Identification of rainfall patterns over the Valley of Mexico

The paper deals with the identification of rainfall intensity patterns and spatial distribution of these intensities for the region of the Valley of Mexico. The employed data consists of rainfall values measured by the rainfall gauge network, spread over a 40x50 km2 region. As first step, storm intensity classes were established using a clustering algorithm. Second, storms from 224 days from the rainy period selected over the years 1993 till 2005 were recoded, at station level, with the intensity class values. The obtained matrix (with days per line and station per column) was clustered under two different conditions. First, clustering per lines helped to identify rainfall types as characterized by their spatial intensity distribution. We give a characterization of the identified rainfall types. Second, by clustering per columns we found stations having similar behavior from the intensity point of view. The validation of the obtained clusters is based on three commonly used indexes: connectivity, Dunn index and silhouette value. The results give a good characterization of the rainfall events over the Valley of Mexico; allow the extrapolation of values to regions without measuring instruments and can be helpful in forecast and alert systems.