Seasonal rainfall prediction in the East Mediterranean with statistical downscaling

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 26 January 2011: 10:30 AM
Seasonal rainfall prediction in the East Mediterranean with statistical downscaling
611 (Washington State Convention Center)
Wanli Wu, NCAR, Boulder, CO; and Y. Liu, M. Ge, G. Descombes, T. Warner, D. Yates, T. Hopson, S. Swerdlin, D. Rostkier-Edelstein, P. Kunin, and A. Givati

Seasonal precipitation prediction has significant societal and economic impact, particularly for arid and semiarid regions. Current seasonal predictions generally rely on general circulation models (GCMs), which have coarse resolution (~300Km). The GCM forecasts provide overall guidance in terms of large and synoptic scale perspectives, but lack of regional and local details and accuracy that are needed by hydrological applications and water resources planning and management. On the other hand, high-resolution (~ 10s Km) limited-area models have their own issues for operational seasonal prediction due to unavailability of reliable large-scale drivers and unaffordable computational needs. Thus statistical and dynamical scaling techniques have emerged to overcome scale mismatch between GCM products and regional (and local) application needs. In this study, a k-nearest neighbor (KNN) simulator is used to derive local precipitation based on NCEP Climate System Forecast (CFS) seasonal forecasts and historic rainfall observations. The KNN algorithm is an analog-type approach that queries days within a specified temporal window similar to a given weather feature vector in a GCM forecast. The selected K nearest neighbors is then rank-weighted to derive daily precipitation with the historic observed precipitations. This study is interested in the semiarid area along the southeastern coast of the Mediterranean Sea, which is strongly influenced by the Mediterranean climate. The KNN downscaling algorithm is first cross-validated with NCEP/DOE reanalysis (RAII, 1982 to 2009). Correlations between the observed daily rainfalls at 18 stations and RAII variables are calculated based on 20 plus years data to determine the predictors driving the KNN downscaling. Five RAII fields (sea level pressure, 700 hPa geopotential height, temperature, 850 hPa vertical velocity and 1000 hPa wind are then used for the precipitation downscaling. The cross-validation shows the KNN algorithm is very promising when driven with the reanalysis. We then tested the KNN with archived NCEP Climate Forecast System (CFS) seasonal forecasts (up to 9 months ahead of the initials). The CFS forecasts show certain skills in downscaling local precipitations but they are not as good as the RAII-driven KNN predictions. The mixed RAII-CFS-KNN scheme is proposed and under testing. Finally a WRF-based dynamical enforced statistical downscaling (DESD) is in consideration and will be tested in comparison with the statistical downscaling. The results from all experiments described above will be presented in this talk to illustrate the strength and weakness of the statistical downscaling in bridging the scale gaps between the global model rainfall forecasting and the regional and local hydrological applications.