HR: 17:30h
AN: H24C-07    [Abstracts]
TI: El Niņo/Southern Oscillation Impact on Rainfall Over South America: A Bayesian Approach to Improve its Forecasts.
AU: * Zamboni, L
EM: lzamboni@atmos.ucla.edu
AF: Abdus Salam, International Centre for Theoretical Physics, Abdus Salam ICTP Earth System Physics Section, Strada Costiera 11, Trieste, 34014, Italy
AU: * Zamboni, L
EM: lzamboni@atmos.ucla.edu
AF: Department of Atmospheric and Oceanic Sciences, University of California at Los Angeles, 7127 Mathematical Sciences building 405 Hilgar Ave, Los Angeles, Ca 90095, United States
AU: * Zamboni, L
EM: lzamboni@atmos.ucla.edu
AF: Department of Mathematics and Computer Sciences, University of Trieste, via Valerio 12, Trieste, 34127, Italy
AU: Mechoso, C R
EM: mechoso@atmos.ucla.edu
AF: Department of Atmospheric and Oceanic Sciences, University of California at Los Angeles, 7127 Mathematical Sciences building 405 Hilgar Ave, Los Angeles, Ca 90095, United States
AU: Kucharski, F
EM: kucharsk@ictp.it
AF: Abdus Salam, International Centre for Theoretical Physics, Abdus Salam ICTP Earth System Physics Section, Strada Costiera 11, Trieste, 34014, Italy
AB: The goal of this work is to improve the skill of long-range predictions of rainfall over South-Eastern South America (SESA). We follow a Bayesian approach, i.e. a combination of empirical relationships and numerical model's output. In the present work we focus on the period October-November, during which anomalous precipitation over SESA is significantly linked to El Niño Southern Oscillation (ENSO). The period of study is 1957-1998. The empirical model includes a linear regression of an index based on the observed anomalies in meridional wind over South America at 200 hPa and in mean precipitation over SESA during both El Niño and La Niña events; the correlation between these two timeseries is 0.7. The dynamical meaning of this empirical model lies in an observed relationship between mean anomalies over SA during ENSO events: enhanced precipitation over SESA is accompanied by an anomalous cyclone centered east of southern Brazil during El Niño events and vice versa during La Niña events. The wind index is defined as the mean of that quantity at the location of intense meridional wind over SESA during ENSO events. The numerical model output consists of a non-bias corrected ensemble of 4 runs by the UCLA AGCM in its 2 lon x 2.5 lat x 29 layers configuration. Ensemble members are 42-year long simulations with observed sea surface temperatures. The initial conditions correspond to December 1^st, 1956 plus small random perturbations. We find that our method can significantly improve precipitation forecasts over SESA during ENSO events. The skill score of the bias corrected ensemble, in reference to climatology, is 0.1; the corresponding value for the Bayesian forecast is 0.22 when all the years in the timeserie are considered. If only ENSO events are considered, the skill score increases from -0.7 for the bias corrected ensemble mean to 0.27 for the Bayesian forecast. The results for other seasons will be presented at the Conference.
DE: 1843 Land/atmosphere interactions (1218, 1631, 3322)
DE: 1854 Precipitation (3354)
DE: 3300 ATMOSPHERIC PROCESSES
SC: Hydrology [H]
MN: 2007 Fall Meeting