Evaluation of different interpolation methods for NCEP/NCAR temperature data over the 2nd order watersheds in Iran

Temperature is one the most important parameters in climatology and hydrological investigations. The main purpose of this article is enhancing the resolution of the gridded annual temperature data of the NCEP/NCAR reanalysis dataset to a reasonable level. In this data base، temperature is available at the resolution of 2. 5 degree. For regional studies in Iran، due to the diversity of topographical features، the precision of the available resolution is not enough. In this research، it is intended to replace a finer resolution of 1 degree by applying the interpolation methods embedded in the MATLAB. There are 4 different methods for interpolation in the MATLAB: Linear، Nearest Neighborhood، Spline and Natural Nearest Neighborhood. By applying these methods over the 30 second order watersheds across Iran، a network of 1*1 degree resolution was obtained from the 2. 5*2. 5 degree resolution. To examine the precision of the simulated data، the annual average temperature for each sub-watershed was calculated via the long term data of the 47 synoptic stations distributed unevenly across the country. Using RMSE،MBE، Nash-Sutcliffe and CRM statistics the Nearest Neighborhood and Spline methods was identified as the best methods for interpolation of the NCEP/NCAR reanalysis temperature data. The main result of this research is the generation of gridded temperature data for hydro-climatological studies، introducing a method for its updating and henceforth، omitting the labor of gathering، testing and using local data.