Evaluation and optimization of REGEM model to estimate the gully erosion distribution in three watershed of Ardabil province

Extended Abstract Introduction: Gully erosion is the most severe type of water erosion, which causes large damage in the watersheds. To determine the risk of this type of erosion, it is necessary to estimate its severity and specify the distribution of the risk of erosion in the watershed. One of the models used to estimate the gully erosion is REGEM model, but accurate estimation of erosion requires adaptation of the model to regional conditions. In this study, from REGEM and Adapted REGEM (AREGEM) models, they were used to estimate the gully erosion severity of Ardebil province. Also, the distribution of the two models' outputs where mapped in the three watershed and compared to the chosen ones of optimal model. Methodology: In this research, three watersheds were selected in Orta Dagh, Mollahammad and Sarcham of Ardebil province. Respectively 28, 33 and 20 small gullies were selected from three watersheds. In order to estimate the gully erosion with the AREGEM model, the equations of REGEM base model were used. For this purpose, regression relations between the measured values of the gully dimensions and the estimated values by REGEM model were established and coefficients of equations were corrected. Using the equations of REGEM and AREGEM models, three outputs of the model including shear stress , soil erosion factor (Kd) and gully volume (V) for selecting gullies were calculated and with using of IDW interpolation method, obtained the distribution of these three factors in the three watersheds. The outputs of the models were considered as an indicator of gully erosion severity and the watersheds were divided into four levels of erosion intensity. Finally, using two indicators of density ratio (Dr) and quality sum (Qs), the erosion classes obtained from the two models were evaluated and compared. Results: The results showed that using the REGEM model, the average volume of gullies were in the Orta Dagh region 15.6, Mollahammad 17.2 and Sarcham 16.3 cubic meters while the AREGEM model was 16.1, 17.8 and 16.8 Cubic meter respectively. For the three areas of Orta Dagh, Mollahammad and Sarcham, the average of shear stress with REGEM model was 31.5, 32.2 and 24.8, respectively, and with AREGEM model 32.1, 33.3 and 20.6 respectively. The soil erosion factor (Kd) with REGEM model was 62.1%, 63.3% and 69.1% for the three regions, with AREGEM 1/62, 64.2% and 70.1%, respectively. In general, there is not much difference between the mean values of the outputs obtained from the two models, but the distribution map of the outputs in the two models differed greatly. Visually, in all three regions, the estimated erosion volume by AREGEM model has better compliance with the gully density of watersheds, while estimates of the REGEM base model in all three regions were not a good predict of gullies distribution. Also, the comparing of shear stress and soil erosion factors maps, showed similar results in three regions. Therefore, in order to have a more accurate comparing of the two models' outputs distribution map, the semi-quantitative index of the density ratio (Dr) and the quantitative index of quality sum (Qs) were used. Generally, in all three areas, for maps derived from the AREGEM model, with the increase in the classes of all three outputs of the model, the index of the density ratio (Dr) had an incremental trend, while for the REGEM model this trend was not regular. This indicates that as the erosion rate estimated by the AREGEM outputs increases, the density of the gully in those classes also increases, so there is a greater relation between the estimated erosion values with the AREGEM model and the sensitive classes to gully erosion. Also, the use of the Qs index showed that, the AREGEM model has a higher zoning quality and the ability of this model to determine the classes of sensitivity to gully erosion is more abundant. Discussion & Conclusions: The use of the REGEM base model to determine the distribution of gully erosion is not accurate and this model needs to be modified and applied according to the conditions of the study area. For this purpose, the Adapted REGEM (AREGEM) model was presented and used. Also it has been observed that between three outputs obtained from the model, the use of gully volume is more effective in determining the exact distribution of gully erosion. Among the three studied regions, the AREGEM model in the Orta Dagh region has the highest efficiency for determining the distribution of gully erosion and then, respectively in the Mollahammad and Sarcham.