Seasonal Scale Water Deficit Forecasting in East Africa and the Middle East Region Using the NMME Models Forecasts

Friday, 18 December 2015: 11:20
3020 (Moscone West)
Shraddhanand Shukla1, Chris C Funk2, Benjamin F Zaitchik3, Bala Narapusetty4, Kristi R Arsenault5 and Christa D Peters-Lidard5, (1)University of California Santa Barbara, Santa Barbara, CA, United States, (2)University of California Santa Barbara, Geography, Santa Barbara, CA, United States, (3)Johns Hopkins University, Baltimore, MD, United States, (4)NASA GSFC, Greenbelt, MD, United States, (5)NASA Goddard Space Flight Center, Greenbelt, MD, United States
In this presentation we report on our ongoing efforts to provide seasonal scale water deficit forecasts in East Africa and the Middle East regions. First, we report on the skill of the seasonal climate forecasts from the North American Multimodel Ensemble (NMME) models over this region. We evaluated deterministic (anomaly correlation), categorical (the equitable threat score) and probabilistic (the ranked probabilistic skill score) skill of the NMME models forecasts over the hindcast period of 1982-2010, focusing on the primary rainy seasons of March-May (MAM), July-September (JAS) and October-December (OND). We also examined the potential predictability of the NMME models using the anomaly correlation between the ensemble mean forecasts from a given model against a single ensemble member of the same model (homogenous predictability) and rest of the models (heterogeneous predictability), and observations (forecast skill). Overall, we found precipitation forecast skill in this region to be sparse and limited (up to three month of lead) to some locations and seasons, and temperature forecast skill to be much more skillful than the precipitation forecast skill. Highest level of skill exists over equatorial East Africa (OND season) and over parts of northern Ethiopia and southern Sudan (JAS season). Categorical and probabilistic forecast skills are also higher in those regions. We found the homogeneous predictability to be greater than the forecast skill indicating potential for forecast skill improvement. In the rest of the presentation we describe implementation and evaluation of a hybrid approach (that combines statistical and dynamical approaches) of downscaling climate forecasts to improve the precipitation forecast skill in this region. For this part of the analysis we mainly focus on two of the NMME models (NASA’s GMAO and NCEP’s CFSv2). Past research on a hybrid approach focusing only over equatorial East Africa has shown promising results. We found that MAM seasonal precipitation forecast skill in this region could be improved by harnessing the recent teleconnection between the precipitation in the region, with the precipitation and SSTs in the Indo-Pacific ocean region.