Seasonal Precipitation Forecasting Based on the Teleconnection with Weather Signals in Yazd Synoptic Station

Precipitation forecasting has important role in water resource management especially in arid regions of Iran. This study aims to explore the relationships between the seasonal precipitation and weather signals such as NINO’s SST including NINO1+2، NINO3، NINO4 و NINO3.4 and SOI as well as MEI and NAO. The correlation analysis was performed in two states involving the correlation analysis of weather signals with one year lag in seasonal precipitations and the correlation analysis without the lag. Also, precipitation forecasting was performed through using partial least squares regression (PLSR). Results showed that MEI, SOI, NINO1+2، NINO3 and NINO3.4 have the most correlations with winter seasonal precipitation when the one year lag is performed. The most correlation refers to NINO1+2 equal to +0.68. This value for the SOI is -0.61 which exhibited the inverse correlation of winter precipitation with SOI in the past year. The time series without the lag showed the most correlation between the summer and autumn NAO and winter precipitation of the same year. Also, results indicated the acceptable performance of PLSR for precipitation forecasting. With the one year lag the winter, spring, summer and autumn precipitations were estimated with the RMSE equal to 12, 9.9, 0.85 and 6.2 mm, respectively. Also, the Nash–Sutcliffe (NS) model efficiency coefficient for the mentioned seasons is 0.69, 0.22, 0.2 and 0.72, respectively. The R correlation coefficients for these time series were equal to 0.83, 0.46, 0.45 and 0.85, respectively. In general, precipitation was predicted more accurately in the cold seasons. The development and use of such prediction models could make water resource management programs more successful.