3 Ciosures 7 Closures 12 Closures 4 Closures 1 Closure 27 Closures Events 31 h 77 h 126 h 46 h 4 h 284 h
Monitoring 25 Times 39 Times 48 Times 26 Times 17 Times 155 Times Events 216h 616h 740 h 459 h 122 h 2153 h
More details on the analysis and interpretation of these data are given in the masters thesis of Paul Thompson (Thomson, 1995)
Revised criteria for fog formatio~z
An initial study of the cooling-tower fogs reducing visibility along U.S. Highway 30 was prepared by Radian Corporation in August of 1989. However, forecasts of cooling-tower fog based on criteria from the Radian report are too conservative and can lead to an over-prediction of fogging and a large number of 'false alarms'. This required excessive monitoring of the roadway by IDOT personnel. Analysis of data from 1989 through 1994 allowed us to revise the criteria for weather conditions accompanying and preceding the need for closure. According to on-site IDOT personnel, the revised criteria, which were forwarded to IDOT's contract meteorologists for their operational forecasts, contributed to improved forecasts from the meteorological consultants.
The revised criteria for closure are as follows: * Air temperature: >200Fand<500F * Dewpoint depression: <2" F * Wind direction: 1200 to 2700 * Wind speed: 23 knots * General visibility: < 2 mi * Cloud ceiling: < 1000 ft
Software Developrnelzt
Expert systems are computer programs developed to solve real-world problems using knowledge gained from human experts. The system seeks to capture enough of the human specialist's knowledge so it too will solve problems expertly. Specifically, an expert system solves problems traditionally requiring a human expert and does so using a model of human reasoning to reach the same conclusion as a human expert.
The final version of the expert system developed for the project separates the four meteorological inputs (temperature, dewpoint, wind speed, and wind direction) to produce a probability of each variable to cause fog along U.S. Highway 30 (see Appendix C for complete program listing). These are then multiplied together, along with a correction for dewpoint depression, to give an overall probability that the cooling tower plume will trigger a road closure. The system decides (yeslno), based on this overall probability, if the roadway may need to be closed. The actual rules and probabilities were developed using discriminate analysis techniques and trial-and-error methods to develop the best combination of accuracy (false alarm rate vs. missed closure events).
For the evaluation dataset, the system predicted 90% of the road closures while giving a false alarm rate of 15% during the same period. This evaluation assumes a perfect forecast, since the actual observed conditions were used to test the system. In daily operations, it is likely the system would provide more false alarms and a slightly lower success rate at predicting road closures. Because the actual conditions for a closure are very specific and must be forecast very precisely, any errors in the forecast could have significant impact on the accuracy of the expert system to forecast road closures.
The Expert System was delivered to personnel at Freese-Notis and IDOT in Cedar Rapids in December 1993. However, starting about this same time, ADM discontinued use of the alcohol #4 cooling tower that was the main source of water vapor along U.S. Highway 30. This caused the expert system to over-predict closure events and lowered the confidence of the forecasters using it. For this reason, we developed a supplemental procedure for fog prediction.
The development of a FORTRAN model to forecast plume behavior was started during the fall of 1994. The purpose of the program was to forecast plume behavior at least 24 hours in advance with little or no human intervention. In addition, the reduction of false alarms and improved accuracy over the Expert System in predicting closure events was a high priority since the reduction of roadway monitoring is a priority cost-reduction goal of IDOT.
The logic in the FORTRAN program (see Appendix D for a listing of the source code) is similar to the expert system in that individual probabilities for each factor are determined and then combined to achieve an overall probability of fog potential (P,, = P,, * P, * P,,). However, the FORTRAN version does not require human input for any of the atmospheric variables. Instead, the Model Output Statistics data set from a National Weather Service computer model is used to provide a forecast of surface conditions every 3 hours out to 60 hours after the initialization data are reported. Four categories were developed to provide guidelines for using the probability forecast. The categories used are a high, medium, low, and zero probability of reduced visibility along U.S. Highway 30. The categories are very conservative in nature, so during a forecast period with a category zero there is basically no chance of fog causing problems on U.S. Highway 30.
Observations from past road closures were used to determine the threshold for the high probability category. Using a probability of 80% as the threshold between medium and high probability, the model predicted a high category for the majority of the observed closures. When high probabilities are forecast, there is a very significant probability of low visibility along U.S. Highway 30 near the ADM plant that will require the roadway to be closed. The medium category is included to account for small errors in the forecast surface conditions and variations in the actual conditions at closure. If a medium category is forecast, users should be alert to possible reductions in visibility on the roadway. The low category indicates a small chance for visibility being reduced below ambient levels along U.S. Highway 30, so monitoring of the roadway would generally not be required.
During the winter 1994-1995, forecasts were generated every 12 hours (at approximately 1100 and 2300 LST) and sent electronically to IDOT personnel in Cedar Rapids and forecasters at Freese-Notis in Des Moines. The model produces plume forecasts out to 36 hours and lists the expected weather conditions as well as the fog category for each 3-hour forecast period (0000, 0300,0600, and 0900 for forecasts made at 2300 LST, or 1200,1500, 1800, and 2100 for forecasts generated at 1100 LST). Beginning 1 January 1995 the forecasts were sent via e-mail to IDOT in Cedar Rapids every 12 hours, except for sporadic events when the National Weather Service data were not received at ISU. An example of the forecast sent to D O T is shown in Appendix A. In addition to forecast plume behavior, the forecast form contains space for IDOT personnel to record observations of the plume's behavior. These are then returned to ISU to assist in verification of the model.
Evaluation of the FORTRAN model shows that during the fvst four forecast periods (out to 15 hours) the mean error is about 1°F for the temperature and about 0.5"F for the dewpoint. Wind direction and speed are also generally good during the short-term forecast periods. The trend is similar in the standard deviation of the errors for the early forecast periods, but longer forecast periods show a steady increase in standard deviation. A file containing the average bias for each variable at each period was produced to improve the accuracy of the forecast. This file is then used to remove some of the error occurring in each forecast. Initial results show this procedure works very well for correcting small errors in wind direction and wind speeds, temperature and dewpoint.
Results
The forecasts for the period from 1 January though 28 February 1995 were used to test the procedure. These forecasts were returned to ISU from IDOT personnel with plume and monitoring records for the period. The data include 85 separate forecasts sent to IDOT similar to the one shown in Appendix A. There were no road closures along U.S. Highway 30 during this period, but the road was monitored a total of 114 hours. Of these 114 hours, only 24 hours correspond to periods where the model was forecasting a high or medium fog probability, leaving 90 hours of monitoring when the model indicated there should be no problems with cooling-tower plumes affecting visibility along the road. During these 90 hours, only 6 hours corresponded to events missed by the Fortran program, periods where a low or zero category was forecast which also had an observed plume above the roadway at any elevation.
The model did tend to over predict fog problems during some forecast periods. However, when the plume forecasts are separated into 2 sections using the first four periods as a forecast of plume behavior and the remaining seven as an outlook, the model shows promise to significantly reduce the amount of monitoring required along U.S. Highway 30. This procedure could have cut the amount of monitoring from 114 hours to 60 hours, nearly a 50% reduction in the hours during only a single month. This assumes monitoring is required when a high or medium category is forecast which may not he the case. The decision to monitor the roadway is solely the responsibility of IDOT personnel in Cedar Rapids and the forecasts of plume behavior are only guidance to help in making those decisions. Different levels of probability could be used in determining when and if monitoring of the roadway should be started.
The forecasts produced by the FORTRAN program for the period from 1 January at 00 UTC through 28 February at 00 UTC were verified by IDOT and returned to ISU. During this period there were 21 hours when a plume was observed over the roadway, while 60 hours were predicted to have fog potential by the FORTRAN model, and over 100 hours were suggested by human forecasters to he problematic. Statistical analysis of data for all periods showed the model does have useful skill in forecasting plume behavior at the ADM facility. It should be noted that this only covers a very short period and overall results may he different if a larger data set were available. The threshold of 70% can be adjusted to fit the requirements of IDOT for the number of false alarms and missed forecasts they can tolerate.
The FORTRAN model has a very low false alarm rate while hit rates during the period approach 70%. Achieving the low false alarm rate was a goal of the project; with continued refinement during a full operational season, the hit rate should improve. The 70% hit rate is approximately equal to human forecasters, but the large reduction in false alarms means the model has better accuracy than the forecasters.
The FORTRAN code was delivered to D O T contract meteorologists at SSI In October 1995 and was used on a daily basis for preparing their forecasts of plume behavior at the Cedar Rapids site.
During the winter 1995-96 a test was conducted to see if the FORTRAN forecast system could be ported to another city. This test was stimulated by comments by the Highway Research Board that it might be useful to have a procedure that was portable and could be used at other sites. In response to a suggestion from Mr. Royce Fichtner of Marshalltown, we redeployed the system for application to Main Street in Marshalltown due to the IES plant to the south. The transformation of FORTRAN code for application to Marshalltown required interpolation of meteorological data, since Marshalltown has no local observation site comparable to that in Cedar Rapids. However, the Marshalltown site had no road-closure database comparable with the data from Cedar Rapids, so the system could only provide an advisory based on conditions leading to road closure in Cedar Rapids. The system issued forecasts for Marshalltown on 107 days during the 1995-96 winter period (November 1995 -April 1996). During this time, the system forecast high probability of fog on 2 days, medium probability on 8 days, low probability on 23 days and zero probability on 76 days. There were no reports of road closures during the winter period. From these results, we conclude that the system can easily be ported to another location, but that without local observations to correlate meteorological conditions to roadway closures, the system only can be expected to provide a fog advisory that conditions are conducive to fog.
Summary and Corzclusiorzs
All of the road-closure events on U.S. Highway 30 near the ADM plant are the result of pre- existing fog being enhanced by moisture sources at the ADM facility. The cooling towers and grain-dryer stacks do not, by themselves, produce plumes with enough horizontal extent or optical density to warrant closure of U.S. Highway 30. However, when the ambient air is moisture laden and visibility in the area is less than 1 mile, the added moisture from the plant can form ground fogs that reduce visibility to nearly zero.
The previous addition of new 'plume abatement' cooling towers and changes in winter operating procedures have improved conditions near the plant. The addition of more 'plume abatement' towers (combined with the removal of the existing crossflow towers) and the increase of the exit height of the grain-dryer stacks would further reduce the potential for hazardous visibility along U.S. Highway 30. Similarly, relocation of the existing cooling towers to locations well south from the roadway would also reduce the probability of those towers causing ground fog along the highway.
The fog problem is worse during the winter months since cold air can only hold relatively small amounts of water vapor, forcing the rest of form visible clouds. Most of the closure events occur during only the two coldest months and thentypically during periods that are unseasonably warm. However, events may occur whenever the required large-scale weather conditions are met. Therefore, an unseasonably cold event in the fall or spring may require the roadway to be closed.
To improve short and medium range (6 to 36 hours) forecasts of potential road closure events, an Expert System and Fortran program were developed. Both show good potential for reducing the amount of monitoring along the roadway while still maintaining a high degree of safety. The systems were developed using past closure events to produce rules and the probability of closure for each atmospheric variable. During initial testing, both systems predicted over 85% of the closure events and could have reduced roadway monitoring by over 50%.
A research paper summarizing the results and evaluation of the fog forecasting problem has been published by Takle and Thomson (1996). This paper evaluates the bias and threshold of forecast values for the fog problem and compares these with a previous study of the occurrence and forecasting of roadway and bridge frost (see Appendix B for copy of this paper).
References
Takle, E. S., and P. C. Thomson, 1996: Use of expert systems for roadway weather maintenance decisions. 1996 Semisesquicentennial Transportation Conference Proceedings. Center for Transportation Research and Education. Ames, IA. 183-187.
Thomson, P. C., 1995: Development of software to forecast periods of cooling-tower fog along U. S. Highway 30 in Cedar Rapids, Iowa, MS Thesis, Iowa State University. 112 pp.
APPENDIX E. EXAMPLE PLUME FORECAST
This file created on Tue Feb 21 09:58:46 CST 1995 Reading input file ngm.mos.00.22 Model bias file not found.
Forecast created using NGM MOS data from 21 FEB 95 taken at 1800 CST
DAY /FEB 22 /FEB 23 HOUR 00 0 3 0 6 09 12 15 18 21 00 03 06
TEMP 35. 36. 36. 39. 48. 51. 45. 39. 36. 31. 29. DWPT 29. 31: 32. 33. 36. 36. 35. 33. 29. 25. 22. WOIR 180. 216. 241. 276. 287. 285. 291. 309. 308. 315. 317. WSPD 8. 6. 4. 3. 5. 8. 8. 16. 24. 26. 25.
Plume Forecast for US Highway 30 near the ADM plant
PROB%53. 66. 66. 47. 27. 48. 27. 27. 29. 29. 29.. CAT 0 L L 0 0 0 0 0 0 0 0
Observations of plume behavior
ON ROAD - - - - - - - - - - - ABOVE ROAD
COOLING UNITS USED - - - - - - - - - - - The fog categories are based upon the predicted probability for a cooling tower fog effecting the visibility along Highway 30 near the Archer Daniels Midland plant.
The criteria for each category is:
High Probability Greater than or equal to 80
Medium Probability between 70 and 80
Low Probability between 60 and 70
Zero Probability less than 60
These categories provide only a general outlook of the cooling tower plume behavior based on a national computer model which forecasts weather conditions for the area. This forecast is experimental and users should be alert to possible large errors in these forecasts of plume behavior and weather conditions.
Version 0.94 26 JANUARY 1995
By: E.S. Takle and P.C. Thomson email: [email protected] [email protected]
Use of E
xpert System
s for Roadw
ay W
eather Maintenance D
ecisions
EU
GE
NE
S. TA
KL
E
AN
D P
AU
L f. TH
OM
SON
A" e
xp
n g5.m
i. a <.m
prer-hswd
Ual thn, .,e
rr. 1 n
aic
1 or
human .ip
n
rruo
nin
p *iU
i m
rvDolC
Ed tnoulrdga b
aw
Ud
<.",bin..
Ulew
lo .<ach M,m
<o
"c,"*ion
u.
hu
mln
srpcn
lo
comp~.r
mbhm
. ~c
hrrr d
cw
w
ly.sm
s rw
we
lpm
fic u%b re
latin
g lo
mndrn
y wealher drci%
;on matin
s. The
rinc
i*ily
l~m
ro
rfmu
.ting
h
o~
ro
~r
n~
ai~
~o
n
btid
gesm
dm
rd.
u.y,inrcninl ~
~w
~m
d~
le
sr
rn
ix
li
~a
~~
~t
rm
ro
rrom
ao
ng
rog
on U
S H
ighwar 30 in C
Idu
Rlvid
l.lo-a
. duo to plum
es cMtrcd
sr -,ins
,oven
., . con
-rrnm
cn
poductio
" p,.m
adjacror to U
le road*.#> F
mm
ha
w rlp
rlen
su
wa h
lrr Co
ne
luW
Uut
cip
cn
3p
mrn
s cam
be usrfu
l in re
ndw
ry we.&
" m
lintm
arr
drtirionr Thcx.xpn.nrc,
.is h.re
IUoue&
", (om
n,i*r M
mom g.nrra~
issu
e
or dmi.ion
malin
g
lo UI~
use d
u
nM
r infonution.
or .mp
n<
un
*r
UII b
ed*
an* rnrlW
.~ which a
cmnpnrnd
thL rarrcu
t ih.
dcw
.pin
i arnpErirurr
co d
nennin
c M
likrlihoad o
llmrl
tnp
a I.
th. q
tkn
~a
nril" or U
lc ihm
dsla itrm,
and wre
n
rarc..x rariiirblcs lin
ed
in ~
ab
k
I. ~h
c
iys=rn
war mn r
hu
t
Il:m a.m
. LST lo
tom art condition^ zt %
pp
axin
sely I:m
a
m.
ihr rollo
win
l manrim
$. 1" M
opntian.1 ,r,,ing.
M in
pet ru
i- w
sn
supplied by U
ie Ce
nn
n nnc.m
lagi$u u
6o also n
n
Ure
rpm
*y*w
m.
ncrcrific.iionmiaiior~ab~r
2 .um
muiz.3
prrorm
rna
of ihc system ar m
uu
d aglinlc sccual o
wco
mu. T
he condi60nr a, rcon ro
nrr uui 11) m
e ~
~rn
rr tcm
pramrr
mu
, h
brlou(ircdng,~2lUlslurf~ce~mpcra~mu~lhbcbvih~*W~
lk,, hi" w
cqu
4l,oU
lrdcrr.io
i"ItemP
"w~
, H
lr*c"ec. Uiilcn
. te
dm
n br b
hq
d do t
r~
in
c
Uir in
tl~n
ao
n
Uit h
i<n
s m
d
r.ir.dum
Rlfr U
XL
ibm
w ih( O
~P
~IY
T~
chm
,hold b
r IT
we~..ylngU
lrrrm.Lw
illrorm..m
r.k~i.Uicm
.io..*c.~ surfarc u
mp
nm
ar ,omr
rrrcrrnce
la-&
on i. IT
higher b
an U
le dru
-pin
t Um
pn
m. The pD
1 in F
igm
e I sh
ow3 h
w
~h
~g
"g
m
e hs
h~
~c
hm
au
~
lcU
tmd
h
lw.*
h ria
. Tbr sy3lsm
wu
duig
md, u
llrd, m
d nd
plo
yrd opcr&
tion.IIF
I. pm
ctire
. ~
ir
ra
mutm
npliu
) ~
l~.~
.mm
xrcn
l tim
r, *i~
i din
orra
tcom
bin
sti~n.
pu
vrr~n
in
~
ab
ir I ro
,sro
m~
rorrro
ais
rrcd
rd~
oU
la~
~lo
u~
dp
an
mc
nio
r~~
~-
RIlrrroE~Uimmpm~og~our~.i~~~~n~ann~n~ricininor p
~t
ia
n
(lor. D
OT
) CM
h
&n
pcnann.1. e
qu
ipm
zn~
md
ma=
- ndlua i,~h,~uUic,olrtion,n..dingB,,m,i~n.Und.r,~n~nhip
US
Hi
8h
~.
~~
~i
~C
~~
R~
~i
d~
.1
o~
~.
~u
=~
~m
~~
~~
~~
~~
~
.*a, m
i.** tia
m tin
rum
~ho
ni..l.d
"nio
olin
g
lo- .fe
e lo
r* D
OT
.w=de"=
l%d
rn o
rpcn
~Ys=m
10 rom-r
en* = ac_.,_e_plrntd,
d).ram
,,o Ih.m.du.y.Am
bier .-
I% houri in
adyrm
(1.2). ~
p~
~tic
~o
~itio
n~
ol~
i~d
d~
d,d
iidd
iitilili,~m
pc
~tw
e.~
v-
k e
ipn
sy.!m
n bnckrsrd.ch*;"in% w
vm
and um
pnwc,
md ,de m
.blurr try
mnditio
n,
liru o
f 32 pv
um
lrn m
d ru
iab
ks
and I1 mlu
. Roiidvss m
d
~urmj,,r,+
hcrherMN
u,iing rim
p,umcri,llc.d,ola
and%
* fm.1
dam fm
m W
mbr. Im
uv
l. md
Feb
mT
O
r 'ou
r r;.ib
i~h
rwm
atorumon ~~
gh
w.
~.
m.
~~
~g
~~
u~
io
n.
by
lo
r. m
in
M oVtnr
r0g inr~
ud
r m
c ia
city
Ef.T.Ll~.XlOlcm~ytI.il.lor~SumUnir~nAm..la.KOII,
d""ngUi'cpi'O
dc' L
C h
mh
arrr .d
vcncb. I- *.id
PM
~.
~.
.a
B,,, sw
l A
ccur.rr ro
nru
u or o
nw
t md
armin.tian
or lor vilibilily
X.~
~C
il%M
ilwtiW
l(1.
roixlitio
~sn
rr med
d coru,islox.
UO
Tp
no
nn
rl in M
ir m&.
'2JW S"!"".C.
li",d.P.I uu!,.>U
!L"!l"l!.Uln.
r! .!yl J! ,p!,q
>rou
."'!*l=d"~ .>u.unu!.m
.yl
..#pup . uo ""0
, (,!a ,."A1
18w
=?
us
w,w
~s
d 0
~.
r! w
ay, rev r.ru
ol
n sqncq 01 w
m
.mw= r! ,rd
uio
lon
w r 'aidunx,
ioj .~
~!t!q
!ruo
d.~
JO
uog!.!p ~
oa
-!w
~ln
~ia
irn,=or
riwu
op
xa
q u*r!,sp
fig,& su
!~m
ru r
JO
l*wd
o#
l*~P
>w
pu
.>'~
>a
oJ
lqln
~
p$
o#
!aio
)ob
>r!t>
ps
yl
. .A,'LQ
Y~
~ .B"!""PW
.rn*u,"!
.noq m
oy
-mw
I! 'OJ
-wd
m
,, pup -.l W
!! ,r rw
p a!-
0,X
ro 'lo1 'um
!paw '"a!*
j0 b
da
r, r$!j!9
a~
d
soj m
ao
j pu
!qu
lo>
r ww
=u>8 war*.
r. lo
,"*"~Y
L.~
.lr~s
~d
l..wm
~P
q
XU
.t.vs"!*.qww
" 9
%~
zm
luo
j 0, m
du! %s p)in pur a!' ip
idcn
npa3 U
n ox p*%
El
d.,"!
man
m.p
"W.u
no
q
7.i b"' ""0 ."!O
"ol~ll*lV
Pm
!!eN W
I~~
SO
lV)
~?
~~
S1
Dd
~"
O!
~~
lnO
Nl!
4"
W
P90
pn
n~
w
wla
rl lnw
ii~p
sr!$
4rw
!hd
vriqnm
* pl!S
olm
nx
u
ha )V
I JO .q
c*
mrrio
j rujn
bs w
ae
ir nu
, ey
l 'O~
,=~
OJ
8o
~$
u~
~~
!~
~p
pr
nB
u!
~~
~J
pa
~>
ul
P~
Ap
~~
OW
~J
O>
~~
~O
-1
"s~ p
mu
t puxl "VI
Pv
wp
uo!h8,!uww
o,pur Q
!l!qsln
*=
mspu! ~
YI
Y~
AO
~I
.
~n
1(
1t
>~
m
pn
! mnen i=
w~
wy
l,~rlo
r..,pssur*-uo
so, m
mo
J 0, rm
prm
d 10 "~
nls
n'=
A"!
U- pn
nn
b=
r ma
er
o~
am
'ss,usu
n>
a so
j 101 Am
pro
, ow
s
up
!uo
u
un
q lo
lqw
nv
sw s
np
u a
plu
=e su
!,!~
.om"
,0""p
>x,p
x,*ld
~'M
, Y
, 'mp
ms
su
!,u%
U
rld xn "0
luq
usd
ap
A!yl!y
%! I~
aW
ol
m uo 801 jo
uulu
mu w
-o
~.*olL
za"!p",,"!r"~!,.*u
rqo
)r,u"~q
~$i'zP~"!O
"o~q
~!Ya
.MI V
rn*
usn
orw
68.5, U
qUq*
wo
e n
cp
luU"!um
-P.ol
IM ."a,
per nip
p,!sc,olonxu 10 xn
iq px
lo,=
*v nr w
a
.r~sw~tnos~~unroj~px~wmun~~~aruur~r-sqr~~~.nur
JO u
opsu
!quo>
log
am rr=
!q>r m
rpq
,su iou>pun-!r!r,
pun n
nb
!uy
urr!r~
yu
pa
.u!u
lw!p
jo~
n~
qp
do
~a
r~p
~~
isn!!!qe
.qd
pm
np q
-u
ns
ep
psol e 1a8Y9
saw
nld
1>
rtol-8
~!
-,~~
w>
.wlr!,!q
.q~
scs;uas.U
nu
l"!uu
~n
po
l-wa.y,yl!r
~Y
~~
O~
PU
S~
O~
~J
L~
!~
!~
~~
~~
A~
P=
!I
~P
IO
YI
~
ws
!sm
!m rm
.uo"rnp
pu!n
pu..p
dr
pu
w .w
ua
ius
lu!o
d-n
ap
'w=
d
~WJ~'~"l~r~~OJ~JA!PE~!p"!~1lJL>!~~94dSVI~~!~
.m+
"wu
rr r!a
ulp
.qw
r.pits
no
!*d
u~
jg
cx
l~
~~
w
u! p
xn
'. ,T
isour iun
r sw B
u!m ~
o~
~~
~p
an
y
=!ri%
ud
ra
uy IC
>""II9
"U
n
. - y
o!iln
D~
UPIV
JYI~
S h!I!y+
Cw
ei lo
-I*"!
- , d (,m
. .9>."1 r.0-1 dl>
- 8 - "
u~
,~Y
D!I'>
P,
I" - 1, ,A
- L. - ,9 - I="*
10 lap"! 13. ,lid
- % - "!m
'">
"!'!*a ,,,(F
q - * - 'B
'bU,
..!,.,.I o,.lw
rs ,, * .I,. - 3
- *a".>'"
" '*'3
<P r yll%
- W - "" '!W
0
.>p
- J - -'" w
. or,., LF
$>I@ - H - ".'l!il
P +
. q +
I - h
l- "
YW
IOiq
Yn
*
,Pa
l
"A Oh'
"A W
PU
XY
U
""tw-
P".'"I.'V.-#"~"9,, ,.""I>
muuouloq ~9
oi wa!a
pw
m
o~
l uuw
lp
PY" P~I
p?pds P~
X
38
~9
1~
1
uu
~A
J
nl,*o
uW
o, ny 9 a
,71001 y
id I Y
O1
00!1.l!d!ald
*W
-OI
KV
9 >. wn>~
=d
ius
$
~m
od.-*a
S
u!u
ao n
ou
oo
lol o
mu
sdu
nl w
nY
!mH
I.W
, ~
n).>
drn
~, ro
nD
ljnn
. .
equbpmnl. rn
rl mnteienili7 b
ut rlro
ihc pole
nlilt lo
r an ircidcn
x and p
o*lib
le litip
sion
nlu
lllnp
Durn n
ot u
lyng
rction
.Aa
ad
cos o
f U
ie firs* itow
lor in
dirid
ua
lrrtnll but large in rg
grcp
aw
l mnrr b
w
cishcd sgsinst ihe mre
riai co
scof~
io srcond rp
osib
iy wyiw
$1 iiK
<ur,).T
"~
n,~
nm
an
ce
,"prrirm
cm
no
,nk
ihc ro,ceu,cr lo
girr 2
"pr" or 'no'.
on Itoat. kcsu
se ihir w
ould
foica the m
creorologi%t lo m
lXc I WiiE
y dccirio
n b
ud
an =me
icv.1 a1
"r*. w
hich
i* *c
mrin,cnrncr .upni.ai',
,~m
on,ibilil) Rlih.,.
ihs ~
nrrculo~ho~~di~ruerpmorage~hm~.~od~lc~uprrao~
m"llu
ub
lirhr~
,ho
ld.a
drc
irinn
d~
"on
. hyo
nd
wh
ichfm
a.
9Y
mn
Sio
n actio
n is u
km
. ~
me
vio
d~
1~
i~n
~id
ci~
cd
on
mr
er
l~s
un
~~
.~.6
,n~
i~o
~~
u~
to
ev
dv
alr ih
c pm
babiiilics o
l a "hi<'
"miss:'
"lalse alum." or
't-tno
no
crum
ncd
and
their ~
l3tilio
nd
ip. 3cpam
lciy. lo lo*.
crrtscium
ry tlhe
rnio
".ihililyo
lU.
l"r.cllOilandd.ceionc"-
cnon ,ha re.punribili,y
or*
n,u
"lmm
cc ,,,pni.o,1. ,nc,crv.
in8 ihr h
it rrlr aim
incrru
cs uls laire
-aim
rrs. BY u
.ieol %
;snal d
clrrtion
iheaiy, ihr mlin
cmm
c. rup
rrisor cm
b.lrn
CC
hi, ,ate
rgY
Nl irire
.rlvm
rr,r. in
dcp
nd
mliy ollom
ra,,*rc",asy F
fom
SOT, an inlc
r ofa
rc"ncy,
6. ir ,he
number o
llun
da
d
dcvillio
n,
Imu
.lin~
UI. m
ean, orb
e ,l,,"rn.d
nom.l,dilvih"-
. .
tion
, a1 doci.io"
"uirh
lc, pm
ceding occun.nce and prercding
no
n~
cum
nce. Thus. il d' =
0. *err
ir no rLii, be
crux ih
r pro&
lhiliil o
f hi, md
ralw
re""" u
r equal. A w
cnnd indc.. 8
. is Be
iike
liho
od
i.liD
,ha, th
e gilt" d.,.
,up*e,, O
CC
",,."EC
c,".,
no
n~c"".n
cc. lk cnle""
plrcemL
", 3% con,ili.x* ""b
ievd
if
B = 0. b
is~
d
loru
d m
aincauning n low
hitc
-alu
m m
e (at Ihc
Fo
rcc
as
l Hit R
ae F
orc
ca
sl False A
larm Rsu
. . . . . . . . . . . . . . O
ullo
okH
it Rale
OulIookF
alse Ala
m Rat:
0 ', -
. .. .- - ...
54% 60%
66% 72%
78% 84%
90%
Probability Threshold
T&
Dd
lh
m
~rpnrea~~lovrrhilrse)BB>
!.and birsrd
iow
ud
mria
ain
ing
r h
ilhh
ilrlieCIih
ocip
u.o
lrhig
h.rfd
.e-ll.m"tr)ilB
<
1. A
lh1.d p
mm
na
r. A. ir ,he rie
aunder b
eo
we
orih
cs
m X
I*. ti"$
oa
rsin
g chlra<
tl",tic, cu
wr (I) and ca" be in
triprrrd
u
Urp
mn
ug
.ol,in
rVlrrU
lr*~,ern
.u,d
i.tinp
ui,h
bc,w
~n
con
. .aldonl lLlding
la oerurr.nr.
rrom eondilianl
~~
~d
i~
~
to n
ollo
reu
mn
rrof lo
go
r lmlt.S
~W
(71E
onsid
rr,r ~
).~trm
wilh
A
rrrlucs be
lor 70 to h
sre in
sufid
en
t rcc
ui~
y
lor m
uch pn
ctirnl
raiuc md
iy%m
ms ric
h rd
ur3 bem
een IOm
d W lo
be
vw
fvl tor
'"mc
"W
plL
,.
Tnbio 3 g
ives Uite
ltime
M rrlw
l old
'. 8. &
A
lor U
lr rip
s
,ysrom
tor bridge and rn.dw.7
lma
md
la Ulc F
avrn
pro
grrvo
"!ad for fa
g lo
rrcutin
g.A
I1 I,l*rn
, ,how
,MI n
di~
min
.ting
a
cu
mn
ru lro
m n
onacvm
nel (6
> 01,
md, la
ihc dccilio
n
mire
riauud.slliysu
m,sn
b
itwd
tow
~m
aln
lain
ing
aIo
~M
w.
alu
m nlc it ih
r eip
nm
ofalovrr hiln
n. m
"aiw
lof A
*"g-
g~
~v
iuirll~
y~
rm~
erh
ibi~
~lil1
ing
rrdic
ting
~irm
~p
cti~
~ro
~d
. vlyronditionr.FaUrb~~po~rw~obmiddnosndsndmpird
m.iogou,%
mcircir$
br b
um
an
~rrc
uu
rsIo
m
inT
zbio
3.lb
se
rriulll show
!hat !he hum
an lom
.lteo
*err
la11 biu
cd o
wu
d
m.~
n~dnin
8.
lor rdx-.sm
"tc.
.nd ihlt
0. SW
I cornpa.
or riigh
lly lovcr b
n ihe
rxprn
3pum..W
e rmv
hu
iaih
si
mi, m
mp
u>
.an
I' not .m
rll, .d,d
brrn
iv U
icrrp
n r,,irm
,,
c.2iralrd
.I iu
pv
no
u *st
ar.w
r
c "ale .9."rn<d
'prh
rl rorrcacu- lo! me ,n
oit ,m
ao!e, ila
blr illo
vlr #,,u
rn
Fur vr
fog pu
blern
. h hhrm
ln lom
au
n h
ld a h
rl r8v
01 10 mn
ro
~
RE
CO
MM
FND
AT
ION
S FO
R U
SE
OF
EX
PE
RT
SY
ST
FM
S
&r
cip
ticn
ie in
dore
logng ih
rx syllcm
t and ouiobwrrm
dons aro
lher ~
rpn,y*,.m,
Ulr, h.rr
be." d
crd
op
d m
pm
*d<
mom
gcnrnln,nrom
,ogic* l-u
uh
uu
ug
hi".
Ih.l,".h."m,y,.
sm, ur m
ore likly
to b
,"Ccc..r*, i( VI.7
srrd..i*".d lo rr-
<=a *r
pfir
r*eaC
.g..h
osro
n.
btidgr, rslher ihu, morr gen.
era
cond1tionsR.g.-rnnc.011111
weaU
rr).%hi.
is bcs-
,he number o
< N
IL, needrd io
diro"mina<
c K
cu
mn
cc
1mm
n
on
oa
vm
m ir llirh
llmilrd
IlBv
l M lo
r ouriy$
mm
sl. Venfi-
cation slsimp
lcr 3yslm
l il m
vrh rn
on m
an8g<
*lc last.
AP
PE
ND
IX C
. E
XI'
ER
T S
YST
Ehl
SO
UR
CE
CO
DE
The
following is t
he sou
rce listing for
rhe
expe
n sy
scer
n delivered to [
DO
T and Freere-Notis
pers
onne
l during
Dec
embe
r 19
93.
DO
NI\
IN :: C
OOVTWR3
ROOT P
WE
:: COOLTWR
Global KB
da
ta
--=-..ss----==-s=--.s-
CW
E STRUCTURE ::
ea
inm
ec
er groups :: ICOOLTWR-PLRMSI
groups : :
IVINDVEL-IULES DEWPOINT-RULES WINODIR-RULES
TC
XP
-RU
L~
COOLTWR-RULES HETI\-RULES
I H
um
ber of rules ::
46
Nu
mb
*< Of ineEa-rule,
:: 1
hi
:: iDOt4AINI
TEXTRGS : :
11
Frame :: COOLTWR
IDENTIFIER :: COOLTWR3
TRIWSLRTION :: la 3y3cem
Lo
predict
Che
foima2ion of cooling
tow
er
foq an
HU
Y 30 in c
ed
ar ~ap~dn,
IA baled on sLatisrlsal data.
I w
as
::
IFOGYES C~RRECTEDI
PROHPTEVER :
: l~his is a
PROTOTYPE ~xpecc sysrern to deeernune the
forma. cidn bf fog covcrinq ~uy
30 in C
edar ~apid, near the
arch
& oanie~s "idland
I rnow I
corn sweeteners p
lant. =he
Syscem
w6s
developed by E. 5. Takle and P. C. Thomen. The
developers are not responsible for a
ccidents, damaqc;or
1n1u
cy r
esulting from the use at Lhia sysrern. =his System
was
developed for forecarring fog nround the WH pl~nr in
Cedar Rapid.,
Iowa and may
no
t Ds accurate
ac
ocher
locacronl. :line 2 '*.
Thls
Sy
ste
m i
s s
~pyrighred and
shall not be copicd
or used In a
ny
form
without vriiten
perrmasron from E.
5. ?axle '-.
:line 2 [if you need help
at
any prompnpr,
press :ATTR (YELLOW BLINK1 PI :fiTTR IWHITEl
key! :line 2
Pioro=ype Y
ers
lon
3.0s.
14 oc
r 1333
I D
TS
PII
IIR
ES
IIL
TS
:: iTEnP DEWPOINT UlNDDlR YTNDVEL C
ORRECTED TOGYES1
RULCGROVPS ::
ICOOLTWR-RULES TEMP-RULES IIINODIR-RULES DEWPOINT-RULES
WINDVEL-RULES
I COOLTWR-PLRNS
:: ICORR CORRECTED DEWPOINT POGPROB FOGYES PB
-DEW
POII
IT
PB-TEM
P PB
-WIN
D PB-WINOIIIR
EB-WINOVEL TENP TEST
YILIDDIR WlNDVEL
I COOLTYR-RULES
:: IRULE026 RULE031 RULE038 RULE039 RULE040 RULE042
RULE043 RU
LE04
'I R
ULEO<S RULEOeB
I TEMP-RULES
:: IRULEOlS RULED19 RULE020 RULE021 RULE022 RULE023 RULE024
n,r
,eA
,q
, ---
- ,
WINDDIR-RULES
:: IRVLEOOP RULE010 RULE011 RULE012 RULE013 RULEOX*
RULE015 RULE016 RULE017 RULB036
I DEVPOlNT-RULES
:: IRVLEOOI RULE002 RULE003 RVLPO04 RULE005 RULE006
RULE001
RU
LE
008
I WINDVEL-RULES
:: (RULE029 RULE030 R
ULE031 RULE032 RVLS033 RULEO3e
RULE035 I
.-. .~
--
-=
TRIWSL~TION :: lcorreciion faccorl
TYPE :: S
INGLEVaLVED
EXPECT :: MIMBER
UPDATfD-B1 :: (RULE031 RULT039 RULE040 RULE0381
USED-BY :: (RULE045 RULE0461
CONTfiINED-IN :: tRVLEO46I
WG
E :: I0 4.1
CORRECTED
-==-
=.**-
TWSLRTION :: ICOcleCLed prDbahillty of fog1
TYPE i: SINGLEVIILUED
EXPECT :: POSITIVE-NUHBER
UPDATCD-BI :: lRVLEO45 RULE0461
USED-8Y :: IRULEOO R
ULE0421
WG
E :: I0 2.1
DE
VP
DL
M
..==..-* TWWSL~TION :: ithe forecast devpolnr
at
~u
y
301
eRoneT :: iznter me formcarr daupolnt
at
~u
y
30 in cedar ~apld.1
TYPE :: SINGLEVULUED
EXPECT :: N
UWER
USED-BY :: IRULEOOI RULE002 RULE003 R
ULE004 RULE005 RULE006 RVLE007
RV
LE
OO
B i
~.
HELP :: (Enter the d
ewolnf forecast in
degree* 1-
20
co lor)
.I
CONTfiINED-IN :: ISREFWK RULEOI4I
WC
E :
: 1-20 701
TWSLRTION ::
IProbdbllLiy of foq on
Hu
y 30 lo.
Lo 1.
))
TYPE :: SINGLEYnLUED
UP
OIIIE
O-B
Y
:: ISREFWUIK RULE0261
USED-BY
:: ISREFWUIK RULE0261
USED-81-THE-WX?
:: INRULEOOlI
CO
NT
IIINE
D-IN
:: (RULE037 RULE039 RULE040 RULE0381
TO
GY
ES
=
-==
=-
TWSLATiON :: ~~here
will be a cooling
rov
er
fog on nry 30
ne
ar
the
m
~ plant
v~
rh
the focecasc c
onditions.
1 TYPE ::
YE
SIN
0
UPDhTED-BY
:: ,RULE043 RULE0421
USED-BY-IHE-W&Y
:: lHRULE001)
- -=-====ss-=
TwIU\T~~~
:: (the p
rob.+bi~z~~
for fog d
ev
La
a qiven derpaincl
TY
PE
:: SINGLEVALUED
EXPECT :: NUMBER
"PORTED-BY
:: IRULEOOI RULE002 RULE003 RULE000 RULE005 RULE006 RULE007
RU
LE
00
8
1 CONTRINED-IN ::
IS
RE
FW
K RULE0261
WC
E :: 10
1.1
PB-TEMP
-*=-=
==
TWSU\TION
:: (<he probability of fag due to the temp.)
TYPE :: S
INGLEYRLUEO
EXPECT : : N
UM
BE
R
up
om
En
-8y
::
inL!LEO18 RULE019 RULE021 RULE022 RL~LEOZ~
RULE024 RULE025
RULE020
> CONTIII>IED-Ill : :
ISREWK RULE0261
WG
E :: 10 1.1
PB WINDDXR
- -==,,--==== T~SLATION
:: [pi~bablli~y
of f
og due
La
th
e vind direccionl
TYPE :: SINGLEYRLUED
EXPECT ::
NUMBER
VPDI\TED-BY :: lRULEO1O RULE011 RULE012 RULE013 RULE014 RULE015 RULE016
RULE017 RULE036 RULE009
I CONTAINED-I>! :: ISREPWK RULE0261
WG
E :: I0
1.1
78
WINDYEL
- -=
==
-==
==
- TRRNSVVTIOS ::
lprab of f
ag due to vrnd velocity1
TIDE :: SINGLEYaLUED
...-
EXPECT :: NUMBER
IIPDIITEII-81 :: IRULE029 R
ULEOlO &"LEO31
RULE032 RULE033 RULE034 RULE0351
WG
E :: I0
1.1 CONTAINED-IN :: ISRETWUIX RULE0261
92
TWSLATION ::
ithe
foreca3i a
ir temperature
ar
Huy 301
PR
ON
PT
:: l~n~er
the forecast air temperature
ac
nuy 30 in Cedac ~apids
TYPE :: SINGLEVALUED
EXPECT :: N
UMBER
VS
ED
-BY
:: IRVLEOlO RULE019 RULE021 RULE022 RVLEO23 RULE024 RULE025
RVLE020
1 HELP :: lEntei the foicca=i airiemp in degieca 1-30 to 70rt
.1 CONTAINED-IN ::
ISRE-K RULE0441
WG
E :: 1-30
70)
TE
ST
=
=-=
. . -. . . . . .
zm..-=m
=
T-SU
\TIO
N
:: ctne farecaa~ vind direction
at
~w
y
301
PROMPT ::
IEnLer the
fore
cs
ac
vind direction at Huy 301
TYPE : : SINGLEVALVED
EXPECT : : M
BER
USED-8K :: (RULE010 RULE011 RULE012 RULE013 RULE014 RULEOIS RULE016
RULE017 RULE036 RULE009
I HELP ::
(Enter th
e e
xpected rind direction
in
degrees 1120
Lo
2801
.I W
GE
:: 1120 2801
WINDVEL
=--=
==
=
TWSLATION
:: lene
toresa
sr vind velocity
at
nuy 301
PROMPT :: IEnrer the forecast vlnd speed iknoCs1 Huy 301
TYPE :: SINGLEVALUED
EXPECT :: I
NTEGER
USED-BY :: ~IRULEO~?
RULE030 RULE031 RULE032 RULE033 RULE034 RULE0351
XELP :: [Enter the expected wind speed in knots I0
to 4OkCS)
.I RANGE ::
10 401
RULE026
*=-=
--=
SUBJECT ::
CO
OL
WR
-RU
LE
S
If Probabiliry of foq an Hvy 30 0. to 1.
Is not known.
Then it is definite IlOOIl that Probsbiliry
oC fop on Huy 30 0
, to 1.
is
11 m
lnu
. 11111 rn
~n
u* I
the probsblli~y of fog due
to th
e temp. divided by 1.511
Limes 11
minus Ithe probablllry
for
fog dew re
s given derpoint divided by
1.0 Ill
rimes 11 m
rnv
a lprababillry of foq due
to The vlnd direction d
ivided
by 2.
111 fines I1
minus Ipcab
ot
fog due to wind vetoclcy divided by
1.71111.
> " A c < 0 0 0 .* . 4
: fL m 4
m u m 0 w v m * . k
m a c a" m 0
I O U m L.
O D " m e c " m .xu* .0 z e " 0 s " " " e 0 0" C "
C a * v 1 V
5 U
rn m r " u
.4 0 " e U "
w u m Y) a m , " " w , 5 E
w * a u U C : ?".2 _1 " Y
u > - 3 m 8 -0
e - G corn , m m = .4 0
F - Z E I A i r n - 3
F - " : o n " z JN 0 -
0 - " U 0 - u
u U " " 8 - " " .. ad ., * " .. .. V .." "
..un.,s .. ., .. c .*
a"", W 0" e e l 0 " O Y U E e s g : ,,,,,
- I ? U P P U - I I W " . D l m m " """mu- 4 l n O . u E 2 " 0 "
s Y = I
* " S 2 : ': .. - 5 C
" 0 v r . u G w
N O r U Li
u 2: Y " Q
.I * m Y n o I Y Y rn " n n o 4 4
A 0 0 " m 3"
4
$ $2 0 ".r: # m u " " a u 2 2 5 " " " m * " " U V " * 2 25
M U U W * u - 4 s
2 .5 ,: g , m e. 'S E" u W n n Y *.I n u
V " 4 .." Y C .. I LI .* " 8 0 :
z j g : V " " 0 . m u 253 Y . 3 4 , Y I - _ " 3 1 " N d u ,
c . . " W "I I : . - N C Q
?.,* z ..
n n 3
' C " i z " 7 Y I I 1 P
0 c . m C mno "I.'. D N
* " I * *:a .." * * .. 2 w .. * " C -1
D I " r 2 - " rn
C , o - 3 o l l c - 3 P
- 2 : " " e :, " G " " c, 5 " * " m a : 1IT
c u , a C C C " m * a 0 " I. 0 .." 0 " Y) * a" c 9 0 m
" s 0
" " 5 0
0 II
" w $ ' - P " . o
.a - %; P'" m 2 u a 2 . , N r t C "- - m v c .
C I W * " W " m * o U 5 1 5 G l N o o n n n u
PI. ..2 n o 0 r m ".. 3 ; ;.. n * n.? * " Y " - n " l
..a m , a,. ,.P a n " C e
- s :GI f
" " " w e
0 " ,- r i E P
rn " " 0 LI a " m
m * z v I ' T T C C w u 0 - a T
P 3. O Y " .. r Q .." "
o u e 4" * a n n c 5 " ., * , " " 5 0 - w .. 2 ". 5 a
a * w r . 5 C.. m n . .a 2 " m _ l C N .. * C c- - m * v
c t m n m s o
Y 5 , 5irYN " a "".N
0 2.. m.3 I. 0 ,, 2 . . 3 " - 0 t ' ' n w c.4 O Y U r n - n "l v w v * 3 0.. * m on n c P -0 ""
m a m " 5 4 3 7P P e n D " " M * m " " " 5 s L: 0 " " m D L1 " W * 0 " 0
P, = r z z " T T ,- * w W
" a a C
0 - Y .. ,- Y) .." " o m D
4 Y II
Q C n c 5 " * w 2 " " 5 O W - *
m 2 ". 5 :
U m C I I C - o ' " , , c -
O C * " c m m ~ o 7".IT,ir "I" I W ,. 81 0 " " .. @ .. .. O .. * z u-"" " N Z m 0 z - z
i ? a c ? c. " rn " m 2
4 * - 5 2- 2 .. w
, P ".. z 0 w : s
0 c . n r n n m m . 0 "5 Z C 0 L . W
D r.TD.0 %, "%: 2: Em.. 1 * " .. .. :S $ " = o * u - I - " " N Z
Ln 0 - 3 - 2 0" O 7 e n 2 : - "5 5 S 5 .. - 0 : z " 0 0 * r P " " n r r V I I * 0 "
n l 1:
?4 s .nu u
0 0
2- " 0 " v C r Y1 0 s "
Y 0 P " 1 r X " F " s r 4 . 6 e 0
r 0 .. 0
u
0 ..
t 2 :: $ 5
! 2 :z25'~%",,. " O C l n r n n m m . 0 r.*55m'.(I0 U 01.*n,m Ot""%: . " O D * , , W * " " YZ I * ( . . . . . Y . . " " . . . " W * O
" U U - F I ""*I - m m
wan-: O D "
z4.a 7 - 0 0
$-* II " 3 1 C 9"" r m C . W b, " " ? Z Z P rr
" " W 0 00
4 .. *. L T U " r r r Y 1 W " " " U .
' " * , " " 0 : 4" 3 " " ir
0 a * * p ;l t : n r? 0 C
* r
i * * $ 2-
111 r.2 r . C n - -OM",.
O C a m i ? s:i?z:: 0
D " * " " N
n n ? . % S n o 0 " r " 7 . a . . : . .. 0 " a 2 " . 0 84 - " : z z gs s " $ 0 2 *a X 5' -
< < z " 0 7- z g * 0 " 8 " - w m n n 'tr 'E
O i ( l II " " ?.= = F z a" r
4 " W 0. 0 0
c r + m a
" W 4 D O n
U n S Y , * * $ " n < a C
C.. o w : 0 -
0
Y" " e ?
% 0 r
APP
EN
DIX
D.
FOR
TR
AN
SO
UIl
CE
CO
DE
1 hc followtng is
!he F
onra
n S
ourL
C c
ode lor !he lllcrl (J~nu3ry 1995) r
eri
lon
ofthe model u
sed
IJ forecsrl caoltng-tower pidine bellarlor i
n C
cdar
Rxptdr
1 hl
r vc
rrlo
n jrcluder ine abtlny la
curreel obrened model bmr when aua8lable
INTEGER I,J,TSTISI,TEUPI5,191,DWPT1S.191,DIR15,19~,YELlS.191,
+HN
, 00, X
Y. HH,DOl,DD2,m12,m13
c
V-.90
original
Be
ca
version. 16 ~
ec
en
h=
. 94
-Us
ed
szmple average. of acH variables
c
c
V-.91
19 Decembci 9~
E -added
a c
res
sm
n scheme on W
H variables
5
-added BRL
a3
4th
da
ta $Ate
c
5
Vn.92
20 December 94
c
-breaX wind into componcnrs
c
-so
rro
sx
ed
122 oupour pcoblems
c -corrected end of m
anrh
problams in ourpvc
c
-fixed preblcmwhen no
dsra found
C
c W.93
30 Dectmbe.
94
G
-added
are
a for W
T people
to verlty lorecaats
c -changed outpur tomar slightly - (
PBTOG-~OO~
c -fixed d
ata
<c
ad
problem ot large negacivc tempss
E (this W
as also added
Co
0.921
c -fixed d
ac
e
ourp
ur
problem (added to 0.921
C
c
V-.94
26 January 95
c
-added model blaa
co
rr
cc
~ ourput
C
E ....
........
........
........
....
........
......
c read
tex
t tile tor
cu
cra
nc
nqm tile
an
d time
........
........
........
........
........
......
OPEN (UNIT-99, FILE-'.name. tilo', STATUS-'QLD.I
REUD 199.101
INPUTrlLC
OP
CN
(UNIT-98, FILE-'.Lime.file',
BTRTUSr'OLD'I
RCUD 198,101
CURRENT
"zed
Lo
select a
specific ngn tiie
..........
...........
c
INPUT?ILEP'ngm.rno1.00.02'
........
........
........
........
........
........
........
. PClnL'.
'This file c
rea
ted
on
.,CURRENT
PrinC-:Reading inpvc file
',IN
PU
TF
ILE
10
TO
RN
I\T
IR601
CLOSE (UNIT-99)
OPEN IUNIT=lOO. FILE=lNPUTFILE, STRTUS='OLD'I
- c read the temp.
ec
c.
from ngm frie
c ..............................................
TSTIII-0
14
RED llOO,llrEND-19) ST&TION.CREST
IE ISTA(Il.NE.STIITIONi
WT
O 14
E
1%
PRINT',5TRlI1,'
da
ta found .
RLm 1100. lZ.EIID-19.ERR-18l
DRY
REUD ~lOO.12.END-1~,ERR-181 HOUR
READ IlO0.13.CND-19,ERR-18>
THelli
RE-
1100,16~EIID~191 TMP1ll.TEMP~I,li,TfWP~I,21,TENPl1,3~,
+TEHPlI,4IrTEHP~I,SI,TEMP~I,61,TEHPII,7l,TEHP~I,81,TEHPII,9l,
+TEMP~I,1O1~TEHP~l,111.TEHP11,12l,TEHP1I,131,TEMPII,14I,
+TENPII,lSI,TEHPII,161,TENPlll 17~,TEMPll.18l,TEnPl1.191
If ITHP(ll.NE.'TEMe
'1
WT
O 18
REUD 1100.13.END-191 TMP411
$8;;:;;?8 *,"-"-;" 21- E : ? ; " - - - -* 2 " - , , a .
.L) -̂ I,,," " . - e g 5;:;t; e R-
???Z e w e - - . . "
r e - ;;;; = * O f - w - p ..--* - . a >*: ' A " = - - o m . o m = .
N + ? - g : "- rn -. : ? - m r - " - -
3%%:5;s*z0?j:g*ga o z z - ~ - * - , < - f - " " x e a
. 3 r l * Z " Z O * - * * z " m '". .. 0-4 0 - " " " " 3 N . . O " a r e - - 0 n - 3 - ."..-,. w < " - - c a > - C X * r r r z m m - a . e z x 7 rF;za;
8 - n . - r u w " Z""" " ' a - - .
: 0 ? Z 4 0 :: -2z??F c m - >
2,. " ' A "
* .3 * w x
= - x*",--. r " * " m , P,
e r z - m - r ! s : ? : N - 3 .
" 0 . - 2 g ='k 2 0 3 ; m E ? G " " " 0 - - I
H a * .. .
< w -
c apply
the
average blaa
to m
e forecasc
C..............-.
............*.
................
W 69 I=I.I9
00
68 J-1.4
BIAS IJ.11-0.0
68 CONTINUE
69 CONTINUE
OPEN iUN11-96, ERR-72, TILE-"BIAS". STATUS-'0LD.l
70 CONTINUE
COHENT-*R~~~~~~
model bias dara'
GOTO 13
11 COHENT-"Error icadlng modal blas flle.
blns
GW
O 13
72 COHENT-.'H~~~I
bu
s tila
no
r found."
13 PRIM., COHENT
find the lndlvidual probablilitzer
c i
$a
n=
as
adm-2.L new as
et 18 DEC 941
6.................*.**.................--......
RE
WIN
D i1GZl
READ 6102.361
wn,Do,rr
READ 6102,371 HH
36 FORMAT ilX, 12.2% 12.2~. 121
31 FORWIT 61x.121
HH-HH-6
IF INH.LT.11
THEN
HN-HHt24
00-DD-1
END IF
IF IDD.LT.1, THEN
HN-wn-1
IF (MH.LT.11 THEN
m=m+12
YY-YY-1
EN
D IF
ODl-DOIl
,842-m IF (HHZ.EQ.Z.RN0.DOl.GT.281 THEN
HH2-3
001-1
END IE
IF IHH2.EQ.4.OR.~2.EQ.6.ORR~2.EQ.9.0~!HH2.~9.1l.
rlUIO.DDl.W.301 THEN
m2-m2+1
DDl-1
END IF
TT IHH2.EQ.1.1WD.I)01.GTT311 T
HE
N
m2=HN+1
Doll1
END
IF
IF IEM2.EQ.3.lUID.DO1.GT.311 THEN
m2-m.1
001-1
.-. ~
EN
D IT
IF (m2.EQ.lO.IWD.ODt.GT.31I
TH
EN
m2-m+1
DD1-1
END IF
IF 1~2.EQ.12.WD.DD1.GT.311
THEN
HHz-m+1
DD1-1
END IF
IF (m3.EQ.2.IUIO.DD2.GT.281
TH
EN
m3-3
DDZ-1
END IT
IF IHH3.EQ.P.0R.~3.~Q.6.0R.~3.EQ.3.0RRHH3.EQ~11.
rAND.DD2.GT.301 THEN
m3Im3rl
00211
END IF
IT (m3.EQ.1.RNO.DD2.GT.3ll
THEN
109
M13-bWtl
DOZ=l
END IF
Im3.EQ.3.IWD.DD2.GTT3i, THEN
m3-mr1
DD2-l END IF
If ~~~~~Q.IO.RNO.DD~.GT.~I~
THEN
m3nm+i
052-1
END
IF
IF (MJ.EQ.~~.wD.DDz.GT.~II TH
EN
m3-m+1
002-1
END IT
3s
POR
MA
T ('Fprecd=t created using NGM MOS
da
ta from a,I2,. .,
fA3.' ',12.
taken a
t 0'.11,
.oo CST.>
39 fORHI\T
1 'rere
ca
st c
reated usrng
NG
N NOS
da
Ca
cram
., 12, +
., *A3.
',72: ta
ke
n
ac
',12. .no
CST*~
IF ITIHE.EO.'lZOO'I
THEN
PRINT 38,OD.HONTH,YY. H
H
ELSE II ITIHE.EQ.'OOOO'I
THEN
PRINT 39. DD, HONTH. YY, HH
' END
IT
IF ITIHE.EQ.'O000'1
THEN
PRINT 41,HONTH. DDl.HONTH3.002
ELSE IF lTlNE.EQ.'1200'~ THEN
PRINT 4Z.NONTH.DO.MONTH2.DDl
END IF
31 CORMAT lii5.11IfS.0.1XIl
32 FORHRT I'YDIR '.lllTS.O.lXII
33 mRmT IA5.3x.111A1.5x1b
34 PORHRT
35 FORHRT
41 F
OWT
42 FORM0.T
SO FORHRT
si S
OWT
IF ITIHE.EQ.'0000'1
THEN
PRINT 35,'00'.'03';06',
'09'. .12., .IS', '18.. '21'.
'00'.
+'0
1'.
'05
' ~
.
, ..
ELSE IF ITINE.EQ.'1200'1
THEN
PRINT 35, '12'. '15'. '18'.
'21'.
'00'. '03'.
'06'. .09',
'12'.
+'IS8, '18,
EN0 IF
PRINT-. '
' PRINT 31,'TEMP ~.AT~HP<ll,ATEM?l2I,RTEHel3l,RTENPl4l,RT~~PlS~,
+IITEHP161,RTEHPl71.ATEHPla1,ATEHPl91,ATEHPllO1,ATEHPlll~
PRINT 31,'tlWPf
'~~PTI1I,RDHPTI2l,RDWPTl3l,RDWPTl4I,RDWPTlS~,
+RDWPTl6IrRDWPTl7I,RDYPTI8l.RDWPTI91,RDWPTl1OI,ROllPTlllI
PRINT 32,RDIRl11,~IR121,~lRl3l,RDIRI4~,I\DIRlSI,~IR16l,
+RDlRl7l,RDI~l~l,RDIRl9l,RDIRllO~,RDlRl1ll
PRINT 31,'WSPD ',AVELI1I,~VCLl2~.AVELl3~,AVELl~l,AYELlSl,
+AVELI61,RVELl7l,AV~LI8I.AVELl91,AVELIIOI,AVELl111
PRINT','
' PRINT*.
' PRINT-,'
Plume Foresa=r f
~r
US Hxghuay 30 n
ear ~hc
RDH plant'
PRINT-. ' .
PRINT 34,'PROB l',PB~GI1I,PR€OG121,PBFOGI3I,PBFOGl4l,PBFOGl5l,
+PBFOG(61.PBSOG17l.PSFOGl8l,PBFOG191rPBFOGIIOIrPR€OGI111
PRINT 33..CAT
~,C~TI11,U\Tl21,U\T131,U\T141,CATl51,CATl61,
IUITI~),U\T~~I,~T~~I,~T~~O~,CRT~~~I
PR
TM
V..
7
a
PRINT.;
' PRINT ', '
ABOVE'
PRIWT $0, 'ROW','-'
' . .
. 1 . 8
. .
. I . ,
, -, - 9 - <
-,
-,
-,
*,*_
.,*
I
a .
-.
."R
,,,T
. . 8
-
. . .- . . . ,
PRINT 51, 'COOLING'
PllNT .. '
UNITS'
PRINT S0,'WSEDq, '-'
' .
. .
1
. I
,
,-
I_
,_
_.
+
I
I
13
,
.
.. ,
-.
-
-,
PE
E&
,
* -,
-, -
pcini.,'Ths
fog caiegories are based span the predicted
'p'obabilizy
for a
' pcinc.,'coolinq
tow
er
fog e
etec~ing the Yl3ibiliiy along
+Highway 30 nea
r'
prlnt.:ehe
archer Daniels Hldiand plant.'
print..
' '
pi
c,
The criteria for each category is:
PEI"L-,'
' print..'
High
Piobabllicy giearer chan oi equal
to 80'
P'i"f.,'
' p
rin
~'.
' Medium
Probabrliiy
be
twe
en
70
an
d 80,
PCi"L.,'
- pzinc'.'
Lou
Probabilrry between 60 and 70.
p'inL..-
' pcinr.,'
ze
io
ProbabiliLy less than 60
- P'i"L'.
' '
prlnc-.'These
sategoiies provlde only a
general
ou
iLo
ok
of
+th
e cooling
Co
we
r .
print-.'plume behavmr based
on
a
naiisnal compucer lnodel
+which forecasrs
print*,'ueather conditions far she
are
a.
his fozeCart is
+experimental and
- print.,'u=eis
should be aleic to posalble large
err
oc
s in
+these forecast.
ot
print..'plume
be
ha
vla
r and wearher condr~io~~.,
52
PRINT..
' PRINT.,
' '
PRINT.,
' PRINT-,
' PRINT'.
' PRINT*, 'emarl
PRINT.,'
' CLOSE 11001
CLOSE 1991
CLOSE 1981
END
YIr
ala
n
',Y
ER
SIW
By
: E.S.
Takls
and
9~LlikleBlarca~e.c.u
. . . . . . . ,
PRINT-, ' '
PRINT.,'
OblorvaCions ol plume behavior'
PRIM'.
. ' PRINT ', .
OH'
PRINT .jo.'ROI\D','-',I
' '
' '
. I '
4 '
. '
. .
- <
-,
-,
-,-
, -, -
+
_,-
( '
'