Regionalization of Indian monsoon rainfall and long-term variability signals

Abstract

Large regions in India with homogeneous variability of monsoon rainfall are identified with the help of principal component analysis. Four major regions, called principal regions, are first demarcated. Even though the variance explained by the first PC is only 21 per cent, it can be used to organize 45 per cent of the surface area into a coherent region. A novel sequential methodology wherein at every step the grouped stations are sieved out is also proposed. This helps in identifying 10 sequentially homogeneous regions accounting for 91 per cent of the total area under consideration. The four principal regional rainfall series reflect the interannual variability originally present in the data set. These are mutually uncorrelated and hence represent four independent monsoon anomaly patterns also. The first and the fourth principal regions contain temporal signals well discriminated from white noise.