creasingly valuable if they add value to the observational base, correctly anticipate departures from the norm a season or a year in advance, or help to define either risk or opportunity tied to longer-term trends. Therefore, the scope of the modeling efforts should include the use of model-data hybrids to create long-term climatologies of variables that are not directly observed, a combination of statistical and dynamical models to assess conditions a season to a year in advance, and coupled earth-system models designed to incorporate changes in the factors (e.g., greenhouse gases, aerosols, solar variations, and land-cover change) that force long-term changes to the climate system (NRC 1998b, 1999b).
The range of spatial scales is equally diverse. Climate variability can have small spatial scales, and many climate products will be place-based (specific to one site). Hence, high resolution becomes a critical need in data used in creating site-specific products and in developing gridded products to guide decisions. Where inputs and products cannot be portrayed on common high-resolution grids, the ability to use models to downscale information is required to provide the requisite high-resolution products. Although many problems are site specific, the generation of climate products will rely on data not only across disciplines and time but also across space to points distant from the place of interest. For example, a regional seasonal forecast of precipitation and temperature in the United States will rely on ocean observations and surface marine observations, as well as on soil moisture, to initialize the global coupled ocean—atmosphere—land model used to produce the forecast. At the same time, downscaling of the forecast requires local climatologies, statistics, topography, land-use data, and other local or regional information.
As the decisions vary over space and time, climate services should at once be responsive on the local level and integrative of all the wide-ranging and diverse influences on that place. Fundamental to the development of climate information that serves the needs of the nation is a commitment to a global observing system (NRC 1999b); recognition of the importance of place-based, local and regional observations; a strong service-oriented modeling capability (NRC 1999c; 2001a); and a commitment to a user-centric focus (NRC 1999a).
The observing system required for the climate services is global. At the same time, the place-based studies of climate impacts will often require a higher resolution of observations. Therefore, attention to the various national and regional networks is also required. In response to changing capabilities and needs, the infrastructure of climate services cannot be seen as static. An important aspect will be the design and optimization of the observing system. That