Spatial patterns of sediment dynamics within a medium-sized watershed over an extreme storm event

Abstract

In this study, we quantified spatial patterns of sediment dynamics in a watershed of 311km2 over an extreme storm event using watershed modeling and statistical analyses. First, we calibrated a watershed model, Dynamic Watershed Simulation Model (DWSM) by comparing the predicted with calculated hydrograph and sedigraph at the outlet for this event. Then we predicted values of event runoff volume (V), peak flow (Qpeak), and two types of event sediment yields for lumped morphological units that contain 42 overland elements and 21 channel segments within the study watershed. Two overland elements and the connected channel segment form a first-order subwatershed, several of which constitute a larger nested subwatershed. Next we examined (i) the relationships between these variables and area (A), precipitation (P), mean slope (S), soil erodibility factor, and percent of crop and pasture lands for all overland elements (i.e., the small spatial scale, SSS), and (ii) those between sediment yield, Qpeak, A, P, and event runoff depth (h) for the first-order and nested subwatersheds along two main creeks of the study watershed (i.e., the larger spatial scales, LSS). We found that at the SSS, sediment yield was nonlinearly well related to A and P, but not Qpeak and h; whereas at the LSS, linear relationships between sediment yield and Qpeak existed, so did the Qpeak–A, and Qpeak–P relationships. This linearity suggests the increased connectivity from the SSS to LSS, which was caused by ignorance of channel processes within overland elements. It also implies that sediment was transported at capacity during the extreme event. So controlling sediment supply from the most erodible overland elements may not efficiently reduce the downstream sediment load.