资源环境科技发展态势分析平台

Runoff prediction in ungauged catchments has been one of the major challenges in the past decades. However, due to the tremendous heterogeneity of the catchments, obstacles exist in deducing model parameters for ungauged catchments from gauged ones. We propose a novel approach to predict ungauged runoff with Catchment Morphing (CM) using a fully distributed model. CM is defined as by changing the catchment characteristics (area and slope here) from the baseline model built with a gauged catchment to model the ungauged ones. As a proof of concept, a case study on seven catchments in the UK has been used to demonstrate the proposed scheme. Comparing the predicted with measured runoff, the Nash-Sutcliffe efficiency (NSE) varies from 0.03 to 0.69 in six catchments. Moreover, NSEs are significantly improved (up to 0.81) when considering the discrepancy of percentage runoff between the target and baseline catchments. A distinct advantage has been experienced by comparing the CM with a traditional method for ungauged catchments. The advantages are: (a) less demand of the similarity between the baseline catchment and the ungauged catchment, (b) less demand of available data, and (c) potentially widely applicable in varied catchments. This study demonstrates the feasibility of the proposed scheme as a potentially powerful alternative to the conventional methods in runoff predictions of ungauged catchments. Clearly, more work beyond this pilot study is needed to explore and develop this new approach further to maturity by the hydrological community.

Plain Language Summary For flow forecasting in areas without observed data as reference, we propose a novel approach to transfer from a known area to an unknown area, with the idea of morphing. The approach is a potentially powerful alternative to the conventional methods in runoff predictions of ungauged area.