Spatial Interpolation of Daily Meteorological Data Using Kriging to Predict DailyRainfall in North-Western Europe

摘要

The spatial and temporal variability of precipitation is very complex. Littleexperience exists on interpolation of daily rainfall data at sub-continent scale. Daily rainfall has the character of a regionalized and stochastic variable. Kriging is an appropriate technique to provide reliable interpolation of regionalized variables. Daily rainfall data on five specific days from 55 weather stations in North-Western Europe were used to assess the feasibility of kriging. Semivariograms were made and a Gaussian model was fitted through the semivariogram values. The derived model parameters were used for kriging. The prediction errors and variances per weather station and for total data sets were determined by crossvalidation. The coastal weather stations from the data set were excluded when examining the disturbing effect of the transition between sea and land on the precipitation pattern. The results of the crossvalidation tests showed that the kriging models for the five specific days were suitable. This resulted in a more reliable interpolation. Furthermore, maps with isohyethes were generated for an area of 750 by 600 squared kilometers. (Copyright (c) 1991 DLO The Winand Staring Centre for Integrated Land, Soil and Water Research.)