The Impact of Crop Rotation on Crop Attributes Detection Using Sentinel-1A Observations

Crop rotation increases crop yield by returning nutrients to the soil and increasing soil fertility. Additionally, it can improve soil health by increasing the biodiversity of the farm and improving the soil structure. Although crop rotation plays an important role in agricultural modeling and predictions, it is difficult to get reliable data at fine resolutions. This study investigates how crop rotation affects crop attributes estimation using Sentinel-1A observations. A vegetation sampling has been conducted on three corn fields and three soybean fields in Michigan for the growing season of 2022. The fields are subjected to different irrigation conditions and plowing. The growth cycle of corn and soybean is estimated by training two Machine Learning algorithms i.e., support vector machines and random forests using Sentinel-1A data. The backscatter time series of two different soybean fields were compared; one was plowed and the other had residues from the previous year's corn. As a result of the dry corn effect, it is becoming increasingly difficult to differentiate between a corn field and soybean planted on a harvested unploughed corn field due to the surface backscatter increase. While corn's faster growth rate is supposed to make a distinct backscatter time series from soybeans'. Therefore, to estimate crop attributes on an unploughed field with residue from the previous year's crop, it is necessary to use ancillary data (e.g., optical observations) along with the radar data. We will present some results and outcomes of this ongoing study. We hypothesize that our study will improve the existing crop detection algorithms that will make a significant impact on crop classification, monitoring phenology, and yield estimation.