Determination of effective factors on soil loss caused by gullies using drone, a case study of Fatah Abad, Buin Zahra

Recognition of influential factors is required for management of gully erosion. The purpose of this study was gully classification and determining the most effective variables on the soil loss in the Fath Abad, Buin Zahra in Qazvin Province. Twenty four gullies were selected, surveyed and recorded by GPS, and their morphologic characteristics including length, width and depth were measured. Surface soil samples were taken from headcut and were analyzed for texture, EC, pH, organic matter and some anions and cations. Indices related to the form of watershed were extracted from the digital elevation model which derived from UAV image processing. The Ward method was used for cluster analysis of gullies and their properties. Factor analysis, the correlation between factors and multivariate regression were performed to determine important and effective factors on the formation of gullies in SPSS. The results of the research indicated that gullies have vertical headcuts, trapezoidal cross-sections and linear general plans. The average depth of headcuts, depth at 50% of the gully length and the average gully length were 0.4, 0.6 and 25 m, respectively. Soil texture was loam and average soil texture components including sand, silt and clay were 30, 46 and 24%, respectively. The mean EC and pH of the samples was 8 dSm-1 and 8.4 respectively. Linear correlation between the total volume of soil loss and other gully characteristics showed that the average cross section, average gully lower width and upper width, sand content, average gully depth and total length have meaningful correlation coefficients greater than 0.6 at 1% percent level. Also, multivariate regression results showed that the most effective factors on soil loss by gully erosion were saturation percentage, sand content, roundness coefficient, basin length and area respectively, which had more effect on soil loss prediction, which entered into the loss prediction equation.