Comparison of Satellite-Based Sensing on Accuracy and Assessable Field Numbers for Estimating Soybean Vegetation in a Large-Scale Farm

Abstract

In Japan, farmland accumulation has led to the development of large-scale farms that manage numerous soybean fields together. Excess-moisture injury during the early growth stages of soybean cultivation can reduce the yield. Thus, to enhance the soybean productivity, we need to identify the fields where soybean growth is severely suppressed through satellite-based sensing and constructing drainage. However, weather conditions complicate satellite-based sensing. The early growth stages of soybeans coincide with the rainy season, and cloud cover hinders satellite-based sensing. This study evaluated the practicality of sensing vegetation sensing using Sentinel-2, PlanetScope, and Sentinel-1 satellites. Investigations were conducted from 2021 to 2023 to compare the soybean coverage rate estimation accuracy and the number of fields where sensing was successful. The results indicate that Sentinel-2 and PlanetScope accurately estimate the coverage rate. However, the number of fields in which these satellites were used for sensing successfully varied annually. Conversely, although Sentinel-1 had lower accuracy in estimating coverage rates, Sentinel-1 allowed for assessment across more fields than Sentinel-2. According to our results, effective utilization of satellite-based sensing on large-scale farms requires improvement of the success rate of sensing using Sentinel-2 and PlanetScope or enhancing the accuracy of cover rate estimation using Sentinel-1.