Stochastic Soil Erosion Risk Modelling and Simulation Using Fournier Index

Abstract

Approximately 280 million tons of cereal crop is lost from 108 million hectares of croplands through soil erosion. This study aimed to (i) monitor soil erosion hazard over 48 year- long precipitation time series, (ii) determine if changes exist for informed decision making and (iii) provide management stakeholders with relevant and scientifically sound information for effective proactive disaster management. Fourier index data was computed annually, and for four seasons; spring, summer, autumn and winter, for the years 1967 to 2014. Each series was fitted to a suitable distribution and then, based on the fitted distributions, marginal probabilities were computed according Fournier index categories: <20, 20-40, 40-60, 60-80, 80-100 and >100. After some exploratory analyses trying out various distributions, the following distributions were fitted to the data: Gumbel, three-parameter Weibull and three- parameter log-normal. Q-Q plots were generated to assess the fit of the various distributions. Overall, the three-parameter log-normal distribution provided at least a reasonably good fit to all five data sets. In order to have a model (distribution) that fits reasonably well to all five data sets (rather than fitting possibly different distributions to the five data sets), and to have a model with a reasonably simple functional form, the three-parameter log-normal distribution was chosen. Based on the fitted three-parameter log-normal distributions the required probabilities were calculated. The greatest soil loss was detected in summer and autumn with 89% and 66% respectively. However, no significant loss of soil was found on annual basis.