A method to estimate the water storage of on-farm reservoirs by detecting slope gradients based on multi-spectral drone data

Water storage dynamics in on-farm reservoirs (OFRs) are crucial for irrigation water allocation and utilization, ensuring agricultural development sustainability. Previous studies have primarily relied on the area-storage model to estimate reservoir water storage using meter-level remotely sensed data, which often falls short of accurately capturing the water storage dynamics of OFRs, especially in small OFRs with steep slopes. Hence, we proposed a method to estimate the water storage of OFRs in irrigation areas by integrating multispectral drone data, high-resolution remote sensing data, and ground observations. The water surface area was extracted from multispectral drone data using a Gaussian Mixture Model (GMM) and a threshold segmentation method. Slope gradients were then obtained by identifying the maximum potential slope zone (PSZ) and utilizing high-resolution drone-based Digital Surface Models (DSMs). The dam slopes of the constructed boundaries were automatically computed considering the significant decline of slope gradients. By combining the dam slopes of the OFRs with water depth observations, we estimated the construction depth (H). Subsequently, current water depth was obtained using drone-derived Digital Surface Models (DSMs), calculating the elevation difference between drone-derived OFRs and water surface boundary. Once the OFR morphology was fully constructed, the water storage was calculated based on area-storage and depth-storage methods using the 3D volume module in Arcpy. The derived water storage agrees well with in situ observation (R2: 0.99) using slope gradients, reaching an overall accuracy of 95.2 %, with a root mean square error (RMSE) and a mean absolute error (MAE) of 2785 m3 and 1820 m3, respectively. Notably, discernible fluctuations in water storage were observed during the main irrigation phases, highlighting the essential role of OFRs in promoting equitable water resource distribution and enhancing irrigation water management. This integrated approach offers a robust solution for monitoring and managing water storage dynamics in agricultural areas.