Rainfall Network Optimization using Information Entropy and Fire Fly Algorithm Case Study: East Basin of Urmia Lake

The first step of the implementation of water projects is recognizing the characteristics of rainfall, and awareness of the weather and climate of the region. Because of the spatial and time variations of rainfall, it requires much denser network to supervise compared to the other meteorological factors. Therefore, optimal localization of the stations is very necessary. Accordingly, monthly rainfall data of rain gauge stations were collected at the basin and adjacent areas first. The primary rain gauge network was created by considering the period with the largest number of the stations, the most statistical The first step of the implementation of water projects is recognizing the characteristics of rainfall and awareness of the weather and climate of the region. Because of the spatial and time variations of rainfall, it requires much denser network to supervise compared to the other meteorological factors. Therefore, optimal localization of the stations is very necessary. Accordingly, monthly rainfall data of rain gauge stations were collected at the basin and adjacent areas first. The primary rain gauge network was created by considering the period with the largest number of the stations, the most statistical years, and the lowest rebuilding ratio. Then all of time series data were analyzed by statistical analysis including normal and homogeneous tests and stations with errors and non-homogeneous statistics were removed or modified from the total stations. According to the results of normal tests, annual rainfall in 5 stations, due to their smaller amount of tests’ values toward their critical value, does not follow normal distribution. Also, the results of homogeneous tests indicated that 91.7% of selected stations in the basin were confirmed in term of the accuracy of the recorded rainfall data and could be used in hydrological analysis or water resources after this. With the spatial analysis of the precipitation entropy, the amount of information transmission entropy was calculated in the basin area and was considered as a criterion in determining the points with the potential of establishing a new station. The results indicate that the six stations of Shabestar, Sharafkhaneh, Zarnagh-Heris, Harzandat, Kalibar and Ghoshchi-sarab, while having low ranks in the network, are in critical condition and are the weakest stations in the basin. On the other hand, the stations of Saeedabad, Bashsizovjan, Maragheh, Khormazard and Shirin Kandy have the highest rank among other stations, and are five important studied stations in the basin that produce the most useful information in the network. In order to perform the optimization process, an objective function for the whole basin was determined and then the Firefly algorithm was used to obtain the best localization for rain gauge stations. The best localization is obtained by adding 9 stations after checking the results.