Eyes in the Sky, Boots on the Ground: Assessing Satellite- and Ground-Based Approaches to Crop Yield Measurement and Analysis

However, these methods have been shown to have severe systematic biases, particularly in plot area measures. In contrast, satellite data have improved in terms of spatial, temporal, and spectral resolution, enabling the discernment of performance on smallholder plots. This study evaluates ground- and satellite-based approaches to estimating crop yields and yield responsiveness to inputs, using data on maize from Eastern Uganda. The research team collected unique, simultaneous ground data on yields based on farmer reporting, sub-plot crop cutting, and full-plot harvests across hundreds of smallholder plots. The results show large discrepancies among the ground-based measures, particularly among yields based on farmer-reporting versus sub-plot or full-plot crop cutting. Satellite-based yield measures explain as much or more variation in yields based on full-plot crop cuts. Furthermore, estimates of the association between maize yield and various production factors, such as fertilizer and soil quality, are similar across crop cut- and satellite-based yield measures. The use of satellite-based yield measures at times leads to more significant results due to larger sample sizes. Overall, the results suggest a substantial role for satellite-based yield estimation in measuring and understanding agricultural productivity in the developing world. Improving the productivity of smallholder farmers is widely considered to be one of the most effective avenues for reducing their poverty and food insecurity. With agriculture contributing up to 69% of rural household income in Africa, productivity improvements remain a longstanding goal in many African countries. Accurate measurements of crop production, cultivated area, and yield are at the heart of official agricultural statistics and are key to monitoring progress towards national and international development goals. The most common way to assess outcomes related to the productivity of smallholder farmers is by using information collected through in-person interviews for household and farm surveys. However, these methods have been shown to have severe systematic biases, particularly in plot area measures. A less common but also well-established approach to measure crop yields is by physically harvesting a sub-section of a farmer's plot, also known as crop cutting. Crop cutting provides a more objective way to measure grain production for a part of the plots, but heterogeneity within a plot can lead to sensitivities of crop cut yields to the precise location and size of the crop cut sub-plot vis-à-vis the entire plot. Recent work has explored the ability of satellite data to track crop yields. Burke and Lobell (2017) showed that 1 m resolution data from Terra Bella's Skysat sensors were useful for mapping maize yields for farms in western Kenya. This usefulness was measured both by correlation of satellite-based yield estimates with traditional ground-based yield measures, as well as by the ability of satellite-based yield estimates to track yield responsiveness to inputs. The results of this study suggest that satellite-based yield estimation can be a valuable tool for improving the accuracy of methods used to measure land productivity.