Using Satellite Data to Improve Model Estimates of Crop Yield

Abstract It has been proposed that remotely sensed information from satellites could complement the performance of crop growth models. A study was conducted to determine how a model could effectively utilize satellite data and if model estimates of crop yield could be significantly improved by satellite data. A simple model was developed for simulating growth and yield of grain sorghum Sorghum bicolor (L.) Moench on an individual‐field basis. The model contained three state variables (stage of development, green leaf area index GLAI, and aboveground dry mass) and utilized daily meteorological observations. When available, GLAI values for target fields could be used to adjust the initial values of the state variables until a “best fit” of the observed data was iteratively achieved by the simulation. GLAI values could be obtained by ground‐based measurements, but satellite observations were demonstrated to be an appropriate source of data for initializing the model. The model was developed and verified using data from 10 fields observed in Central Texas in 1976. The model was tested using a completely independent data set containing yield and satellite observations from 37 fields in South Texas in 1973, 1975, 1976, and 1977. Without using the initializing procedure, the average yield for the 37 fields was underestimated by approximately 30%. Use of satellite‐derived GLAI data to initialize the same simulations resulted in a 2% overestimate of average yield. The results of this study confirm the usefulness of the initializing procedure and satellite data to improve model estimates of crop yield.