895Nitrogen (N) use in intensive agriculture can degrade groundwater resources. However, considerable time lags between groundwater recharge and extraction complicate source attribution and remedial responses. We construct a historic N mass balance of two agricultural regions of California to understand trends and drivers of past and present N loading to groundwater . Changes in groundwater N loading result from historic changes in three factors: the extent of agriculture (cropland area and livestock herd increased 120 and 800%, respectively), the intensity of agriculture (synthetic and manure waste effluent N input rates increased by 525 and 1500%, respectively), and the efficiency of agriculture (crop and milk production per unit of N input increased by 25 and 19%, respectively). The net consequence has been a greater-than-order-of-magnitude increase in nitrate (NO 3 -) loading over the time period, with 163 Gg N yr -1 now being leached to groundwater from approximately 1.3 million ha of farmland (not including alfalfa [Medicago sativa L.]). Meeting safe drinking water standards would require NO 3 -leaching reductions of over 70% from current levels through reductions in excess manure applications, which accounts for nearly half of all groundwater N loading, and through synthetic N management improvements. This represents a broad challenge given current economic and technical conditions of California farming if farm productivity is to be maintained. The findings illustrate the growing tension-characteristic of agricultural regions globally-between intensifying food, feed, fiber, and biofuel production and preserving clean water.

Abstract. This study investigates the ability of machine learning models to retrieve the surface soil moisture of a grassland area from multispectral remote sensing carried out using an unoccupied aircraft system (UAS). In addition to multispectral images, we use terrain attributes derived from a digital elevation model and hydrological variables of precipitation and potential evapotranspiration as covariates to predict surface soil moisture. We tested four different machine learning algorithms and interrogated the models to rank the importance of different variables and to understand their relationship with surface soil moisture. All the machine learning algorithms we tested were able to predict soil moisture with good accuracy. The boosted regression tree algorithm was marginally the best, with a mean absolute error of 3.8 % volumetric moisture content. Variable importance analysis revealed that the four most important variables were precipitation, reflectance in the red wavelengths, potential evapotranspiration, and topographic position indices (TPI). Our results demonstrate that the dynamics of soil water status across heterogeneous terrain may be adequately described and predicted by UAS remote sensing and machine learning. Our modeling approach and the variable importance and relationships we have assessed in this study should be useful for management and environmental modeling tasks where spatially explicit soil moisture information is important.

Irrigated cropland represents the largest source of groundwater nitrate (NO 3 ) pollution in the Central Valley (CV) of California. Mitigation, through the use of best management practices that maximize crop nitrogen use efficiency (NUE), will be most effective in reducing pollution if used where the risk of NO 3 leaching loss is greatest. The University of California's Nitrate Groundwater Pollution Hazard Index (HI) tool was used to map the risk of NO 3 leaching below the rootzone in irrigated fields in a portion of the CV. The HI is an expert system that calculates an index value based on soil properties, crop characteristics, and type of irrigation system in use. Depth to groundwater, aquifer recharge rate, and actual farm management practices (e.g., rate of nitrogen [N] fertilizer applied) are not considered in the calculation. Application of the HI to 1,318,000 ha (3,256,848 ac) of irrigated cropland in the four southernmost counties of the CV revealed that 31% of the area is at high risk of NO 3 leaching loss if not managed carefully. Adoption of drip or microsprinkler irrigation on all orchards, vineyards, and vegetable fields would decrease the area rated as most vulnerable from 31% to 20% of the area analyzed. Crop fields on permeable soils and/or irrigated by surface gravity methods contributed the most to the area at high risk. The HI can help the USDA, regulatory agencies, and Cooperative Extension target regulatory, research, and education efforts.

We developed machine learning models to retrieve surface soil moisture (0-4 cm) from high resolution multispectral imagery, terrain attributes, and local climate covariates. Using a small unmanned aircraft system (UAS) equipped with a multispectral sensor we captured high resolution imagery in part to create a high-resolution digital elevation model (DEM) as well as quantify relative vegetation photosynthetic status. We tested four different machine learning algorithms. The boosted regression tree algorithm provided the best accuracy model with mean absolute error of 3.8 % volumetric water content. The most important variables for the prediction of soil moisture were precipitation, reflectance in the red wavelengths, potential evapotranspiration, and topographic position indices (TPI). Our results demonstrate that the dynamics of soil water status across heterogeneous terrain may be adequately described and predicted by UAS remote sensing data and machine learning. Our modeling approach and the variable importance and relationships we have assessed in this study should be useful for management and environmental modeling tasks where spatially explicit soil moisture information is important.