DETECTING RIPARIAN VEGETATION RESPONSES TO GROUNDWATER CHANGES USING SENTINEL SATELLITE IMAGERY AND CLOUD-BASED COMPUTING

Groundwater dependent ecosystems (GDEs) along river corridors are biological hotspots that often exist at the epicenter of intense human activities, but are rarely considered during water resource management decisions. Until recently, technological limitations have hindered assessment of groundwater dynamics on riparian GDE health at the spatial and temporal scales relevant to policy and management. Here, we synthesize large, publicly-available datasets of mapped riparian vegetation, groundwater levels, and stream flow regimes to assess the influence of groundwater on riparian vegetation across California, where sustainable groundwater management is legislated and groundwater depletion is a concern. We used Sentinel-2 satellite imagery to assess riparian vegetation health via NDVI (normalized difference vegetation index) at 10-m spatial resolution - a scale required to characterize these narrow, highly fragmented landscapes. We link riparian NDVI responses to available field-based groundwater level data for ~20,000 riparian vegetation polygons, stratified by hydrologic region, seasonal and interannual trends, and degree of streamflow alteration. We show that NDVI in riparian forests decreases as groundwater levels decline, suggesting a strong dependence on shallow aquifers.We also show that reliance on groundwater is greatest in the summer, particularly for sites without perennial water sources (natural or anthropogenic). We also find that riparian vegetation along anthropogenically modified stream reaches, where vegetation is subsidized by managed water sources, produced higher NDVI values than vegetation in natural stream segments. Despite the prevalence of anthropogenic surface water influences on riparian vegetation, groundwater reliance is still high. Therefore, efforts to alleviate water stress in riparian GDEs require regional monitoring and coordinated management of surface flows, groundwater well levels, and vegetation responses to water availability.