The USLE‐M and the USLE‐MM estimate event plot soil loss. In both models, the erosivity term is given by the runoff coefficient, QR, times the single‐storm erosion index, EI30. In the USLE‐MM, QREI30 is raised to an exponent b1 > 1 whereas b1 = 1 is assumed in the USLE‐M. Simple linear regression analysis can be applied to parameterize both models, but logarithmically transformed data have to be used for USLE‐MM. Parameterizing the USLE‐MM with nonlinear regression of untransformed data could be a more appropriate procedure. A statistical check of the two suggested models (USLE‐M and USLE‐MM), considering two alternative parameterization procedures for the USLE‐MM, was carried out for the Masse and Sparacia experimental stations, in Italy. The analysis showed that the USLE‐MM with the linear regression parameterization procedure was the only correctly specified model, that is, with normally distributed and homoscedastic residuals. With this model, the normalized soil loss, Ae,N, prediction error did not exceed a factor of 5.7 for Ae,N > 17.3 Mg ha−1 at Masse and of 3.5 for Ae,N > 27.5 Mg ha−1 at Sparacia. Stable values of b1 require inclusion of high Ae,N values in the calibration dataset. Using a common exponent b1 for the two stations increases the practical interest for the model and did not imply a substantial worsening of the model performances, especially for the highest soil loss values. Development of a USLE‐MM‐type model having a wide applicability appears possible, and data from other experimental sites could make this conclusion more robust.
Statistical check of USLE-M and USLE-MM to predict bare plot soil loss in two Italian environments
Mannocchi, FrancescoMembro del Collaboration Group
;Todisco, FrancescaMembro del Collaboration Group
;Vergni, LorenzoMembro del Collaboration Group
2018
Abstract
The USLE‐M and the USLE‐MM estimate event plot soil loss. In both models, the erosivity term is given by the runoff coefficient, QR, times the single‐storm erosion index, EI30. In the USLE‐MM, QREI30 is raised to an exponent b1 > 1 whereas b1 = 1 is assumed in the USLE‐M. Simple linear regression analysis can be applied to parameterize both models, but logarithmically transformed data have to be used for USLE‐MM. Parameterizing the USLE‐MM with nonlinear regression of untransformed data could be a more appropriate procedure. A statistical check of the two suggested models (USLE‐M and USLE‐MM), considering two alternative parameterization procedures for the USLE‐MM, was carried out for the Masse and Sparacia experimental stations, in Italy. The analysis showed that the USLE‐MM with the linear regression parameterization procedure was the only correctly specified model, that is, with normally distributed and homoscedastic residuals. With this model, the normalized soil loss, Ae,N, prediction error did not exceed a factor of 5.7 for Ae,N > 17.3 Mg ha−1 at Masse and of 3.5 for Ae,N > 27.5 Mg ha−1 at Sparacia. Stable values of b1 require inclusion of high Ae,N values in the calibration dataset. Using a common exponent b1 for the two stations increases the practical interest for the model and did not imply a substantial worsening of the model performances, especially for the highest soil loss values. Development of a USLE‐MM‐type model having a wide applicability appears possible, and data from other experimental sites could make this conclusion more robust.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
