Estimating index of sediment connectivity using a smart data-driven model

Connectivity is a key topic in investigations of hydrogeomorphic processes in catchments. Mapping and deter-mining quantitative values of hydrologic and sediment connectivity are prerequisites for understanding the linkages amongst various parts of catchment systems with different land use and topographic changes to properly manage sediment and water-related issues. Given the need to develop appropriate methods for determining connectivity indices, we developed a novel approach to estimate index of sediment connectivity (IC) using an artificial neural network (ANN) algorithm. Physical characteristics of the catchments, including elevation, slope, area, length of stream channel, length of overland flow, normalized difference vegetation index (NDVI), and surface soil moisture (SSM) were used as inputs and the IC was assessed as the output of the ANN model. Our findings revealed that the accuracy of the modelling within homogenous sub-sections (based on IC values) of catchments was higher than that at the catchment scale, even with fewer input parameters. Also, the use of dynamic parameters (e.g., SSM/NDVI) along with physiographic data improved performance of the ANN model in estimating IC. The results obtained from this study indicate that the proposed method can be applied as a promising, reliable, and cost-effective tool for estimating IC at the catchment scale and also within homogenous sections of catchment.