MODELS TO PREDICT SLOPE LENGTH FROM OTHER WATERSHED ATTRIBUTES

Abstract

Soil erosion by water is an extensive and increasing problem worldwide. Albeit, this problem has been recognized as a significant hazard in Iraq, yet the number of studies on this topic is very limited. Most of the models used for estimating soil erosion contain parameters for slope length factor (LS). A major constraint is the difficulty in extracting the LS factor. Accordingly, the current study was initiated with the main objective of deriving models to predict the slope length from relatively easy to measure basin characteristics with a reasonable accuracy. To achieve the above objective, standard methodologies were employed to describe 30 main basins with the upper part of Iraq in terms linear, areal and relief morphometric parameters. The majority of the delineated watersheds were characterized by having high slope lengths indicating lower drainage density and higher erosion rate. Linear and non-linear least squares techniques were applied to predict the slope length from other basin characteristics. Different indicators were used to test the performance of the proposed models and the approach was validated using K-fold procedure at independent basins. The results indicated that the 4-parameter regression model outperformed the remaining models of watershed slope length. The regressors of this model are bifurcation ratio, perimeter, and basin length and slope gradient.