On the multiscale dynamics of tropical rainfall

The Tropical Rainfall Measuring Mission (TRMM) is providing some of the first detailed and comprehensive datasets on the four dimensional distribution of rainfall and latent heating over vastly undersampled oceanic and tropical continental regimes. It has enabled study of variation of the timing of heaviest rainfall - particularly nocturnal intensification of large mesoscale convective systems over the oceans, and diurnal intensification of orographically and sea-breeze forced systems over land. Current approaches for detecting strong diurnal patterns are largely based on daily mean rainfall profile. They could yield erroneous identifications when a daily mean rainfall profile has large variations. To more reliably infer regions where rainfall has strong diurnal patterns, we propose two new approaches. One has explicitly incorporated the standard deviations of the daily mean rainfall profiles. The other is based on Fourier transform and allows for an effective significance test of a diurnal pattern. With these approaches, we have exhaustively identified all the regions with strong diurnal patterns from TRMM rainfall data.

We have further extended our analysis of rainfall dynamics to time scales from a few hours to a week using concepts of fractal theory, especially the notion of persistent correlations based on the Hurst parameter. In fact, in some geographic locations, the fractal scaling holds to time scales as long as a few months, or even longer. The temporal-spatial variations of the Hurst parameter enables a more comprehensive examination of tropical rainfall dynamics, especially the connections between the south Asian summer monsoon systems to events of low-frequency climate variations such as the ENSO and the Indian Ocean Dipole (IOD).