College of Engineering | Apply to CSU | Disclaimer | Equal Opportunity Statement | Privacy | Search CSU

Reconciling TRMM Precipitation Estimates to El Niño Southern Oscillation Variability

You must be on the CSU network—either physically or using VPN—to watch this or any of the videos on this site.

March 9, 2017
David Henderson
Hosted by Chris Kummerow (advisor), Steve Rutledge, Sue van den Heever, Branislav Notaros (Electrical and Computer Engineering)


Over the tropical oceans, large discrepancies in TRMM passive and active microwave rainfall retrievals become apparent during El Niño-Southern Oscillation (ENSO) events, where TMI retrievals exhibit a systematic shift in precipitation seemingly correlated with ENSO phase, while the PR does not. To investigate the causality of this relationship, this dissertation focuses, both spatially and temporally, on the evolution of precipitation organization between El Niño and La Niña conditions and their impacts on TRMM TMI and PR retrieved precipitation through the use of ground validation (GV) and satellite-based sources. The precipitation validation is performed as a function of convective organization through implementation of defined precipitation regimes, which have physical characteristics consistent across meteorological regimes.

The robustness of radar-based GV rainfall estimates from the Kwajalein S-band KPOL radar are examined through comparisons with the Kwajalein rain gauge network. The TRMM-GV 2A53 rainfall product is found to heavily underestimate convective rain types, where prominent biases occur as precipitation becomes more organized. To further examine these rainfall biases, GV and polarimetrically-tuned rain rates are compared, where GV biases in both the 2A53 product and convective and stratiform Z-R relationships are minimized when the rain rate relationships are developed specifically as a function of precipitation regime. The results demonstrate that exploration into precipitation regimes should be considered when deriving and evaluating rain relationships to establish the source and range of uncertainties existing within different precipitating systems.

TRMM radar (PR) and radiometer (TMI) rain rates are then evaluated though multiple case studies of collocated TRMM and KPOL rain rates at the 1°x1° and TMI footprint scale. The results of this study indicate that TRMM TMI and PR rainfall biases are best explained when derived as a function of organization and convective fraction. Large underestimates in both TMI and PR rain rates are associated with predominately convective rainfall across all regimes, where TMI rainfall underestimates both PR and GV rain rates. While PR rain rate estimates typically underestimate GV rainfall, TMI rain rates are heavily overestimated in rainfall regimes containing predominantly stratiform precipitation. Over the Kwajalein region, differences in TMI and PR rain rates seem to be driven by the occurrence of organized precipitation, where TMI-PR differences during El Niño conditions largely derive from MCS-like precipitating systems containing large stratiform precipitating regions. Application of the resultant biases helps mitigate the TMI-PR differences occurring between the ENSO phases and explain uncertainties introduced by the TMI Bayesian retrieval.

TRMM discrepancies directly relate to a shift from isolated deep convection during La Niña events toward organized precipitation during El Niño events with the largest variability occurring in the Pacific basins. During El Niño conditions, an increase in stratiform raining fraction leads to an increase in TMI rain rates that is less prevalent in PR rain rate retrievals. Reanalysis and AIRS data indicate that higher occurrences in organized systems are aided by increased mid- and upper-tropospheric moisture accompanied by more frequent deep convection. During La Niña events tropical rainfall is dominated by isolated deep convective regimes associated with drier mid-tropospheric conditions and strong mid- and upper level zonal wind shear. Application of the known TMI and PR biases yields increased consistency in PR rainfall with the radiometer-based TMI and GPCP rainfall estimates. The resultant satellite-based rainfall estimates are in general agreement when describing the response of tropical precipitation to ENSO induced variability in tropical SSTs.