Statistical evaluation of proxies for estimating the rainfall erosivity factor

Abstract

Considering the high-temporal-resolution rainfall data requirements for calculating the Rainfall Erosivity factor (that is, the R-factor), studies have developed a large number of proxies for the R-factor (PR). This study aims to evaluate 15 widely used proxies, which were developed in various countries using daily, monthly, or yearly rainfall data, in terms of correlation and statistical equality with the R-factor by using the 6-min pluviographic data from 28 stations in Australia. Meng’s test was applied to rank the correlations. Although the Meng’s test indicated that the correlation between Rainfall Erosivity (R) and Rainfall Erosivity calculated by the proxy model (PR) generally increased with a finer time resolution of the rainfall data (in the order of year, month, and day), the 15 PRs under examination were all highly correlated with R (r > 0.62, p < 0.004), implying that all of them can be reasonably used as an R predictor. A direct estimation of the R-factor using PRs produced a mean relative error (MRE), root mean square error (RMSE), and Nash–Sutcliffe efficiency coefficient (NSE) with a mean of 50.0%, 1392 MJ mm ha−1 h−1 a−1, and 0.17, respectively. The linear calibrations improved the accuracy of the estimation and produced an MRE, RMSE, and NSE with a mean of 36.0%, 887 MJ mm ha−1 h−1 a−1, and 0.70, respectively. Finally, suitable proxies for instances where only daily, monthly, or yearly rainfall data are available were recommended.

Annual and Seasonal Characteristics of Rainfall Erosivity in the Eastern Rhodopes (Bulgaria)

“…The utilisation of the MFI has been recommended by several authors as a reliable index for the assessment of the potential of the precipitation to generate erosion. The authors in [38,45] found a significant correlation between the MFI and the rainfall erosivity factor (R) in the universal soil loss equation (USLE).…”