The potential value of seasonal climate forecasts for use in agriculture depends on many factors, including those related to the agricultural enterprise and its sensitivity to climate variability, the type of forecast and its lead time and accuracy, and the ability and willingness of farmers and others to change their decisions in response to climate forecasts. Crop models are useful for translating predicted climate anomalies into estimates of crop yield anomalies and for estimating the value of forecasts. However, these models should be evaluated for local conditions to make sure they accurately simulate differences in production under the range of climate conditions anticipated in the study area.

For this exercise, a crop model will be used along with daily weather data provided on a CD and diskette, respectively. Instructions for installing the software and data are provided in an attached sheet (Attachment 1). The crop model is embedded in a user-friendly software package designed for use by farmers and their advisors. This package was developed by funding from the United Soybean Board in the US and is made available through the Internet by private agricultural supply and service companies in the soybean-producing states. An Abstract of this software is provided (Attachment 2).

Weather Data

In this exercise, we provide daily weather data for a number of years for a farming community in Southern Georgia (Tifton). Daily data from 1950 through 1998 were available for use. These yearly files of daily weather data were divided into 3 categories depending on rainfall during May, June, and July of each year: (1) years with rainfall amounts falling in the highest 1/3 of the cases, (2) years with rainfall falling in the middle 1/3, and (3) years with rainfall amounts falling in the lowest 1/3 of the cases. A fourth set of weather data were created by selecting at random from all of the available years of data. Ten years of data are available for each category (Wet, moderate, Dry, and All) on the diskette that is provided.

For this exercise, consider these daily weather data to be realizations of a climate prediction that is presented using terciles. The format would thus forecast that rainfall during the next 3 months will be in the lowest 1/3, middle 1/3 or highest 1/3 of the historical amounts. In the exercise, we assume that this forecast is perfect in the sense that the forecast correctly predicts the tercile to which the next 3 months of rainfall will belong. Thus, if the forecast is for rain to be in the lowest tercile, the 10 years of "Dry" weather data can be used as different possible realizations of this forecast.

Estimating Crop Production for Each Tercile

We will use soybean as an example crop, and we will compute the expected yield using daily weather data for each tercile as well as for the randomly selected years. In addition, the distribution of yields can be estimated from the 10 years of results that are simulated, for each category. This step will provide an estimate of the differences in yield one might expect for different (perfect) tercile forecasts, using the same management practices for each case. These results will clearly show differences in expected yields for each case, but they will not provide any information about the value of forecast information to the farmer. In other words, in this case the farmer always uses the same management practices without considering climate predictions. Changes in yield can not be attributed to the use of climate predictions. Changes in management must be investigated to determine if there is potential value to the farmer by conditioning management practices to climate predictions.

Steps for this exercise are:

1. Run PCYield. You can run it from the Start button, or from the MS Explorer by double clicking on the PCYield.exe file under the C:\Program Files\PCYield directory.
2. Edit the "Test Field" by using the menu on the top bar of PCYield. First, select the Weather Location to be "Tifton, Georgia" using the pull-down menu. Secondly, select the soil, "Silt Loam, Productivity=0.9, Shallow, Well-Drained, Moderately Permeable". Then click the "OK" button to save these changes for the field. The same weather location and soil will be used for all runs.
3. On the main PCYield screen, set the "Selected Projection Date" to July 19, 1997. NOTE, you must use the year 1997 because we have not retrieved current weather data from the Internet for this field. This date will be used for most of the simulations.
4. Enter the Planting dates for "Current" and "Normal" years to be June 1. Select the Maturity Group 5 variety for both the current year and for normal years. These are normal practices for Tifton.
5. Now you are almost ready to run PCYield for the first time. First, however, you must copy the correct weather data to use for the first run from the diskette. Copy all 10 files from A:\Dry into the directory C:\Program Files\PCYield\MyFarm\. Files already in the directory will be replaced; answer "Yes" to replace existing files. This will allow you to simulate results for the lower tercile rainfall years.
6. Click the "Compute" button on the bottom right corner of the main screen. This cause the CROPGRO-Soybean model to simulate soybean results for the lower tercile case, using the normal planting date and variety.
7. The next screen shows average, lowest and highest predicted soybean yield for the "Dry" tercile case, in the bottom table. Record these numbers on a piece of paper for comparison with the results from other cases.
8. Look at the graphs of rainfall and growth by clicking the buttons at the bottom of the page.
9. Repeat Steps 5-8 for each of the other sets of weather data (copied from the diskette from directories Mod, Wet, and All for middle and highest terciles and random years, respectively).
10. Be ready to discuss what the results mean.

Investigating Possible Changes in Management

Now that you know how to operate PCYield, you can investigate the possibility for increasing yields for each set of weather files by changes in management. Only two management variables can be changed with this simple interface: variety and planting date. Here, we will attempt to do this for the Dry case first. Copy the lowest tercile weather data from the A:\Dry directory into the C:\Program Files\PCYield\MyFarm directory. Make several runs, one after the other, each with different combinations of planting date and variety. You should not use a planting date earlier than April first nor later than July 1. And, you should consider only varieties from Maturity Groups 4-8. Record the results nad which variety and planting date that you use in each case.

Repeat this process using the randomly selected years. Compare results for these two cases. If you have time, repeat the process for the highest and middle terciles (using weather data from the Wet and Mod directories on A:\, respectively.

Be ready to discuss these results and compare them with what others find.

Installation of PCYield

1. Insert the PCYield CD into the CD Drive and run the setup.exe program on the CD.
2. Select the state of Georgia for isntallation.
3. Select default paths for installation. Make sure that the PCYield directory was created under the C:\Program Files\ directory.
4. Insert the Diskette into the drive for later copying weather files for each case into the C:\Program Files\PCYield\MyFarm directory. Check to make sure that the diskette has 4 directories, each with 10 weather files.

Page Last Updated: September 9, 1999

Contact Information: Guillermo Podestá, Institute Science Coordinator
e-mail: gpodesta@rsmas.miami.edu
Telephone: 1.305.361.4142
FAX: 1.305.361.4622