Spatial correlation of radar and gauge precipitation data in high temporal resolution

摘要

A multi-sites precipitation time series generator for engineering designs iscurrently being developed. The objective is to generate several time series'simultaneously with correct inter-station relationships. Therefore, a modelto estimate correlation between stations for arbitrary points in a projectarea is needed, using rain gauge data as well as radar data.Two methods are applied to compare the spatial behaviour of precipitation inboth the rain gauge data and the radar data. The first approach is tocalculate precipitation intensities from radar reflectivity and use it asgauge data. The results show that the spatial structure in both data sets issimilar, but cross correlation varies too much to use radar derived spatialcorrelation to describe gauge inter-station relationship. Thus, a secondapproach was tested to account for the differences in the spatial correlationassociated to the distribution. Using the indicator time series, crosscorrelations for different quantiles were calculated from both the rain gaugeand radar data. This approach shows that cross correlation varies dependingon the chosen quantile. In the lower quantiles, the correlation is verysimilar in rain gauge and radar data, hence a transfer is possible. Thisinsight is useful to derive cross correlations of rain gauges from radarimages. Correlation data for rain gauges thus obtained contains all theinformation about heterogeneity and anisotropy of the spatial structure ofrainfall, which is in the radar data.