Entropy-based assessment and zoning of rainfall distribution

Rainfall distribution has become highly erratic due to climate change and intensive human activities. Hence, the estimation of rainfall distribution has an extraordinary significance in understanding the hydrological cycle and is crucial for water resources management. This paper presents a study on the large-scale spatial rainfall distribution in the Pearl River Basin of China using the information entropy theory and the fuzzy cluster analysis. The Directional Information Transfer Index (DITI) was used to describe the similarity between rainfall gaging stations, and the fuzzy cluster analysis was utilized to classify rainfall gaging stations into distribution zones with the proximity relation defined by the DITI. This research shows that the DITI integrates the rainfall feature at respective stations and the mutual influences among them. Further, the DITI-based fuzzy cluster analysis has a great advantage over the conventional pattern recognition method. Considering the unique temporal and spatial distribution characteristics, the DITI-based model combined with the fuzzy cluster analysis method provided more accurate classification of the rainfall distribution zones. Based on the monthly average rainfall data from 1959 to 2009 at 62 stations, the rainfall distribution in the Pearl River Basin is classified into 10 zones with their unique temporal and spatial distribution characteristics. The correct classification of rainfall distribution zones is crucial for the management and allocation of water resources in the Pearl River Delta to meet the increasing demand of domestic and industrial usage not only within the basin but also as a complementary source for Hong Kong.