Tropical Precipitation Extremes

Abstract

Classifying tropical deep convective systems by the mesoscale distribution of their cloud properties and sorting matching precipitation measurements over an 11-yr period reveals that the whole distribution of instantaneous precipitation intensity and daily average accumulation rate is composed of (at least) two separate distributions representing distinctly different types of deep convection associated with different meteorological conditions (the distributions of non-deep-convective situations are also shown for completeness). The two types of deep convection produce very different precipitation intensities and occur with very different frequencies of occurrence. Several previous studies have shown that the interaction of the largescale tropical circulation with deep convection causes switching between these two types, leading to a substantial increase of precipitation. In particular, the extreme portion of the tropical precipitation intensity distribution, above 2 mm h 21, is produced by 40% of the larger, longer-lived mesoscale-organized type of convection with only about 10% of the ordinary convection occurrences producing such intensities. When average precipitation accumulation rates are considered, essentially all of the values above 2 mm h 21 are produced by the mesoscale systems. Yet today's atmospheric models do not represent mesoscale-organized deep convective systems that are generally larger than current-day circulation model grid cell sizes but smaller than the resolved dynamical scales and last longer than the typical physics time steps. Thus, model-based arguments for how the extreme part of the tropical precipitation distribution might change in a warming climate are suspect.

Contrasting the co-variability of daytime cloud and precipitation over tropical land and ocean

“…Previous studies averaged precipitation rates to a single value representing grid mean (e.g., Lee et al, 2013;Rossow et al, 2013). In this study, a marginal histogram of 3B42 0.25 • × 0.25 • grid precipitation rates is created for each 1 • × 1 • grid cell.…”

“…Using the International Satellite Cloud Climatology Project (ISCCP) WSs defined in the extended tropics (35 • S-35 • N), Lee et al (2013) provided a comprehensive picture of precipitation characteristics for each WS, with an additional focus on the relationship between the most convective regime (WS1) and precipitation. Rossow et al (2013) also conducted similar analysis but for precipitation extremes using ISCCP WSs for the deep tropical zone of 15 • S-15 • N. While such CR-based approaches provide valuable information about the cloud-precipitation relationship at large scales, the precipitation composites by CR encompass large spreads which obscure details of the relationship. Since CRs contain mixtures of clouds types by design, and therefore contain considerable cloud variability, ambiguities in the cloud-precipitation relationships are hard to resolve.…”

Contrasting the Co-variability of Daytime Cloud and Precipitation over Tropical Land and Ocean

“…Previous studies averaged precipitation rates to a single value representing grid 20 mean (e.g., Lee et al 2013;Rossow et al 2013). In this study, a marginal histogram of 3B42 0.25° grid precipitation rates is created for each 1°×1° grid cell.…”

“…Using the International Satellite Cloud Climatology Project (ISCCP) WSs defined in the extended tropics (35°S−35°N), Lee et al (2013) provided a comprehensive picture of precipitation characteristics for each WS, with an additional focus on the relationship between the most convective regime (WS1) and precipitation. Rossow et al (2013) also conducted similar analysis but for precipitation 10 extremes using ISCCP WSs for the deep tropical zone of 15°S−15°N. While such CR-based approaches provide valuable information about the cloud-precipitation relationship at large scales, the precipitation composites by CR encompass large spreads which obscure details of the relationship.…”

Bivariate copula in Johor rainfall data

“…The characteristic of rainfall data in tropical countries is believed that belongs to extreme values [12]. Gumbel, Clayton and skew t copula are good in detect the dependent extreme values [13][14].…”