卧虎山水库不同算法LS因子值适用性分析

在总结以往LS因子相关算法研究基础上,结合McCool的LS因子参照值,对LS算法的适用性进行评价分析。研究表明,除去复合算法和Remortel修正算法外,其他不同算法LS因子值都小于参照值;研究区LS因子值的最优算法为基于Remortel迭代运算的修正算法;其次为结合刘宝元的陡坡公式和Remortel改进L指数因子迭代运算的复合算法,以及Remortel第4版AML程序算法;再次为B?hner算法,而Moore算法和Desmet算法,由于其与参照值的相关性相对较差,而且其RMSE相对较大,不推荐在该区使用。

The Applicability of LS Factor Value from Different Algorithms in Wohushan Reservior Basin

The Universal Soil Loss Equation (USLE) model and its principal derivative and the Revised Universal Soil Loss Equation (RUSLE) model have been widely used in the past decades. However, the use of USLE and RUSLE has been limited by the inability to generate reliable estimates of the LS factor. Several different LS factor algorithms from the previous studies were briefly summarized in this article and their applicability was evaluated in Wohushan reservoir basin. According to the Agriculture Handbook No. 703 and 537 of US Agriculture Department, the LS-values in McCool's table are the same as the LS algorithms in USLE/RUSLE. Although there is some regional heterogeneity in the specific regional applications for LS calculations, the difference is very limited within a certain slope length and slope gradient. Based on these reasons, the LS-values from McCools are primarily preferred as the reference value. There are four basic LS algorithms which were Remortal, Moore, Desmet and B?hner used to be compared with reference value. In addition, two revised algorithms, i.e. the improved iterative Remortal algorithm and complex algorithm, were presented. The slope-length exponent (m) in the former algorithm was revised from low rill/interrill ratio class to moderate class. The complex algorithm was composed of L-factor and S-factor from different research, of which the latter was from the above mentioned improved algorithm of Remortel and S-factor was made up of S algorithm from McCool and that of Liu BY. In this article, the LS values of the six above algorithms were compared with that of McCools by RMSE (the Root Mean Square Error), the correlation coefficient and the slope of the regression equation. The results indicated that, other than the improved algorithm of Remortel and the complex algorithm, the LS-value obtained by different algorithms are all less than that of reference value. It is also found that the optimal algorithm in the study area is the improved iterative algorithm of Remortel, followed by both the AML program of LS factor from RUSLE Version 4 of Remortel and the complex algorithm. The B?hner’s algorithm could also be used in this area. However, the algorithms from Moore and Desmet were recommended not to use in the study area because of their relatively higher RMSEs and relatively poor correlation coefficients.