Development of a dynamic Natural Resources Conservation Service Curve Number (NRCS-CN) to account for the vegetation and soil moisture effect on hydrological processes.

摘要

This study proposes an approach that makes use of remote sensing-based products to automatically adjust the Natural Resources Conservation Service Curve Number (NRCS-CN or simply CN) hydrological model to improve runoff estimates. The CN is adjusted to account for the effect of vegetation density changes and soil moisture content on hydrological processes. The proposed approach consists of two stages; first, we propose a new method to integrate the effect of vegetation growth on hydrological processes in the determination of CN, which does not include this factor according to its standard version. Second, we investigate the adjustment of CN based on antecedent soil moisture conditions prior to rainfall events. Then, we have integrated the changes in a hydrological model to assess their impact, specifically, on FFG as their determination is based on the CN method. The MOdel Parameter Estimation EXperiment database (MOPEX ) is used to develop and test the proposed approach. The information used includes data from over 9 watersheds across the U.S., which includes the daily gauged precipitation (P) and runoff (Q ) observations from 1948 to 2003. The normalized difference vegetation index (NDVI), derived from a 5-year (1985-1990) Advanced Very High Resolution Radiometer (AVHRR) observations, has been used to estimate the Greenness Fraction (GF) as a proxy for the vegetation density. The vegetation growth throughout the year was assessed via estimation of monthly averaged CNs using P-Q pairs which were then compared to the monthly averaged GF. The improvement in the performance of the CN methodology was assessed with respect to the standard approach, which does not account for the vegetation growth over time and only uses static inputs related to soil texture and land use. The results evidenced how the vegetation-adjusted CN (CNveg adj) compensates the underestimation of the standard CN (CNstd). The correlation coefficient (R2) between the simulated and observed runoff when using the unadjusted and adjusted CNs was 0.63 and 0.80, respectively. In the same order, a Nash-Sutcliffe coefficient (NSC) of -0.17 and 0.67 and the Root Square Mean Error ( RSME) of 5.22 and 2.75 were also obtained.