2000 Articles

Capabilities and limitations of dynamical climate models.

Goddard, L.

Within the many research initiatives around the world that seek to use seasonal climate forecasts to improve management of agricultural production systems, there is growing interest in linking seasonal climate forecasts based on atmospheric models with dynamic crop simulation modelsfor impact prediction and decision-support
applications. Although the utility of the general approach has been demonstrated using categorical indices of climateteleconnenctions(e.g.,ENSOphases), direct useof predictionsfromdynamic atmospheric models have been hampered by the mismatch in spatial and temporalscale between these climate models and the point (i.e.
station) spatial scale and daily time step of the crop models. The IRI and START cosponsored a workshop that brought climate prediction scientists, crop model users and other interested scientists together to identify and address the technical issues related to the mismatch between the format and spatial and temporal scale of
output from seasonal climate prediction models, and the requirements of crop simulation models. A long-term objective of the workshop is to contribute to the development of techniques for linking climate model outputsthat the IRI can implement either for routine use in house, or distribute to interested agricultural users of seasonal climate forecasts. Specific workshop objectives were to:
1. Clarify issues related to the use of climate prediction model output as input to crop simulation models.
2. Communicate relevant crop model requirements to climate prediction community, and climate model
characteristics, capabilities and limitations to crop model application community.
3. Describe existing and potential approaches for addressing the spatial and temporalscale mismatches
between dynamic climate prediction and crop simulation models.
4. Develop a strategy for further developing, evaluating, and implementing promising approaches.
Workshop presentations highlighted the general issues, and discussed alternative methods to downscaleseasonal climate prediction model output in space and in time. Participants then discussed a range of related issues, and explored options for evaluating the various methods using a common set of climatemodel hindcasts
and associated data for a few regions where predictability is good. This report presents summaries of invited presentations, and highlights and recommendations based on the discussion.