Approaches for Regionalization of Precipitation Climates in the Context of Global Climate Change Monitoring

Abstract

Two precipitation regionalization methodologies are used and evaluated. In one, precipitation in the tropics, consisting of 34 stations, each considered to be representative of a 20-degree square region, was analyzed using the time scale identification techniques of singular spectrum analysis and the multitaper method of spectrum estimation. A second approach consisted of defining a spatial domain based on the distribution of the principal vegetation types dominant in that region in order to develop area-mean indices from the climate stations located within it. The geographical domain of this second approach was the general region of the North American Great Plains.

The set of tropical stations with significant variance in the 3–7 yr ENSO time scale were all located in areas where the ENSO phenomenon significantly modulates the amount of seasonal precipitation over time. However, even in areas of the western tropical Pacific where an ENSO signal in precipitation is shown to be of the same sign, correlation between stations can be relatively low. In the specific example discussed here, the association between Darwin, Australia and the island station of Apia to the east was only r=0.35. In the region of the U.S. Great Plains, however, strong decadal variations were evident in a number/)f grassland regions, encompassing a rather large area. Additional tests for regions with different vegetation types are needed to evaluate the validity of using dominant vegetation type as a regionalization tool.