ICTS

Indian Monsoon affects the lives of over a billion people. For over a century, scientists have been trying to understand the physics behind it, and make predictions. In this work, we make use of the availability of high-resolution data and advances in Data Science. We consider monsoon rainfall as a spatio-temporal random process, and introduce a random field of discrete variables through which we identify large-scale coherent spatio-temporal characteristics of the underlying physical process. We find that daily spatial distribution of rainfall over the Indian landmass during monsoon can be represented using a few canonical patterns. These patterns and their inter-relations give us a clear picture of the intra-seasonal oscillations of monsoon – its onset, "active spells", "break spells" and finally retreat. We also identify regions where daily rainfall characteristics during monsoon are near-identical, and this allows us to partition the landmass into homogeneous zones. We explore possibilities predictions at short/medium spatial and temporal scales.