ANALYZING THE SPATIAL DISTRIBUTION OF ANNUAL PRECIPITATION ACROSS KAZAKHSTAN THROUGH DIFFERENT INTERPOLATION TECHNIQUES

Various climatic factors, including temperature, precipitation, humidity, and wind, play a crucial role in plant growth, yield, and quality. Therefore, understanding a region’s climate characteristics is essential before beginning agricultural activities to determine suitable crops for cultivation. This study aims to map the spatial distribution of annual total precipitation in Kazakhstan by applying different interpolation methods to long-term (20-year) precipitation data from 80 meteorological stations. Precipitation data from 2000 to 2019 were obtained from the MERRA-2 Meteorological Re-Analysis database. To identify the most accurate interpolation method, seven techniques were tested: Inverse Distance Weighting (IDW), Radial Basis Functions (RBF), Local Polynomial Interpolation (LPI), Ordinary Kriging (OK), Bayesian Empirical Kriging (BEK), CoKriging (COK), and Diffusion Kernel. Correlation coefficients and Root Mean Square Error (RMSE) values were used as criteria to evaluate each method's accuracy. Results showed RMSE values of 35.568 for IDW, 30.777 for RBF, 30.031 for LPI, 25.751 for OK, 28.767 for BEK, 45.9 for COK, and 25.379 for Diffusion Kernel. The Diffusion Kernel method, with the lowest RMSE and highest correlation coefficient, was found to be the most suitable for creating the areal distribution map of precipitation in Kazakhstan. Selecting the best interpolation method is essential as variations among methods can lead to different spatial distributions. Using the most appropriate interpolation method for each point dataset allows for more accurate mapping of water, soil, and climate data, leading to improved resource management.