Abstract/Summary

Indian summer monsoon precipitation is significantly modulated by synoptic-scale tropical low pressure areas (LPAs), the strongest of which are known as monsoon depressions (MDs). Despite their apparent importance, previous studies attempting to constrain the fraction of monsoon precipitation for which such systems are responsible have yielded an unsatisfyingly wide range of estimates. Here, a variant of the DBSCAN algorithm is implemented to identify nontrivial, coherent rainfall structures in TRMM-3B42 precipitation data. Using theoretical considerations and an idealised model, an effective capture radius is computed to be 200 km, providing upper-bound attribution fractions of 57% (17%) for LPAs (MDs) over the monsoon core zone and 44% (12%) over all India. These results are also placed in the context of simpler attribution techniques. A climatology of these clusters suggests that the central Bay of Bengal (BoB) is the region of strongest synoptic organisation. A k-means clustering technique is used to identify four distinct partitions of LPA (and two of MD) track, and their regional contributions to monsoon precipitation are assessed. Most synoptic rainfall over India is attributable to short-lived LPAs originating at the head of the BoB, though longer-lived systems are required to bring rain to west India and east Pakistan. Secondary contributions from systems originating in the Arabian Sea and south BoB are shown to be important for west Pakistan and Sri Lanka respectively. Finally, a database of precipitating-event types is used to show that small-scale deep convection happens independently of MDs, whereas the density of larger-scale convective and stratiform events are sensitive to their presence - justifying the use of a noise-rejecting algorithm.