Wikindx Resources

Communication incl. Poster: BibTeX citation key:  Berga
Berg, A., Sultan, B. & De Noblet, N. 2009. Simulating crop yields over West Africa with ORCHIDEE: sensitivity of model skill to rainfall forcing. Work presented at Third International AMMA Conference, July 20—24, at Ouagadougou, Burkina Faso.
Added by: roussot 2009-10-19 22:20:06
 B  
Categories: Land surface processes, Society-Environment-Climate interactions, Weather to Climatic modelling and forecasting
Keywords: Agriculture, Vegetation
Creators: Berg, De Noblet, Sultan
Publisher: African Monsoon Multidisciplinary Analyses (Ouagadougou, Burkina Faso)
Collection: Third International AMMA Conference

Number of views:  880
Popularity index:  55.94%
Maturity index:  published

 
Abstract
Studying the large-scale relationships between climate and agriculture raises a twofold issue: the impact of climate on crops, and the potential feedbacks to climate from croplands, through biogeochemical and biophysical processes. One relevant framework to address these interactions between climate and crops is to extend the land component of Earth System Models, in order to account explicitly for croplands. Following this approach, a representation of tropical crops has recently been introduced in the IPSL land surface model ORCHIDEE: based upon processes and parameterizations derived from the well-characterized crop model SARRAH for tropical cereals, this new version of ORCHIDEE, called ORCH-mil, was first applied offline over West Africa, forced with the NCC dataset (NCEP reanalysis Corrected by CRU data).
Here, we investigate the sensitivity of the model skill, in terms of yield simulation, to the rainfall forcing. The model skill can be defined, against observations from the FAO, as the model ability to correctly simulate large scale (i-e, national) millet yields, its ability to correctly capture the observed relationship between weather and yield, and its ability to simulate the observed year-to-year yield anomalies (i-e, the model score).
Thus, the model was ran offline over 1968-1990 forced with a range of datasets, differing mainly by their representation of rainfall: from three principal datasets (NCEP reanalysis, NCC, and an observational daily dataset based on interpolated rain gauge measurements, by the IRD), eight 1.0°, six-hourly, datasets were derived, resulting in a range of increasingly realistic rainfall forcing datasets, from “model-only” (NCEP) to “observation-only” (IRD). Analyzing the sensitivity of the model results to these forcings allows us to assess the dependence of the different components of the model skill on the realism of rainfall characteristics (total, seasonality, frequency).
Added by: roussot