Remotely-sensed water budgets for agriculture in the upper midwestern United States

Process based hydrological models are often the most accurate and comprehensive tools available for answering hydrological questions. However, the resources needed to parameterize and calibrate them can be problematic for small organizations with limited resources or for management questions where emerging problems require rapid investigation. To address this need, this study calculates a simplified monthly water balance using gridded remote sensing actual evapotranspiration (ETa) data and PRISM precipitation data for 223,355 agricultural parcels in Wisconsin totaling 2,966,250 ha of agricultural land. Agricultural management features such as crop rotation and irrigation as well as physical features such as soil drainage and hillslope gradient were compared to identify differences in monthly water budgets. Results showed that Wisconsin was energy limited in most years such that additional precipitation was more likely to suppress ETa. Significant differences were identified with dairy and pasture rotations having the highest annual ETa and the potato/vegetable rotation having the lowest annual ETa rates. Growing season length was identified as a likely predictor of annual ETa. Surprisingly, results also showed that there were no appreciable differences in ETa between rainfed cropland and irrigated cropland. Lastly, results indicated that across all rotations, the months most susceptible to infiltration and runoff were April, May, October, and December. This research demonstrates that cropland management plays a significant role in mitigating or accentuating potential infiltration and runoff relative to variability in precipitation in mesic areas. Without accommodating hydrological variability in mesic, energy-limited areas, water or nutrient conservation policy strategies designed to target average weather conditions may be regularly under-protective or over-restrictive.