Meteorological drought analysis using artificial neural networks for Bursa city, Turkey

Abstract

Climate change is one of the most important environmental events of recent years. Floods and droughts may occur more frequently with climate change. Droughts could be classified as meteorological, hydrological or agricultural. When meteorological drought appears in a region, agricultural and hydrological droughts follow. In this study, the standardized precipitation index (SPI) was applied for meteorological drought analysis in Bursa City-Turkey. Analyses were performed on 3-, 6-, 9- and 12-month-long data sets. According to the analyses most of the percentages (67-72%) for all SPI values (3, 6, 9, 12 months) was in the near normal class close to Marmara Region-Turkey. Also meteorological variables and SPI-12 values were simulated with ANN models and had different structures. It was found that R and MSE values calculated were in the acceptable ranges and rising of hidden layers and input numbers in the model structures was ensured for more efficient model run. Modelling the precipitation from the meteorological parameters (Model 2) was possible with some error. Therefore simulations of SPI-12 and soil temperatures were indicative to meteorological and agricultural droughts. Comparison of the observed values and the modelling results showed a better agreement with SPI-12 and soil temperature parameters. Drought predictions made by ANN models would be useful for local administrations and water resources planners and would be extremely important for drought risk management.