Remote sensing offers a low-cost way to obtain crop production statistics across large spatial and temporal scales. Yet it has been difficult to map the production of smallholder farms, given that the size of an individual field (< 2 ha) is typically smaller than the spatial resolution of most readily-available satellite imagery (e.g. Landsat, MODIS). My work develops new methods and uses new micro-satellite data that provide high spatial and temporal resolution imagery to map the production of smallholder farms.