Agricultural drought risk assessment: An operational model for Nebraska

An agricultural drought risk assessment model was developed for corn and soybean on the basis of feature variables derived from the Standardized Precipitation Index (SPI) and Crop-Specific Drought Index (CSDI) using multivariate techniques. This model can be used to assess real-time agricultural drought risk on specific crops at critical points prior to and during growing season by retaining previous and adding current weather information as the crop passes through the various development stages. This model will be helpful to decision makers ranging from agricultural producers to policy makers from local to national levels. The results of the model validation using three different datasets show that the risk assessment accuracy improves as the crop develops. At the end of April before corn is planted, the average correct assessment rate of drought risks on final yield is 65%. At the beginning of July when corn is at vegetative development, the average correct assessment rate reaches to 73%. In late July when corn is at Ovule development, the rate increases to 80%. The rates are 88% in the second half of August as well as the end of September, when corn is at reproduction and ripening developments, respectively. The model assessment accuracy for soybean is lower than that for corn at earlier growth stages because the relationship between soybean and weather at earlier stages is not as highly correlated as corn. A reliable assessment with 79% correct assessment rate begins at mid-August when soybean is at pod formation stage. In early September and October, when soybean is at pod fill and ripening stages, respectively, the model is able to assess risks on soybean yield with 85% correct assessment rate. To provide a better visualization of agricultural drought risk assessment, the state drought risk assessment map was presented by combining three data layers, including a county level map of Nebraska, a map of leading counties in crop harvesting, and a map of assessment results of each county in GIS