Monitoring water storage of on-farm reservoirs using remote sensing and ground observation

The development of artificial reservoirs plays a considerable role in regulating the spatial and temporal distribution of irrigated rainfall and guaranteeing sustainable agricultural development. Many studies have used the area-storage relationship to obtain the storage capacity of on-farm reservoirs (OFRs), but it does not work for OFRs with persistent water surface area. In this study, we proposed an effective method to estimate the water storage of irrigated OFRs by combining multi-source remote sensing data and ground observation. We quickly derived the location of irrigated OFRs by using seasonal characteristics of irrigated OFRs and obtained high-precision water surface area using an object-oriented segmentation. We estimated water storage of irrigated OFRs by combining three different ways (i.e., Lidar-based, ground observation-based (photos), and surface area-based). The method performs well in three aspects, i.e., identifying on-farm reservoirs, extracting water surface area, and calculating water storage. The accuracy of identification reaches 94.1%, and the derived water area agrees well with the surveyed results, i.e., an overall accuracy of 97.8%, the root mean square error (RMSE) and the mean absolute errors (MAE) are 962 m² and 766 m², respectively. The obtained water storage is reliable using three different ways (i.e., the area-storage, Lidar-based, and photo observations-based methods), with accuracy of 98.8%, 95.2%, and 94.1%, respectively. The proposed method enables monitoring of the storage of multiple types of irrigated OFRs, particularly the photo observation-based method can deal with the storage of OFRs with persistent water areas, showing huge potential to promote irrigated water resource utilization efficiency.