Analysis of interpolation methods to map the long-term annual precipitation spatial variability for the Republic of Bashkortostan, Russian Federation

摘要

In meteorological modeling it is very important to have accurate data about the amount of precipitation over particular territory. This data can be obtained by the interpolation of the point sources. In this case several interpolation methods (inverse distance weighting, ordinary kriging, geo-regression and co-kriging) are used to map average long-term precipitation over Republic of Bashkortostan, a region of the Russian Federation. Data of more than 30 years of observations from 41 stations have been processed. Several variogram models for ordinary kriging and co-kriging methods have also been fitted. It was found out that no variogram in ordinary kriging method fits best to the observed amounts, the closest one is linear model. In order to make the geo-regression method, elevation of each station was taken. The correlation between elevation and precipitation of all points was not good enough, so cluster analysis was carried out to find out points with good correlation. Results of cluster analysis show small correlation between elevation and precipitation, despite that fact, results of geo-regression (when divided on 4 clusters) show better results, than IDW or kriging. Among all methods co-kriging with linear modeled variogram and which previously was divided by 4 clusters shows the closest result to observed amounts.