Application of artificial intelligence in simulation of spatial distribution of snow density in semi-arid regions: A case studyof upstream regions of Yazd-Ardakan plain

Water crisis is one of the most important problems in arid and semi-arid regions, so snowfall occurring in upstream parts of mountainous basins has an enormous role in hydrological balance. In this paper, the spatial distribution of snow density in Sakhvid, Yazd has been studied using an artificial neural network. Snow density is an important parameter for assessment of water resources in mountain basins, and, with thecorresponding data and snow depth, snow water equivalent values can be calculated. For this purpose, 216 in-situ snow density data were measured using the Mt. Rose sampler. Then, using SGA-GIS software, 32 geo-morphometric parameters were calculated from DEM. The best network was 1-9-32 with a multilayer perceptron model, the back-propagation algorithm,the sigmoid activation function, and a linear output. In order to evaluatethe network, the ANN correlation coefficient and the root mean square error (RMSE) were used. The results showed that the correlation coefficient and RMSE of the observed and estimated data were 86 percent and 5.1 respectively. So, application of artificial intelligent can simulate the spatial distribution of snow density very well.