<< Back to WCRP CMIP3 Subprojects
- Rusticucci, M., J.Marengo, O.Penalba and M.Renom, 2006: COMPARISONS BETWEEN OBSERVED AND MODELED PRECIPITATION AND TEMPERATURE EXTREMES IN SOUTH AMERICA DURING THE XX CENTURY (IPCC AR4 MODELS). 8th ICSHMO Proceedings of 8 ICSHMO, Foz do Iguaçu, Brazil, April 24-28, 2006, INPE, 379-389., ISBN 85-17-00023-4.
One of the key aspects of Climate Change is to understand the behavior of extremes. It is recognized that the changes in the frequency and intensity of extreme events are likely to have a larger impact than changes in mean climate. We propose to assess the expected changes in climate extremes over southern South America through the analysis of the indices of the IPCC 4th Assessment Model Output for the present climate (IPCC20C3M). These "extreme indices" are derived data, from simulated daily temperature and precipitation, in the form of annual indicator time series. In this paper, for the common period 1960-2000, the mean, standard deviation and mean square error between the grid point from different models and the nearest station was calculated. The indices that could be comparable are: FD: annual occurrence of frost days (days with MinT <0ºC), Tn90: percentage of days where MinT was above the 90th percentile of the 1961-90 base period. R10: number of heavy precipitation days > 10mm (R10), CDD: consecutive dry days and R5D: maximum 5-day precipitation. The available models are CCSM3 (CCSM),USA CNRM-CM3, (CNRM), France GFDL-CM2 .0 (GFDL0) GFDL-CM2.1 (GFDL), USA, INM-CM3.0-Russia (INM), MIROC high-resolution MIROC), MIROC medre-resolution (MIRMED), Japan, and PCM, USA (PCM)
In general, it is necessary to think that the values of the models on the Andes cannot be evaluated by the failing of the models in interpreting the orography. If we center the analysis in the Southeast of South America, a low land region which has more dense information, one sees that the average value is well simulated, the station values has similar values over regions, and are of the same order of magnitude, as in case of the models. In some cases, as the model GFDL, FD's average values are well simulated. The interannual variability from days to weeks in average, also is in good agreement. Another temperature extreme, a warm one, Tn90, (being an index percentile-based, is relative to the local climate), It seems to be better represented that the FD. The consecutive dry days are more difficult to be simulated, since the region has a marked precipitation gradient that is not properly represented. The maximums of dryness over central Argentina Andes could not be represented for any model. On the other hand, the extensive dry season of the Amazon, is displeased and exaggerated in GFDL, whereas it does not exist for CCSM3. An index that measures the quantity of extreme rainfall, (R5d) shows in all the models that the quantity of rainfall is underestimated and there the differences of rate of rainfall are not clear. The one that better approaches the average values is the MIROC3.2. When the number of days is evaluated by extreme rainfall (R10), without considering how much it precipitated, the maps are more similar. The Amazon interannual variability is well simulated.
Full Article: http://gemini.dpi.inpe.br/col/dpi.inpe.br/Gemini@80/2006/04.05.15.27/doc/capa.h
Last Updated: 2006-05-17
<< Back to WCRP CMIP3 Subprojects