Rain gauge network design using coupled geostatistical and multivariate techniques

A methodology for the design of a rain gauge network is developed in this study. To the best of the authors’ knowledge, this is the first time a combination of geostatistical tools and factor analysis, along with a clustering technique, has been used to prioritize rain gauge stations in terms of information content over the study area. The whole study area is divided into homogeneous subregions and a conventional variance-based approach is implemented in each subregion to rank rain gauge stations. For this purpose, factor analysis coupled with ordinary block kriging is used to identify the number of homogeneous subregions, and then, ordinary point kriging is used to assign rain gauge stations to each subregion. The developed scheme is quite time-efficient as it is not sensitive to initial guesses on cluster centers, there is no need to specify the number of clusters in advance and, above all, it is highly relevant to the overall objective stipulated in rain gauge network design. The proposed methodology is implemented on real data set in the south west of Iran. The results show that the proposed approach compares well with existing paradigms in rain gauge network design and only six rain gauge stations are required to provide the necessary information. In particular, the measure of network accuracy lies somewhere in between the so called time consuming and more simplified approaches used in rain gauge network design.