%0 Journal Article
%T Analysis of the spatial distribution of snow water equivalent in the watersheds of West Azerbaijan
%J Iranian journal of Ecohydrology
%I Faculty of New Sciences and Technologies, University of Tehran
%Z 2423-6098
%A Yousefi, Kamran
%A Dusti Rezaei, Mehrang
%A Ahmadaali, Jamal
%D 2023
%\ 08/01/2023
%V 10
%N 1
%P 77-91
%! Analysis of the spatial distribution of snow water equivalent in the watersheds of West Azerbaijan
%K Snow water equivalent
%K multivariate regression
%K Artificial Neural Network
%R 10.22059/ije.2023.353231.1706
%X A significant part of the precipitation in the upstream areas of the watersheds of West Azerbaijan province is in the form of snow, especially in the cold months of the year. Analyzing the data of these sources is very important for the optimal management of reservoir dams in the region, especially in spring, which is the season of snow melting in the respective basins. In common traditional methods for estimating water equivalent of snow storage, the height gradient relationship (linear regression relationship between water equivalent of snow and the height of snow measuring points from the surface of the oceans) is used, which sometimes due to lack of proper fit, some points are removed from the calculations. In this study, for the data of 65 snow measuring stations in snowy months of the water year 2020-2021, instead of the relation of altitude gradient, the four-variable regression of water equivalent to altitude, longitude and latitude (in the UTM coordinate system) was used. It improved the regression characteristics (an increase of more than 3 times the A.R.S., i.e. the square of the adjusted correlation coefficient, or the coefficient of determination, and a decrease of more than 10,000 times the significance level, i.e. the alpha of the correlation coefficient). Also, in order to further investigate, the simulation of relationships with artificial neural networks (selected network: three-layer perceptron with 3, 8 and 1 neurons in the input, middle and output layers) was used; So that the correlation coefficient of the estimated data with the available observations for the selected artificial neural network model was 0.97.
%U https://ije.ut.ac.ir/article_91854_7465b53b660549c24b71eecf92652ab6.pdf