(UNTITLED)
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP85T00287R000600580001-4
Release Decision:
RIPPUB
Original Classification:
S
Document Page Count:
18
Document Creation Date:
January 12, 2017
Document Release Date:
July 19, 2011
Sequence Number:
1
Case Number:
Publication Date:
May 13, 1983
Content Type:
MEMO
File:
Attachment | Size |
---|---|
![]() | 820.89 KB |
Body:
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
.' " ?' 4Y 1333
MEMORANDUM FOR: -(See Addressees List)
FROM: 25X1
1e , Strategic Resources Division, OGI
SUBJECT: PRED: A Predictive Weather Model for Crop Forecasting 25X1
1. The attached memorandum describes a new weather forecasting model which
will be used primarily in future assessments of the Soviet grain crop.
2. This paper was prepared byl ~ Agricultural Assessments
Branch, Strategic Resources Division, Office of Global Issues.
3. Comments and queries are welcome and may be addressed to the Chief,
Agricultural Assessments Branch, on
Attachment:
PRED: A Predictive Weather Model for
Crop Forecasting, May 1983, GI M 83-10129
25X1
25X1
25X1
25X1
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
SUBJECT: PRED: A Predictive Weather Model for Crop Forecasting
Mr. Elmer Klumpp
Special Assistant to the Under Secretary
International Affairs and Commodity Programs
Department of Agriculture
Dr. Terry N. Barr,
Acting Chairman, World Agriculture Outlook Board
Department of Agriculture
Mr. . Jimmy Murphy
Acting Assistant Administrator
International Agriculture Statistics
Department of Agriculture
Mr. Donald J. Novotny
Director, Grain and Feed Division
Foreign Agriculture Service
Department of Agriculture
Mr. David Schoonover
Director, Centrally Planned Economics Division
Foreign Agriculture Service
Department of Agriculture.
Mr. Gerald A. Bange
Director, Foreign Production Estimates Division
Department of Agriculture
Mr. Anton F. Malish
Chief, USSR/Eastern Europe-Branch
Economic Research Service
Department of Agriculture
Mr. Keith Severin
Foreign Production Estimates Division
USSR/Eastern Europe
Department of Agriculture
Mr. Frank Game
Foreign Agriculture Service
Department of Agriculture
Mr. John Danylyk
Bureau of Intelligence and Research
Office of Economic Analysis
Department of State
Mr. David Peterson
Director, Office of Intelligence Liaison
Department of Commerce
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
I I
SUBJECT: PRED: A Predictive Weather Model for Crop Forecasting
OGI/SRD/AAA I(13 May 1983)
Distribution List: (Attachment with Each Copy)
Orig - Each Addressee
1 - D/GI
1 - DD/E
1 - DD/SOVA
1 - C/SOVA/SE
1 - SOVA/SE/R
1 - C/ECD/CM
1 - C/SRD
4 - C/SRD/AA
8 - OGI/PS
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
I I
Central Intelligence Agency
DIRECTORATE OF INTELLIGENCE
3 fi~~AY 1x33
Pred: A Predictive Weather Model For Crop Forecasting
Sutunary
The estimation of grain output in the USSR will be significantly improved
this year by the introduction and use of the PRED model. In the past the
estimation of crop yield has been handicapped by the lack of a reliable
predictive weather input to UPSTREET's agronomically based, weather driven,
crop-growth simulation model. In the absence of this essential data, analysts
could only forecast the maximum yield. This required an evaluation of the
impact of elapsed weather on crop yield and the assumption that ideal growing
weather would prevail for the remainder of-the season. With the completion of
PRED, which has the capacity to produce representative weather data for an
entire season, the acceptance of the ideal weather assumption is no longer
necessary. Furthermore, the accuracy and utility of the assessments produced
during the early season are enhanced. In this report the use of the PRED
model, as an adjunct to the UPSTREET analytical effort, is examined and
evaluated.
This paper was prepared by thel (Agricultural Assessments
Branch, Strategic Resources Division, Office of Global Issues (AAB/SRD/OGI).
Comments and queries may be addressed to the Chief, AAB/SRD/OGI
25X1
25X1
L_ 25X1
25X1
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
PRED: A Predictive Weather Model for Crop Forecasting
PRED was developed to provide predictive weather inputs to the crop model
used in the UPSTREET analysis of grain production in the USSR. By its
calculated employment of historical weather data to determine the likely
character of future weather, beyond the date on which the crop model is run,
PRED provides information that has previously been unavailable. By generating
a variety of representative weather scenarios, an average, (i.e. best
estimate) crop yield can be determined-as well as information regarding the
uncertainty of this estimate. This approach enhances the applicability of the
UPSTREET methodology and lends credence to assessments of crop production made
early in the growing season.
The following sections describe the PRED model. Section I describes how
the historical data is processed and details the creation of 200 independent,
season long weather scenarios. Section II describes how a sample of weather
scenarios (typically 25) are selected from this collection of 200 to reflect
specified long term weather projections; Section III then describes the
calculation and analysis of summary weather and crop-yield statistics.
Section I: Data Collection and Assembly
The first section of PRED compiles historical weather data2 to provide a
base for generating parameters to describe weather conditions and develop
weather scenarios. The historical weather data base includes at least eight
years of information on daily maximum and minimum temperatures, precipitation,
and evapotranspiration (ETP) for up to 3000 locations.
To create future weather scenarios, the historical data are processed
according to a serifs of statistical routines which establish probabilities,
expressed as curves , for the following weather parameters:
o occurrence of precipitation
1 UPSTREET - The UPSTREET methodology is an interdisciplinary approach to
grain production forecasting developed by CIA. It was first employed to
forecast Soviet grain production in 1975. The methodology is dynamic and
improvements are made nearly every year.
2 The historic weather data is obtained from ETAC, the Envirormental
Technical Applications Center, of the US Air Force. 25X1
3 ETP - The potential mount of moisture lost because of the combined effects
of evaporation from the earth's surface and transpiration from leaves and
other parts of plants.
4 These are actually cumulative probability curves with, for example, maximum
temperature values on_ the x-axis and cumulative probability values ranging
from 0 to 1.0 on they-axis.
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
o amount of precipitation
o maximum temperature level
o minimum temperature level
o ETP
There are two sets of probability curves for the temperatures and ETP: one
for days with precipitation and one for days without. For maximum and minimum
temperature and ETP, probability curves are derived for each half-month period
in the year. For precipitation amount, a probability. curve is derived for
each month.
From these probability curves, 200 weather scenarios are created. The
occurrence of precipitation, temperature, precipitation and ETP values for
each day are selected from the probability distributions using a randan number
generator. It is assumed that these 200 scenarios are a representative sample
of all possible weather situations, given past history.
In the second section of PRED, a subset of scenarios, usually 25, is
selected to represent the weather at each location and to serve as input to
the crop model. Each scenario created in Section I of PRED is characterized
as above normal, normal, or below normal in terns of both precipitation and
temperature.
Scenario selection is initiated by the user who provides forecasts for
each location. These are usually monthly forecasts but can be for any
interval of time. The user can derive forecasts independently or use
forecasts that are available from several universities and the National
Weather Service Long Range Predictive Group. Each forecast consists of four
declarations:
o temperature - below, near, or above normal
o confidence - a subjective probability that the
temperature will actually fall into the above
categories .
o precipitation - below, near, or above normal
o confidence - a subjective probability that the
precipitation will actually fall into the above
categories.
Since weather is never totally predictable, this type of declaration lets some
of the 25 scenarios chosen fall outside of the forecasted precipitation and
temperature categories.
Table 1 presents some examples of scenario selection for different
forecasts. For example, if the forecast is for near normal temperatures and
near normal precipitation,' with 90% confidence levels for both forecasts, then
the vast majority of the scenarios selected will be normal in both temperature
and precipitation. With the same forecast but only a 60% confidence level on
both forecasts, only five of the scenarios selected will be. normal in both
temperature and precipitation. As the confidence level decreases, the
dispersion in the scenarios selected increases.
25X1
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
I I
Examples of 25 Scenarios Selected for 3 Different Forecasts
Number of Scenarios in Each
Example Forecast Temperature Precipitation Class
Example 1
B N A
Near Normal Temperature (N) B 0 2 0
Near Normal Precipitation (N) Precip N 1 19 1
with 90% Confidence on both A 0 2 0
Example 2
B N A
Near Normal Temperature (N) B 2 3 21
Near Normal Precipitation (N) Precip N 3 5 3
with 60%.Confidence on both A 2 3 2
imp
Example 3
B = Below Normal
N = Near Normal
A = Above Normal
B N A
Above Normal Temperature (A) B 2 1 3
Above Normal Precipitation (A) Precip N 2 2 3
with 60% Confidence on both A 3 4 5
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Section III: Utilization of PRED Output
Once the sample of weather scenarios has been selected for each location,
it is simply a matter of appending them, one at a time to observed weather to
date, and running the crop model. The results are 25 (or whatever number was
selected) end-of-season percent of maximum-yield-remaining estimates for each
location. PRED averages crop yields over all scenarios to give a best
estimate of the final yield at that location. The average yield is given
along with ?ne standard deviation above and below the average (68% confidence
Applications and Limitations
PRED significantly enhances the capacity of UPSTREET-type analysis by
developing in a rational and objective way, predictive weather which can be
used to drive the UPSTREET crop model. PRED is not restricted in use to the
USSR; it can be used in any country for which historical weather data is
available. In addition, it also has potential for use with CROPCAST, a
generic version of the UPSTREET crop model, also used primarily for grain
estimates; with regression-type models designed for the assessment of grains
and non-grains; and with the Ritchie generic crop growth model which can be
employed to forecast yields of a wide variety of crops. F__]
PRED also can be used to assess the effect of weather on crops in any
given season. In this operational mode the user inputs a forecast of average
temperature and precipitation for the entire growing season. The resulting
yield values. can be compared with yield values obtained by running the crop
model on actual weather for the full growing season.
Furthermore, PRED has the potential to project the range of grain yields
within established time frames by forecasting extremes in weather, i.e., in
seasons that are excessively dry or very cool. It does not have the capacity
to forecast short-term weather conditions, however, and it can neither
anticipate such phenomena nor accept forecasts of the likelihood of their
occurrence.
Finally, PRED is structurally limited to 40 stations or locations per
run, but multiple runs can be made. In an area as large as the USSR, it will
therefore be most useful when applied to key indicative areas.
5 See the appendix for examples of output.
25X1
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
I I
Appendix 1
This appendix describes the functions and interrelationships of files and
programs that collectively canprise the PRED model. It will be most valuable
when used in conjunction with the diagram of the entire PRED model (see
figure 1). Sections I, II and III are divided by dotted lines on the
figure. The crop model and random number generator module shown on the figure
are exterior routines. Appendix 2 describes the files and their required
Section I of PRED compiles historical weather data and arranges it for
use in developing weather scenarios. It contains a total of six files (the
Historic Data Base file; the Historic Monthly file; the Monthly Check file;
the Station Statistics file; the Scenario file; and the Scenario Attributes
file) and three programs (MONFILE 2, CLIMSTAT, and SCENARIO).
Section II of PRED selects predictive weather for input to the crop
model. Within this Section there are three files (the Forecast Input file;
the PREDMET file; and the PREDMET Output file) and one program (PREDMET).
Section III identifies the locations in the crop model that correspond to
the PRED stations, and provides PRED weather to the crop model so that it can
compute final yields. This Section contains a total of six files (the PRED
Cell List; the Yield Strip file; the Last Day Model file; the Last Day Strip
file; the Last Day PRED file; and the LOSS file) and the PREDSTRPR (PRED
Stripper) and LOSS programs.
This file is canprised of daily weather records of maximum and minimum
temperatures (in degrees C), precipitation amounts (in millimeters) and
evapotranspiration (ETP) rates (in millimeters) for a maximum of 3,000
locations. Stations for which less than eight years of data are available are
not included. The data, arranged chronologically, are derived fran reports
generated by stations participating in the World Meteorological Organization
(WM10) reporting program; they are quality controlled and reformatted for
agricultural applications by the US Air Force Environmental Technical
Applications Center. The data base includes information on the USSR dating
from 1968.
Historical Data Base Variable Description
Station Number
Year
Julian Day
ETP (MM)
Precipitation (mm)
Maximum Temperature (C)
Minimum Temperature (C)
I4onfile 2 Program
The MONFILE 2 program reads the Historical Data Base file, groups and
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Figure 1
PRED Flow Diagram
Historic
Data Base
Station
Statistics
File
RANDOM
M RINAE
Predmet
Input
File
PRED
Cell
List
Yield
Strip File
MON W 2
CLIMSTA'
REDI5IET
PREDSTRPR
Historic
Monthly
File
Monthly
Check
Data
File
Scenario
File ~7
Scenario
Attributes
File
t
ff et
1
Strip File
Last Day
-----
I
I CROP
MODEL
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Last Day
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
I I
reformats the information into monthly increments, and stores the results in
the Historic Monthly file.
Historic Monthly File
The Historic Monthly file is the repository of data reformatted by the
t4)NFILE 2 program. In this file the data for each location are grouped into
monthly records, which became input to the CLIMSTAT program.
Monthly Check Data File
The Monthly Check Data file is built by the user and contains information
used to control operation of the CLIMSTAT program. The information
includes: the station number, the total number of months of weather data in
the station's data base, and a processing flag which indicates if the station
is to be processed.
CLIMSTAT.Program
The CLIMSTAT program reads the Monthly Check Data file to establish, for
each half-month in the year, precipitation, temperature, and
evapotranspiration parameters for each station. The probability of
precipitation for a certain day is defined by a first order Markov chain which
expresses the probability of precipitation as a function of whether or not
precipitation occurred on the previous day. The values of temperature and ETP
are assumed to follow normal distributions that are defined by a mean and
standard deviation computed from historical data. The amount of precipitation
is assumed to follow an Incomplete Gamma distribution defined by two
parameters--gamma and beta. For maximum and minimum temperatures and ETP,
distributions are defined for. each half-month; for precipitation amount, the
resolution is at the whole month level.
This file receives and stores the output of the CLIMSTAT program, in
which one set of statistics per station is produced. Each set of statistics
contains climatological parameters for each station, and these parameters (a
total of 16) are organized by half-month increments.
Station Statistics File Parameters
Precipitation Probabilities for the Next Day:
1. Probability when previous day had precipitation.
2. Probability when previous day had no precipitation.
For a Precipitation Day:
3. Maximum temperature - mean
4. Maximum temperature - standard deviation
5. Minimum temperature - mean
6. Minimum temperature - standard deviation
7. ETP - mean
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
8. ETP - standard deviation
9. Gamma -- (Used to define the Incomplete Gamma Function
10. Beta -- that describes the amount of precipitation).
For a Dry Day:
11. Maximum temperature - mean
12. Maximum temperature - standard deviation
13. Minimum temperature - mean
14. Minimum temperature - standard deviation
15. ETP - mean
16. ETP - standard deviation
Scenario Program
The Scenario program uses data in the Station Statistics file, in
conjunction with a random number generator, to build 200 season-long weather
scenarios for each station. Random numbers are used to translate the
parameters from CLIMSTAT into weather outcomes for each day. The program also
builds the Scenario Attributes file, which contains the day-today
accumulations of temperature and precipitation for each scenario.
The Scenario file is created by the SCENARIO program and is the
repository of the 200 season-long weather scenarios produced by the latter.
This very large file is used by the PREDMET program. To save space the .
records contain no information on stations or date; hence PRED software must
be provided with accurate information or, the sequence of stations as well as
the start and end dates of the scenarios. The user must keep separate records
on the order of data in this file.
The Scenario Attributes file contains accumulations of temperature and
precipitation for each scenario in the Scenario file. The records in this
file are used by the PREEMET program to evaluate relative warmth and wetness
of the developed scenarios in the scenario selection process. In addition,
the Scenario Attributes file allows the user to make comparisons of any
. portion of a scenario with a like portion of any other scenario.
Forecast Input File
The Forecast Input file is created by the user and contains weather
forecasts for each station for specified time periods. The forecasts indicate
whether temperature and precipitation will be below average, average, or above
average. Forecasts can be given for a maximum of 11 time intervals. A
confidence level is also established for each forecast. This file contains a
temperature forecast code, atemperature confidence level, a precipitation
forecast code and a precipitation confidence level for each of the 11 forecast
intervals.
25X1
25X1
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Forecast Input File Description
3-Month Forecast - 3 Locations
Location
Month
1
1
1
2
1
3
2
1
2
2
2
3
3
1
3
2
3
3
A = Above Normal
N = Normal
B = Below Normal
Temperature Precipitation
Forecast-Confidence Forecast-Confidence
A 90% N 80%
N 70% N 70%
N 60% N 60%
B 80% A 80%
B 70% N 70%
N 60% N 60%
N 90% A 90%
N 70% N 70%
N 70% A 70%
PREDMET Input File
The PREDMET Input file is created to control the operation of the PREDMET
run. These controls establish starting and stopping dates for each run, state
the number of iterations to be included in the run, define the forecast
intervals to be employed, and list the stations that are to be assessed. The
file is updated by the user as needed.
The Random Module uses random numbers fran 0.0 to 1.0 as probabilities to
determine whether or not precipitation occurs, the amount of that
,precipitation, maximum and minimum temperatures, and ETP for each day in
building a year-long scenario. It is also used in selecting the subset of
scenarios by the PREEMET program.
PREDMET Program
The PREE1ET program uses the Forecast Input file, PRE71DET Input file,
Scenario file, Scenario Attributes file, and the Random Module to control the
selection of scenarios covering the period that extends fran a selected date
to the end of the growing season or other selected stopping date. The 25
scenarios thus selected for each location will be biased toward the forecast
but with a statistically meaningful sample spanning a broader range of
conditions. Program output is stored in the PREEMET Output file and serves as
the meteorological input for the crop model.
PREEf1ET Output File
The PREEMET Output file is structured exactly like the meteorological
data input file (MET) used by the crop assessment model. Information in it
can thus be used by the crop model without change. This file contains the
daily meteorological data of the 25 scenarios selected for each location.
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Each record in this file contains a day's worth of meteorological information
for a single location and one of the 25 iterations.
Last Day Model File
This file is created by the crop model, using actual weather up to the
day PRED is run. PRED uses this file to generate the Last Day Strip file and,
via the crop model, the Last Day PRED file.
The PRED Cell List is a control file that matches the selected locations
in PRED with locations in the crop model.6
PREDSTRPR Program
The PREDSTRPR program (PRED Stripper) reads the output of the Last Day
rbdel file and selects records that pertain to the locations specified in the
PRED Cell List. It creates the Last Day Strip file, which inputs to the crop
model, and the Yield Strip file, which inputs directly to the LOSS program. F
Last Day Strip File
The Last Day Strip file, generated by PREDSTRPR, is. identical to the Last
Day Model file except that it also contains an iteration number that
corresponds to the weather scenario being processed. It is input to the crop
Yield Strip File
This file contains the estimated percent of yield remaining, as of the
beginning day on which the crop model is run on PRED data, at each crop model
location.
Last Day PRED File
The Last Day PRED file is the output Last Day file derived fran running
the crop model or. PRED data. It contains end-of-season values, expressed in
yield percent, for each location and each weather scenario. This is input to
the LOSS program.
LOSS Program
The LOSS program canputes the mean and standard deviation of the results
of the Last Day PRED file. Wien weather data are not available for all
locations in a region, the results of the Last Day PRED file are used in
conjunction with the Yield Strip file to determine the mean incremental change
in yield and its standard deviation. These results are assigned to locations
using a nearest neighbor assignment scheme. This output forms the LOSS
6 The crop model runs on a gridded cell format.
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
file.
The LOSS file is the output mean and standard deviation for all locations
in the LOSS program. Figure 2 demonstrates these results for one location.
Tb illustrate, this average end-of-season crop yield for spring wheat is given
to be 54.1% of the historical maximum yield for location 1. Some gather
scenarios deteunined crop yields that were less than this, sane more, as
indicated by the spread in the distribution. The actual standard deviation
(SD) was calculated to be 6.2, so that a 95% confidence interval about 54.1%
would be 54.1% +,12.4% (i.e., + 2 standard deviations). The interval 41.7% to
66.5% will contain the true or actual yield 95% of the time. Table 2 contains
a sample output for 30 locations.
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Figure 2
Distribution of Crop Yield as a Percent of Maximum for Location 1
Percent of Maximum Yield Remaining
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-R DP85T00287R000600580001-4 25X1
I I
Location
Estimated Crop Yield As A Percentage of Thtal
With One Standard Deviation Limit
Spring Vheat
47.9
54.1
60.3
59.8
66.3
72.8
54.4
58.7
63.0
43.6
50.0
56.4
52.6
56.3
60.0
66.8
70.7
74.6
63.3
68.8
74.3
45.3
48.8
52.3
57.0
61.8
66.6
55.2
60.4
65.6
64.0
67.5
71.0
53.5
58.7
63.9
.66.9
72.3
77.7
44.7
47.2
49.7
69.8
73.0
76.2
68.9
73.5
78.1
72.1
76.1
80.1
69.1
73.8
78.5
63.9
67.3
70.7
62.4
66.2
70.0
81.1
84.5
87.9
50.6
54.0
57.4
55.6
57.9
60.2
61.0
65.0
69.0
81.8
85.3
88.8
73.1
76.9
80.7
78.3
81.5
84.7
73.4
78.3
83.2
68.5
74.1
79.7
69.3
73.4
77.5
25X1
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4
I I
Appendix 2
Description Files
Created By
Used By
Medium
Storage
(Bytes)
Historic Data Base
External
MONFILE2
Tape
8000/location/year
Historic Monthly
MJNFILE2
CLIMSTAT
Tape
6000/location/year
Monthly Check Data
User
CLIMSTAT
Disk
14/location
Station Statistics
CLIMSTAT
SCENARIO
Tape or Disk
1536/location
Scenario
SCENARIO
PREDMET
Tape
600,000/location
Scenario Attributes
SCENARIO
PRELIU11ET
Tape
300,000/location
Predmet Input
User
PREDMET
Disk
1000
Forecast Input
User
PREDMET
Disk
200/location
Predmet Output
PREDMET
Crop Model
Tape or Disk
28/location/iteration/day
Pred Cell List
user
PREDSTRPR
Disk
7/location
Yield Strip
PREDSTRPR
LOSS
Disk
32/location
Last Day Strip
PREDSTRPR
Crop Model
Disk
264/location/iteration
Last Day Pred
PREDI OD
LOSS
Disk or Tape
264/location/iteration
Last Day Model
Crop Model
PREDSTRPR/LOSS
Tape
264/crop model locations
LOSS
External
Disk or Tape
92/crop model locations
Notes:
number of locations :
20-40
number of years :
10-15
number of iterations:
25-50
number of days :
30-300
25X1
13
Sanitized Copy Approved for Release 2011/07/19: CIA-RDP85T00287R000600580001-4