The first paper of this series identified several parameters to which predictions of the CREAMS erosion model were sensitive. Two of these, soil erodibility parameters nbov and K, cannot be measured directly. Therefore, this paper reports procedures for deriving nbov and K for the CREAMS erosion model from rainulator data. The procedures identify confidence regions for best fit combinations of nbov and K. Fairly specific values of nbov were obtained, but a wide range of K values provided similar near-optimal predictions. Optimum regions of nbov and K for plots where rilling was the dominant erosion process were significantly different to those for plots dominated by rain-flow erosion for two of the three soils studied. The parameters derived from rainulator data were used to predict soil losses for field catchments on two clay soils, for conditions where surface cover by stubble was 110%. Measured rainfall and runoff were used as inputs to the model. For an Irving clay site, the model gave excellent prediction of field soil losses. For a Moola clay site, the larger soil losses were considerably underpredicted, because resistance to rilling on the rainulator plots resulted in lower K values than required for erosion events where rilling is fully developed. The need for parameters to be derived from data reflecting the erosion processes of importance to field soil loss is illustrated. Nonetheless, this study shows that parameters for the CREAMS model can be obtained satisfactorily from rainulator data, provided that the erosion processes studied are relevant, and that there is sufficient replication within the rainulator studies.
Storm-wise sediment yield prediction using hillslope erosion model in semi-arid abundant lands