Description
Agriculture is sensitive to changes in water supply, especially in irrigated, arid regions. Within a region, water availability varies between users by position in the watershed, water source, community affiliation and farm size. The Mexicali Valley in Northern Mexico experienced water shortages due to groundwater over-exploitation beginning in 2009 and an earthquake in 2010. Managers providing program support to vulnerable farmers may not have access to data at spatial resolutions that reflect variability in water supply. We ask which farmers are most impacted by scarcity events and how they adapt. We combine satellite imagery (MODIS 250m and Landsat 30m spatial resolution) of Normalized Difference Vegetation Index (NDVI) with farmer surveys and interviews to document drivers behind regional patterns in agricultural change in the Mexicali Valley. We classified timeseries of mean-annual NDVI based on statistical criteria such as magnitude and p-value of the slope, absolute change, range and maximum residual from the trendline. Major categories of spatially clustered agricultural activity included reduced (brownspots) or enhanced NDVI (greenspots) which may have occurred from fallowing, change in crop type or number of seasons. We interviewed farmers in remotely identified brownspots and greenspots to attribute these changes to water scarcity and other drivers like market prices, labor prices, etc. Next, we conducted contingency tests between drivers of agricultural changes, responses to drivers and socioeconomic traits. Farmers in regions affected by groundwater over exploitation or earthquake damage were associated with brownspots, confirming that satellite image time series can identify agricultural water scarcity. Contrary to expectation, declining groundwater levels along the Mexico-U.S. border resulted in a small reduction of cultivated area (1.5 km2) while an increase in agricultural area (13.2 km2) occurred near the border. Damage to a main canal from the 2010 earthquake temporarily caused fallowing over a larger cultivated area (24.5 km2). Farmer responses to drivers depended on their resources and entitlements; smallholder and communal farms were disproportionately destabilized by water scarcity. Remote sensing products and interviews from this study may be useful to social scientists for targeting populations affected by agricultural change and to water managers for improving farm resilience to future water scarcity.
