Determination of cultivar coefficients for the CSM-cropgro-peanut model using variety trial data

Detailed growth analysis and developmental data are normally required for calibration of crop simulation models in order to determine the cultivar coefficients that correspond to the unique characteristics of each crop and cultivar. The objective of this study was to evaluate the suitability of using variety trial data in which no phenology data were collected for estimating the cultivar coefficients for the CSM-CROPGRO-Peanut model. We selected three peanut cultivars of different maturity, which are commonly grown cultivars in the southeastern U.S. The cultivars were an early maturity cultivar (Virugard), a medium maturity cultivar (Georgia Green), and a late maturity cultivar (C99R). The variety trial data for these cultivars were obtained from the research reports on variety performance tests conducted in three locations in Georgia from 1994 to 2001. A similar cultivar in the same maturity group was used as a starting point from which to calibrate each of the cultivars. Synthetic cultivars with corresponding values for the different combinations of the selected cultivar coefficients were created for the CSM-CROPGRO-Peanut model; the model was then run for each of these cultivars. The set of coefficients for the synthetic cultivar that produced the lowest root mean square error (RMSE) between simulated and observed yield from the variety trials was adopted as the final values for the candidate cultivar. The accuracy of the cultivar coefficients was evaluated using data on plant growth, development, and yield that were collected during the 2005 growing season in a research station and two farmers' fields in southwest Georgia. Our results demonstrated that variety trial data in which no phenology data have been collected could be used successfully to derive cultivar coefficients for the CSM-CROPGRO-Peanut model. Simulated values of various development and growth parameters were in good agreement with their corresponding observed values for almost all parameters that were evaluated. At final harvest, the RMSE between observed and simulated yield components was 259 kg ha ^-1 for seed yield, 301 kg ha ^-1 for pod yield, and 608 kg ha ^-1 for aboveground biomass. The calibration procedure described in this article could facilitate the estimation of cultivar coefficients for new peanut cultivars if local variety trial data are available.