Comparison of interpolation methods for estimating spatial distribution of precipitation in Ontario, Canada

ABSTRACT In this study, different interpolation techniques in a geographical information system ( GIS ) environment are analysed and compared for estimating the spatial distribution of precipitation in the province of Ontario, Canada. A high‐resolution regional climate modelling system [Providing Regional Climates for Impacts Studies ( PRECIS )] is used to simulate the present (1961–1990) and future (2071–2100) precipitation events for 12 meteorological stations over Ontario. The results verify that for the present case PRECIS simulates well the precipitation events when compared with observed data. The future precipitation events can be projected after the validation of PRECIS . Six interpolation methods are then used to generate spatial distribution of precipitation based on the projections of future precipitation of 12 meteorological stations; they include inverse distance weighting ( IDW ), global polynomial interpolation ( GPI ), local polynomial interpolation ( LPI ), radial basis functions ( RBF ), ordinary kriging ( OK ), and universal kriging ( UK ). Cross‐validation is applied to evaluate the accuracy of interpolation methods in terms of the root mean square error ( RMSE ). The results indicate that LPI is the optimal method with the least RMSE for interpolating the PRECIS precipitation. LPI is then used to analyse spatial variations of the average annual precipitation for the period of 2071–2100 over Ontario.