County Level Crop Yield Prediction Using Smap Derived Data Products and Deep Learning Model

This paper aims to examine the impact of SMAP derived soil moisture data on the crop yield prediction of the deep learning neural network models. In this study, the crop yield model based on Bayesian neural network (BNN) is used. Corn crop yield prediction experimental results with different type of SMAP Level 4 soil moisture data as inputs are compared to illustrate the impact of the soil moisture data on deep learning crop yield prediction models. The preliminary results show that either rootzone or surface, or rootzone plus surface soil moisture data can improve the BNN model crop yield prediction or produce comparable results though which type of soil moisture data (surface or root zoon) will help improve the prediction accuracy is uncertain. Soil moisture data and precipitation plus NWDI data can produce comparable yield prediction results due to high correlation between soil moisture data and precipitation along with NWDI.