School of Meteorology (Defense)
Climate Downscaling over Northeast Brazil: Simulation, Prediction and Application
Ceara Research Institute for Meteorology and Water Resources (FUNCEME)
03 May 2012, 3:00 PM
National Weather Center, Room 1313
120 David L. Boren Blvd.
University of Oklahoma
Regional climate model (RCM) is a useful tool to understand and predict climate variability at regional scales, and climate downscaling using RCMs enables climate information on spatial scales more relevant to sectorial applications. The NCEP Regional Spectral Model (RSM), with horizontal resolution of 60 km, is used to downscale the ECHAM4.5 AGCM (T42) simulations forced with observed SSTs over northeast Brazil. An ensemble of 10 runs for the period of 1971–2000 is used in this study. The RSM can improve the regional climatology as compared to the driving AGCM, and can capture the observed rainfall variability at regional scales (e.g., mean rainfall and weather statistics). The EOF analysis reveals that the interannual variability of rainfall’s sub-GCM scale component is largely modulated by the tropical Atlantic SST anomalies. The RSM primary deficiency is a dry bias over many locations.
Ceara Institute for Meteorology and Water Resources (FUNCEME) has issued seasonal forecasts over Northeast Brazil since December 2001, in collaboration with the International Research Institute for Climate and Society (IRI). The NCEP RSM, the Brazilian Regional Atmospheric Modeling System (RAMS), and the IRI Climate Prediction Tool (a statistical downscaling tool) are used to downscale the IRI ECHAM4.5 AGCM ensemble forecasts. Forecasts have been issued monthly, for four upcoming running 3-month periods. Forecast ensembles of 75 members are postprocessed and merged into final probability forecasts. Verification of the first 10 years of FUNCEME forecasts reveals that the overall rainfall forecast skill, measured by the ranked probability skill score, is positive over a majority of northeast Brazil. Skill levels are generally higher for the rainy season than the pre-rainy season, and in northern region than the southern region of northeast Brazil. Over a period of as brief as 10 years, the variability in the amplitude of ENSO extremes is likely to govern forecast skill more strongly than incremental improvements in models or forecast methodology. The skill of the downscaled forecasts is generally higher than that of the driving global model forecasts, indicating the added values of the regional models. However, the downscaling forecasts do not capture the observed shifts in the rainfall climatology.
Successful application of climate downscaling forecasts in climate risk managements requires creativity to address users’ needs. We defined a weather index, using daily rainfall time series during the growing season for corn, to measure the severity of drought and flooding conditions. Corn yields are more sensitive to the weather index than the seasonal mean rainfall. The skill of corn yield prediction is largely attributed to the skillful forecasts of weather index. The reliability of streamflow forecasts is of a great concern to water resource managers. We use multiple hydrological models to increase the pool of streamflow forecast ensembles and apply multiple model ensemble combination techniques to improve the reliability of the streamflow forecasts. Since 2012, Ceara water allocation committee has used the FUNCEME’s streamflow forecasts for water management.