Analysis of corn and soybean yield variability at field scale using VHR satellite data

Crop yield assessment and forecasting is one of the major components of crop production monitoring. Timely and accurate information on crop yield at global, national, and regional scales are extremely important. In particular, spatial variability of crop yield at the field scale can help provide objective information, for example, to farmers to improve management practices and identify yield gaps, or to insurance companies to feed this information into insurance models.

The increased number of very high spatial resolution (VHR) sensors aboard space-borne platforms, including micro- and nano-satellites, provide new opportunities to exploit multiple VHR data sources for generating new products for the study of land cover and land use change. A high temporal resolution is usually required for agricultural applications to monitor and capture the dynamics of crop growth. Exploiting high spatial resolution (1-3 m) data provided by WorldView-2/3 (WV-2/3) satellites and PlanetScope satellites enables better spatial resolution compared to moderate spatial resolution satellites, such as Landsat 8 (30 m) and Sentinel-2 (10-20 m). In this work, we aim to explore VHR data to explain in-field variability of yield of corn and soybean. In-field yields of soybean and corn were collected in Iowa, USA, and were correlated with multi-spectral satellite data acquired by WorldView-3 and PlanetScope. Results show that the most important spectral bands explaining corn/soybean yield variability are green/yellow, red edge and NIR. High temporal frequency of Planet allowed identification of best suitable date for yield assessment: PlanetScope's spectral bands at 3 m explained 60% of in-field corn/soybean variability.