• The importance of the parameter to the specific concerns of a given naval mission, such as intelligence gathering, use of certain weapons systems, and aircraft operations. Some of the parameters may be mission critical; others may be of much less importance.

• The accuracy and precision to which the parameter value must be known. In many cases, the parameter may have a critical threshold for making a go/no go decision. It may be sufficient to know the value with only modest accuracy when the value is far from the threshold but necessary to know it to high accuracy near the threshold.

• The spatial and temporal variability of the parameter. In some cases, an average value over relatively large areas and long times may be sufficient; in other cases, pinpoint accuracy at specific times and places may be required.

• The relative uncertainty of the predicted value. In practice, the decision to proceed with a particular naval mission depends on many factors, one of which is the environment. The role of the environmental factor in the ultimate decision process is determined in part by the certainty of knowledge of particular key parameters. The more certain the value, the more probable its use in decisionmaking.

One approach to dissecting this problem into manageable components is to sort and organize parameters by spatial and temporal scales and associated predictability. The following section discusses the relationship of mission time lines to physical knowledge. The next section focuses on predictability and model scale. The final section outlines data availability and its relevance to mission timescales.

Figure 3-1 presents a schematic space-time diagram illustrating the spatial and temporal scales of variability of many environmental systems. Of particular importance in evaluating this diagram is development of an objective method for understanding present and future capabilities for predicting environmental conditions across different spatial and temporal scales. For example, at the largest spatial (global) and temporal scale (decadal), climatological data can be treated statistically to arrive at estimates of the state of the global environmental system for any interval of the present year. These statistics are largely probabilistic functions defining the likelihood that the system will be found to lie within defined bounds. At this scale of observation, present-day predictive skill is reasonably well developed. However, as the spatial and temporal scales of processes become smaller, the influence of nonlinearities in each environmental system becomes larger, resulting in greater uncertainty in the ability to predict the state of the environmental system. At the smallest time- and space scales, nonlinearities result in environmental processes that are dominated by stochastic processes such that