21 Combining GEFS with CHIRPS Precipitation Estimates to Bias Correct Precipitation Forecasts for Food Security Analysis

Monday, 8 January 2018
Exhibit Hall 3 (ACC) (Austin, Texas)
Martin Landsfeld, Univ. of California, Santa Barbara, CA; and C. C. Funk, L. S. Harrison, P. Peterson, S. Shukla, and G. Husak

NCEP's Global Ensemble Forecast System (GEFS) provides medium range (16 day) precipitation forecasts at 1 degree resolution for a relatively long term period (1985 – present). These forecasts can provide a valuable insights into future scenarios and are used by the Famine Early Warning Systems Network (FEWS NET) for time sensitive food security analyses of developing and ongoing drought episodes. The Climate Hazards Group Infrared Precipitation with Stations (CHIRPS) dataset is a gridded land based, quasi-global (latitude 50N-50S), 0.05 degree resolution and has a similar period of record (1981-present). CHIRPS is based on a well developed climatology and incorporates topographic and latitudinal effects which express small scale rainfall distribution much better than the 1 degree GEFS.

We combine the GEFS and CHIRPS products to produce a bias-corrected field at the CHIRPS resolution (GEFS-CHIRPS). The bias-correction follows the quantile-quantile approach based on matching cumulative distribution rankings for both the GEFS and CHIRPS. The results and the raw GEFS forecasts are compared to in situ data at the dekadal (10 day) time scale. We describe differences in accuracy between the two products with a focus on drought prone areas of the developing world. We evaluate the impacts of forecast data choice for several drought monitoring scenarios. These include drought risk assessments made midway through the growing season and forecast accuracy in semi-arid areas and areas with complex topography. In conclusion, we outline cases for which using the higher resolution GEFS-CHIRPS product leads to substantial gains in information.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner