The principal temporal focus of GOALS is seasonal-to-interannual, and the breadth of its spatial interest is global. However, to accomplish its scientific objectives, the program also will have to consider temporal variability over a broad range from diurnal to decadal time scales. For the design and implementation of process studies, specific regions embedded in the global domain will require special emphasis.

The range of time scales of interest is shown in Figure 3-1. GOALS concentrates on phenomena that include the interannual (e.g., ENSO), biennial, and annual variability of the coupled ocean—atmosphere—land system, as well as intraseasonal fluctuations (e.g., monsoon break and active periods, and the Madden-Julian Oscillation) that act to create patterns of weather and may conceivably influence seasonal-to-interannual time scales. At the higher-frequency end of the spectrum and where land surface processes are more directly involved, GOALS plans on specific interfaces with GEWEX. Investigations on low-frequency (long-time period) phenomena within the GOALS program include the decadal modulation of and trends in ENSO, interannual variability of monsoon and rainfall patterns, and the seasonal and interannual variability of the North Atlantic Oscillation and the Pacific North American Oscillation. Of importance also are the climate signals and their long time scale modulations found in ice cores, tree rings, coastal corals, pollen, etc. For these studies, GOALS will require an interface with the DecCen program.

The GOALS program gives due recognition to the possibility that anthropogenic effects on climate may modulate seasonal-to-interannual variability of the coupled ocean—atmosphere—land—ice system. The consideration of the anthropo-

Figure 3-1

Temporal scales, regional foci, and elements of the GOALS program. Note that the overlapping time scales of interest between the GOALS program and those of GEWEX and DecCen implies the need for well-coordinated interfaces between the programs.

genie effects on climate and climate impacts is primarily handled by DecCen and ACC with which close coordination is planned. Combining the focus of GOALS (seasonal-to-interannual time scales) and those of GEWEX and DecCen, leads to a complete program that addresses key issues concerning the predictability of climate on all time scales—a primary concern of the USGCRP, and the international CLIVAR program of the WCRP.

The spatial focus of GOALS is, in effect, one of nested domains. The largest scale, global, is the domain for global models. At this scale, "global" observations are needed for the initialization of models and for quantifying and evaluating their prediction skills. Empirical and diagnostic studies are proposed to document the seasonal-to-interannual variability of the natural global system. Both modeling and data analysis investigations should search for evidence of relationships between sub-components of the Earth system and low-frequency variability that may be exploited for improving prediction capabilities. This last activity is sometimes referred to as prospecting for predictability and will have a

bearing on which parts of the global domain are chosen for concentrated GOALS subprograms and projects.

Within the global domain, there are several regions in the tropics that will become foci of GOALS research. Within these regions process studies, long-term observations, and modeling efforts will target specific elements of predictability. The recommended strategy for U.S. GOALS for choosing the regional foci is defined below:

• The predictability found in the Pacific Ocean through ENSO should be followed to other regions of the tropics to determine how much of the variance can be explained in terms of ENSO.
• The regional foci should encompass the major tropical heat sources and sinks. There is considerable evidence of interannual variability in other regions of the tropics (e.g., the Asian—Australian monsoon system) together with predictive skill illustrated using empirical methods. Although there appears to be coupling between ENSO and variability in the other major heat sources and sinks, there also appears to be evidence of independent variability.
• The predictable elements found in the tropics from ENSO, or associated with the other heat sources and sinks in the tropics, should be projected to higher latitudes to see if they imply extratropical predictability. There is some evidence that the extratropics may not possess significant inherent predictability. However, this conclusion is based on limited observational studies and on coupled model experiments of relatively crude formulation, especially the oceanic component that has often been represented by fixed SST distributions. Investigations are needed to determine whether predictability on seasonal-to-interannual time scales, wherever it exists, has roots within the tropical system or whether there are predictable modes in the higher latitudes. To isolate predictable elements of higher latitudes (if they exist) it will be necessary to include more sophisticated land surface and sea ice processes than currently exist in climate models.

Figure 3-2 displays the regional foci now thought to be important for the attainment of the scientific objectives of GOALS. These are: (A) the tropical Pacific Ocean; (B) the Pan-American region extending from the eastern Pacific Ocean to Africa—the focus of the PACS program; and (C) the Asian—Australian monsoon system that encompasses the Indian Ocean, the western Pacific warm pool and the oceanic regions to its northwest and southwest, the oceanic throughflow through the Indonesian Archipelago that connects the Pacific and Indian Oceans, and the land masses of Australia, South Asia, and East Africa.

It is recommended that the initial emphasis for U.S. GOALS should be placed on the Pacific basin and the Pan—American region because of their significant and direct effects on North American climate. It is noted, however, that as our understanding of the global climate system develops, priorities may be expected to change and new regional foci identified.

Figure 3-2

The three regional foci of GOALS: the tropical Pacific Ocean (A), the Americas (B), and the Indian Ocean and surrounding continental regions (C). Background shading represents the long term August precipitation rates as estimated  by the MSU (units: millimeters per month). Regional foci encompass major precipitation  regions of the tropics.

The panel suggests that the extension of GOALS to the global domain and the expansion of present observational capabilities beyond the tropical Pacific to other regions should occur in an orderly manner and in parallel with a regionalization of prediction problems. Improvements in regional prediction skills are particularly important for applications and interfaces with society, both of which are high priorities of GOALS.