and Multivariate Regression in Prioritizing Climate Factors Affecting Runoff Generation in Research Plots: A Case Study of Sangane Station, Khorasan Razavi

Estimating runoff in watersheds is of great importance in water resources management.The aim of this study was to compare the efficiency of Artificial Neural Network and Multivariate regression in prioritizing climate factors affecting runoff generation in research plots (areas of 10, 20, 30 and 40 m2) of Soil Conservation Research Database of Sanganeh. Sanganeh has an area of 50 hectares and is located in Khorasan Razavi province. For this purpose, the data of rainfall – runoff of 72 events was used in 32 plots. The multivariate regression relationships were created between the input variables (rainfall amount and intensity) and the height of the surface runoff collected in the selected output plot (10, 20, 30 and 40 2), plots with the same conditions on a slope, plots on different slopes and finally, the total plots existing in the area. The results were indicative of a significant and positive effect of climate variables on output runoff volume. The study showed a greater impact of rainfall variables than rainfall intensity in the spatial scales under study. In addition, according to the parameter coefficient and Root Mean Square Error (R2, RMSE), it can be concluded that multi-layer perceptron artificial neural network models are more accurate than multivariate regression models.