A comparison of different interpolation methods using the geographical information system for the production of reference evapotranspiration maps in Turkey

Eight different interpolation methods (inverse distance weighting,spline with tension,thin-plate spline,completely regularized spline,ordinary kriging,simple kriging,universal kriging,and linear regression (lr) model) were comparatively analyzed to determine the spatial distribution of monthly reference evapotranspiration (et0) values calculated using the hargreaves method (et0-hg). to compare the different methods,observation stations were divided into two data groups (58 stations for analysis and 14 stations for validation). the correlation coefficients for all months ranged between 0.68 and 0.90. of all the methods evaluated,the lr model was found to give the optimum results. the highest correlation coefficient was observed with the lr model for all months except march,april,june,and september. uk showed lower correlation coefficients than the lr model for all months except june and september and was found to be the second-best method. secondary data contributed positively to et0-hg interpolation; therefore,certain spatial data was incorporated into the lr model. the relationship between the monthly et0-hg values produced using the most efficient interpolation method,and the values calculated with the penman-monteith method (et0-pm) were also investigated in this study. correlation coefficients between the methods varied between 0.65 and 0.86 for all months. the largest difference between the two methods was observed in june (-2.27 mm day-1),and the smallest difference was seen in february (0.02 mm day-1). in conclusion,hargreaves method was found to be easier to use in calculating et0 for the creation of et0 maps,in which the use of lr was found to be more reliable than other interpolation methods. © 2014,meteorological society of japan.