Comparison of Published and Proposed Models for Estimation of Soil Erodibility in a Semiarid Region

As the in-situ determination of the soil erodibility factor is in need of elaborate setting up and observation for lengthy periods of time, the use of pedotransfer functions is a good alternative for measurement at a large scale and is highly suitable for soil mapping. Therefore, a study was initiated to select the best model or index for estimating soil erodibility (K) based on soil properties under different land uses over the southern and middle parts of Erbil province. To achieve this goal, soil samples were obtained from the top 0.2 m at 30 locations within the study area. The most commonly used models and a proposed model were compared and ranked based on their power of predictability. The results indicated that 67% of the estimated soil erodibility by the reference method fell in the moderate and fairly high classes. It is also apparent from the displayed results that more than 73% of sampling sites have a dispersion ratio of more than 15% thereby indicating that the majority of the soils are highly vulnerable to erosion. Among soil erodibility indices, only Critical Level of Organic Matter (CLOM) exhibits the strongest correlation with K (r = 0.01). It was also observed that the truncated Wischmeier and Smith model has the highest predictability power for estimating K, followed by the proposed multiple linear models, while the Shirazi model exhibited the poorest performance. Further, the soil erodibility estimated by K6 and K2 exhibited better harmony with land use variation than that of the remaining models. It was also noticed that K gained the highest value at the north and tends to decrease from north to south.