GC43B-1195
Characterization And State-Of-The-Art Modeling Of Extreme Precipitation Events Over Africa During The Historical Period
Thursday, 17 December 2015
Poster Hall (Moscone South)
Peter Gibba, West Africa Science Service Centre on Climate Change and Adaptation Land Use, West Africa Climate System, Ouagadougou, Burkina Faso and Mouhamadou Bamba Sylla, West Africa Science Service Centre on Climate Change and Adaptation Land Use, Wascal Competence Center, Ouagadougou, Burkina Faso
Abstract:
The ability of the state-of-the-art climate models to reproduce the mean spatial characteristics of extreme precipitation indices over Africa is evaluated. The ensembles of eight precipitation-based indices as defined by ETCCDI were extracted from seventeen CMIP5 GCMs and twelve CORDEX RCMs simulations based on absolute and percentile (95th) thresholds and computed from the 1975 to 2004 historical period. Daily precipitation indices calculated from GPCP and TRMM satellite-derived observation datasets during the period 1997 to 2012 and 1998 to 2011 respectively were also employed in this study for model validation. Results of spatial representation of the frequency of extreme precipitation events (R1mm, CDD, CWD and R95p) highlight a generally good consistency between the two observations. Equally, in the regional analysis some similarities exist in their median and interquartile (25th and 75th percentile) spread especially for CDD, CWD and R95p for most regions. In the associated intensities (SDII, RX5day, R95 and R95ptot), results indicate large spatial differences between the two observational datasets, with finer resolution TRMM generating higher rainfall intensities than the coarser resolution GPCP. TRMM has also demonstrated higher median and interquartile range as compared to GPCP. The CORDEX RCMs and CMIP5 GCMs simulations have estimated more number of extreme precipitation events, while underestimated the intensities. The differences between the models and observations can be as large as the typical model interquartile spread of the ensembles for some indices (R1mm, CWD, SDII and R95) in some regions. Meanwhile, CORDEX estimations are generally closer to the observations than CMIP5 in reproducing the frequency of extreme rainfall indices. For the estimation of rainfall intensities, CORDEX simulations are in most cases more consistence with TRMM observations whilst the CMIP5 GCMs simulations are closer to GPCP observations.
