Developing a Weather Generator Model to Preserve Spatial Correlations between Neighboring Stations

Considering the importance of preserving spatial correlation between neighboring stations in many of the studies in the fields of agriculture and water resources on a daily time basis, in this research, a weather generator model (WG) was developed to simulated climatic variables in neighboring stations while preserving spatial correlations between these stations. This model uses an extended Markov model to generate precipitation occurrence series which is capable of simulating spatial correlations between neighboring stations with an acceptable performance. In order to simulate precipitation amounts on wet days and other climatic variables, a nonparametric algorithm was proposed. The performance of this algorithm was assessed in relation with the generation of daily mean and standard deviation, daily correlation and lag-1 autocorrelation of climatic variables and also spatial correlation between neighboring stations using Coefficient of Determination (R2), Standardized Root Mean Square Error (SRMSE) and Standardized Mean Absolute Error (SMAE) statistics. The results showed that this model is capable in reproducing statistical properties of historical time series with an acceptable accuracy, while the model underestimated the values of lag-1 autocorrelation coefficients. Moreover, climate change scenarios can be simulated by modifying model parameters while preserving spatial correlations between neighboring stations