Food security will become increasingly threatened over the upcoming decades, due to a growing population, climate change, and natural resource degradation. This is particularly true in India, where climate change impacts are expected to be especially large, with up to a 30% loss in yield for some staple crops by mid-century (Lobell et al. 2008). Furthermore, over 40% of agricultural production relies on groundwater irrigation, however groundwater reserves are becoming rapidly depleted, with some studies estimating that a large proportion of deep wells will dry up by mid-century (Shah et al. 2009). While the impacts of environmental change on production have been well established, there is little understanding of how farmers respond to this change. Yet, it is important to account for farmer behavior as farmers may be able to reduce or eliminate the negative impacts of environmental change by adapting their cropping practices. For example, farmers may be able to reduce the impact of warming temperatures by switching to new hybrid crop varieties that are more heat-tolerant. Understanding how, why, and how effectively farmers may adapt their cropping strategies to environmental change will better identify whether India will be able to produce enough food over the upcoming decades. This proposal will examine land use and land cover change (LCLUC) of agricultural systems, and attribute these changes to long-term environmental drivers, like climate change and groundwater depletion. This will allow us to understand how effectively farmers have adapted to environmental change, and how vulnerable current agricultural systems still are to future change. Specifically, we will derive novel remote sensing products that quantify smallholder crop production, including cropped area and yield from 1995 to the present. To date, mapping the production of smallholder farms has been difficult for several reasons. First, the size of individual farms is smaller than the resolution of readily-available satellite imagery, like Landsat and MODIS, leading to issues with mixed pixels. Furthermore, ground data rarely exist to calibrate models that translate satellite vegetation indices to production measures like yield. We propose to develop unique methods to overcome these problems that build on previous work by the PIs (Jain et al. 2013, Lobell et al. 2015). We will also use remote sensing to quantify adaptation decisions, including shifting sow date, switching crop variety, and increasing irrigation. We will link these remote sensing datasets with gridded weather, groundwater depth, and panel household datasets to examine how farmers are responding to climate change and groundwater depletion. This study will be one of the first to identify specific adaptation strategies farmers adopt in response to medium to long-term environmental change. We will also evaluate how effective these strategies are in bolstering future food security.