Autor:innen:
Ahsan Raza | University of Bonn, crop production, INRES, Katzenburgweg 5, 53115, Bonn, Germany | Germany
Dr. Murilo dos Santos Vianna | University of Bonn, crop production, INRES, Katzenburgweg 5, 53115, Bonn, Germany | Germany
Dr. Seyed Hamid Ahmadi | University of Bonn, crop production, INRES, Katzenburgweg 5, 53115, Bonn, Germany | Germany
Dr. Muhammad Habib-ur-Rahman | University of Bonn, crop production, INRES, Katzenburgweg 5, 53115, Bonn, Germany | Germany
Dr. Thomas Gaiser | University of Bonn, crop production, INRES, Katzenburgweg 5, 53115, Bonn, Germany | Germany
The coupling of soil water erosion models and the dynamics of crop growth and soil water balance in the agricultural landscape has been studied over several decades. To date, however, the accuracy of soil erosion models in agroecosystems with heterogeneous field conditions remains a topic of debate. Numerous questions remain to be addressed, especially related to the uncertainties caused by the approaches to model coupled processes, such as soil water fluxes, crop growth, and soil erosion. In this study, we investigate two widely used methods (Freebairn and Rose) to represent soil erosion and coupled them with a process-based crop model within the SIMPLACE framework. Well-distributed spatiotemporal measurements of the soil and plant dynamics were taken in a heterogeneous field to calibrate and evaluate such model solutions. The accuracy of these coupled models was also compared to a statistical model developed for the same field. Besides model accuracy, soil erosion process representation, data, and calibration requirements were also studied. The simulations of water erosion with dynamic Freebairn and Rose models were influenced by the performance of runoff and crop growth models. However, a pronounced difference was found between modeled and measured soil erosion when these predictions were made with an uncalibrated runoff model. Hence, our results highlighted that large uncertainties in soil erosion modeling were associated with improper performance of the runoff model. Among selected models in the validation phase, the Freebairn model had the highest accuracy of sediment yield predictions (NSE = 0.71, RMSE = 0.69 t ha-1 d-1) than Rose model (RMSE=0.83 t ha-1 d-1) providing insight into the selection of parameter values and calibration for these models. These findings give considerable confidence in the model structure. Therefore, it can be concluded that both Freebairn and Rose models can be used as tools to predict sediment yield within the SIMPLACE framework. However, further improvements of soil erosion models should focus on enhancing the data quality for model applications and improving the representation of these models in terms of their scales and objectives.