1 گروه علوم خاک ، دانشکده کشاورزی ، دانشگاه تبریز
2 گروه سنجش از دور و GIS ، دانشگاه تبریز
عنوان مقاله [English]
The wind erosion occurs when wind speed exceeds the soil erosion threshold and plants or their residues, surface roughness, or other obstacles do not protect the soil surface. Also, wind erodibility is one of the most important determining parameters of wind erosion under certain climatic conditions. The main objective of this research was mapping of soil erodibility through empirical relationship between satellite imagery and physicochemical properties and estimation of soil erosion using a comprehensive assessment model on the east shore of the Urmia Lake. For this research work soil sampling carried out in 153 points of three elevation classes (1271-1273, 1273-1275 and 1275-1278 meters) and 4 supervised classification methods such as, support vector machine (SVM), maximum likelihood classification (MLC), minimum distance and artificial neural network (ANN) were used for classifying and mapping of soil erodibility. Soil physic-chemical properties measured and 26 samples of them randomly were selected for wind erodibility measurement in an artificial wind tunnel. Wind tunnel experiments at a distance of 20 cm from the tunnel floor, revealed wind erodibility of 2.92 ((g m-2 min-1)/ (m s-1)). Also, stepwise regression results showed that among the physic-chemical properties of soils, erodible fraction was the most important soil property which used in estimating erodibility and has a positive correlation with soil erodibility. The mean weight of aggregate diameter had negative correlation with soil erodibility and no relationship was found between soil chemical properties and erodibility. Among the four supervised classification methods, the ANN has a higher capability in classifying and mapping of erodibility. Finally, the results showed that the overall classification accuracy is 57.1%.