Variable rate irrigation of crops is a promising approach for saving water whilst maintaining crop yields in the semi-arid American Mountain West – much of which is currently experiencing a mega drought. The first step in determining irrigation zones involves characterizing the patterns of spatial variation in soil moisture and determining if these are relatively stable temporally in relation to topographic features and soil texture. Characterizing variable rate irrigation zones is usually done with ancillary data that is likely to be related to the variations in soil moisture. Then soil moisture sensors are usually installed in each zone. Very little is usually known about the variation of soil moisture within the zones or in the field in general away from the soil sensor locations. More detailed characterization of patterns of soil moisture variation can be achieved through sampling and lab analysis, but this is prohibitively time-consuming and expensive.

In Southern Idaho, one of the main agricultural areas in the American Mountain West, annual precipitation levels are typically <500mm with most falling as winter snow or spring rain. Svedin et al. (2021) note that in such locations winter snowfall and spring thaw act as the principal wetting event and that the variation of water content at spring green-up seems to be a function of snow accumulation and melting patterns over the winter and early spring. This research investigates whether snow melt patterns measured using time-series Sentinel 2 imagery in Google Earth Engine can be used to infer spatial variation in spring soil moisture.  The study focuses on two field sites in Southern Idaho. Dense soil sampling followed by laboratory determination of soil moisture content has been undertaken for several seasons at both sites to allow accurate geostatistical mapping of soil moisture patterns in the fields. The study quantifies how accurately the patterns in snow melt can predict the spring soil moisture patterns each year. The study also investigates how snow melt patterns are related to patterns in aspect caused by micro-topographic features derived from drone imagery. Aspect is likely to be related to patterns in thermal radiation received at the surface throughout the growing season which can drive spatial variations in within season evapotranspiration rates. Automated analysis of snow melt patterns in Google Earth Engine has the potential to provide a veritable gold mine of relevant information for variable rate irrigation in the semi-arid Mountain West at very little cost to the farmer.