Despite the significance uncertainty in the variables of erosion prediction models, spatial and temporal variability is seldom considered in soil conservation analyses and planning. The consequences of this simplification are misleading soil loss predictions, with consequences for the agricultural and environmental sustainability of agricultural fields. The objective of this paper was (a) to estimate the effect of spatial and temporal variability of the USLE factors on erosion prediction of a homogeneous agricultural field, in the Distrito Federal (Brazil), (b) to incorporate the resulting uncertainty in the analysis, and (c) to identify appropriate mitigation measures. Results indicate that despite the low coefficient of variation of the USLE factors (V = 20 %), the coefficient of variation of the predicted soil loss was 64 %, with a significant propagation of uncertainty in the model. The implications of this uncertainty are that, even when the expected value of soil loss (A) was 19 % smaller than that of the average soil loss tolerance (T), there would be a probability of 43 % that A > T. In order to reduce this risk, it is suggested that the safety margin of the system be increased by the implementation of conservation practices, such as terraces. The probability of failure in this new scenario was only 12 %. The results reinforce the importance of the incorporation of variabilities and uncertainties in erosion modeling and conservation planning processes, using stochastic analyses.
Erosion models; variability; uncertainty; conservation planning
