The SMAP-Based Dynamic Agricultural Productivity Indicator for Improved Corn and Soybean Yield Productivity Forecasts

Agricultural yield productivity can be estimated by calculating the area under the curve of the satellite-basded Normalized Difference Vegetation Index (NDVI) during the growing season. Since NDVI variation can be predicted by water availability in the soil in advance, soil moisture information can potentially contribute to anticipating yield productivity at the end of harvest time. In this work, we propose a two-step process to improve yield productivity predictability for corn and soybeans on a global scale. Our experiments were conducted to predict yield productivity for the most productive corn and soybean regions in the world, as well as two major corn and soybean agricultural regions in the U.S. and Argentina. In the first step, we derive root zone soil moisture (RZSM) from surface moisture data provided by the Soil Moisture Active Passive (SMAP). Concurrently, we forecast NDVI using a well-established lag time between NDVI and RZSM. Accurate NDVI forecasts is vital for better estimating the area under the NDVI curve, especially during the growing season. In the second step, we utilize this area under the NDVI curve, which correlates strongly with agricultural yield, to improve end-of-season yield forecasts. SMAP soil moisture data is shown to significantly contribute to the accuracy of these forecasts. Knowledge of expected end-of-season crop yield prior to the end of the growing season is essential for a variety of users, including agricultural agencies, farmers, insurance companies, and state (local) and federal governments.