Statistical Observations of Water Stress in Kansas Winter Wheat and Corn from Remotely Sensed Evapotranspiration

Abstract:

In this study, we performed exploratory statistical analyses on two satellite-based evapotranspiration (ET) datasets to view the differences in their responses under conditions of water stress in rainfed winter wheat and irrigated corn in southwest Kansas. Our goal is to observe statistical distributions and temporal trends between a Penman-Monteith-based ET algorithm (MOD16) and a surface energy balance algorithm (SSEBop) to obtain a clearer picture of the sensitivity of these ET products to conditions of water stress and overwatering throughout the growing season. We found that with winter wheat, the MOD16 data were heavily skewed and were best approximated by a log-normal distribution. They showed clear deviations from normality in drought years whereas overwatered and sufficiently watered crops consistently followed a normal distribution. SSEBop ET demonstrated consistent deviations from normality in underwatered crops in wet years while closely following a normal distribution in drought years. Additionally, analyses of temporal trends in both datasets suggested lower and later maximum ET values in underwatered crops. This study is a beginning step towards developing a statistically-based ET prediction model that could then be utilized to identify zones of crop water stress throughout the growing season. Knowing the differences between these products and their responses is key to understanding which ET algorithm works best for agricultural applications and why.