Nearly a decade after establishing secular trendsin extreme rainfall (Goswami et al., 2006, Science)and offering an explanation for the stability of theIndian monsoon, we shifted our attention from a regional to a global tropical viewpoint of extremes. In particular, the aim was to assess the validity of the oft-quoted “wet-wetter; dry-drier” warming world paradigm. We showed that there is a cyclical nature associated with tropical extreme annual rainfall (Sukhatme and Venugopal, 2015). The most striking result in this work is that when the El Nino Southern Oscillation (ENSO) goes from a ”cold to warm”phase (say 1960s to 1990s), extremes showed a significant decrease; remarkably, going from a ”warmto cold” phase (1990s to the current decade), the extremes showed a near-exact reversal both statistically and geographically. The implication of this finding is that the footprint of natural variability on extreme rainfall is large enough to notbe ignored, and that one has to revisit the idea of attributing all the observed increase in extremes to warming.
Using daily, 1-degree TRMM 3B42 rainfall observations, we statistically analysed and documented the wet and dry spell characteristics of tropical rainfall. Using an Intensity -Duration- Frequency (IDF)-like approach (common in hydrology, but seldom used in meteorology), we found that while both ocean and land regions with high seasonal rainfall accumulation (humid regions; e.g., India, Amazon, Pacific Ocean) show a predominance of 2-4 day wet spells, those regions with low seasonal rainfall accumulation (arid regions; e.g., South Atlantic, South Australia) exhibit a wet spell duration distribution that is essentially exponential in nature, with a peak at 1 day (Ratan and Venugopal, 2013). The behaviour that we observed for wet spells is reversed for the dry spell characteristics. In other words, the main contribution to the non-rainy part of the season comes from 3-4 day dry spells in the arid regions, as opposed to 1-day dry spells in the humid regions. The total rainfall accumulated in each wet spell was also analysed, and we find that the major contribution to seasonal rainfall for arid regions comes from 1-5 day wet spells; however, for humid regions, this contribution comes from wet spells of duration as long as 30 days. Following this, we assessed the role of randomness in determining the observed features. We find that while the 2-4 day mode might arise purely out of chance (especially in a region where it rains for more than ∼70% of the time during the season), what separates reality from randomness is in the contribution of longer duration wet spells (most likely the influence of organised convection) to the observed total number of rainy days or seasonal accumulation. The importance of our findings is particularly relevant in the context of numerical modelling; specifically, the 2-4 day mode or the contribution of longer duration wet spells could potentially serve as additional benchmarks for meteorological models, where the emphasis is often on a model’s ability to reproduce first-order statistics of rainfall such as the mean or standard deviation.