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

Abstract

Understanding the determinants of agricultural productivity requires accurate measurement of crop output and yield. In smallholder production systems across low-and middle-income countries, crop yields have traditionally been assessed based on farmer-reported production and land areas in household/farm surveys, occasionally by objective crop cuts for a sub-section of a farmer's plot, and rarely using full-plot harvests. In parallel, satellite data continue to improve in terms of spatial, temporal, and spectral resolution needed to discern 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. Using unique, simultaneous ground data on yields based on farmer reporting, sub-plot crop cutting, and full-plot harvests across hundreds of smallholder plots, we document large discrepancies among the ground-based measures, particularly among yields based on farmer-reporting versus sub-plot or full-plot crop cutting. Compared to yield measures based on either farmer-reporting or sub-plot crop cutting, satellite-based yield measures explain as much or more variation in yields based on (gold-standard) full-plot crop cuts. Further, estimates of the association between maize yield and various production factors (e.g., fertilizer, soil quality) are similar across crop cut-and satellite-based yield measures, with the use of the latter at times leading 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.

Sight for Sorghums: Comparisons of Satellite- and Ground-Based Sorghum Yield Estimates in Mali

“…The correlation slightly improves to 0.40 when omitting plots with self-report values above 2500 kg/ha ( Figure 3B). The low agreement between ground-based measures is similar to that reported for maize plots in Eastern Uganda, where self-report and crop cut yields had a correlation below 0.30 even after excluding very small plots where self-reports were deemed least reliable [16]. Moving to the comparison with satellite measures, Figure 4 shows the correlation between the two ground-based yield measures and GCVI for each date of Sentinel-2 imagery, both for the raw GCVI values and the harmonic fit at those dates.…”

“…For the current study we used the Sentinel-2 Level 1-C product, which represents top-of-atmosphere reflectance. From these raw bands we calculated several common vegetation indices that have been found to be useful for agricultural monitoring in similar settings [12,15,16].…”

“…results in weaker, stronger, or equivalent β coefficients (with expected signs) when using satellite-based yields. This approach avoids dependence on the assumption that the ground-based measures are free of error, and was used to evaluate estimates of maize yields in recent studies in Kenya [15] and Uganda [16]. Similar to the prior work, the following (self-reported) variables are included in our vector X: Household size; head of household age, and education level; plot size; plot distance to (i) household, (ii) road, and (iii) market; use of crop rotation and fallowing; frequency of ploughing; quantity of labor, seed, fertilizer, and pesticides used; and rainfall totals for the entire growing season and for the month of August, which is especially important for determining drought stress during flowering.…”

“…Satellite remote sensing offers a potential alternative approach to measuring crop yields, especially as satellite sensors with the fine spatial resolution needed to distinguish individual smallholder plots become more prevalent. Several decades of research has focused on developing and testing algorithms to estimate yields from satellite, initially in large commercial plots [13,14] and increasingly in smallholder systems [11,12,15,16]. These studies have primarily focused on the primary staple crops within a region, such as maize, wheat, and rice, although some recent work has extended beyond, such as to cotton, millet, and sorghum [12].…”

Predicting county-scale maize yields with publicly available data

“…Recent work has shown that some weather variables (like temperature and water/drought) have nonlinear effects on corn yield 27,28 . There also exists another category of yield prediction using remote sensing 29 which we do not consider for this study.…”