The requirements for a program of climate research are often perceived to be in conflict with what is required for weather services: climate-related studies require long-term consistency whereas weather services (e.g., forecasts, severe weather warnings) require rapid delivery of data products. However, climate can be viewed as the long-term statistics of weather. This linkage may be exploited to meet both sets of requirements. Thus variables that are critical for weather forecasting, such as atmospheric profiles of temperature and humidity, are also important for climate research and modeling.
One of the fundamental differences is the time scale over which both the forecasts and the observations must be made. For example, simple weather forecasts based on persistence (i.e., tomorrow's weather will be the same as today's) work well over short time scales. On longer time scales, more complicated models and a richer observation suite are required. For example, ocean processes (such as ocean circulation) and terrestrial processes (such as evapotranspiration) need to be included to produce forecasts with sufficient capability or to understand the critical processes. It is possible to generalize by noting that as the time scale of interest increases, more processes and more complicated interactions become important. This effect leads to the occasional surprises noted in the overview of the Pathways report (NRC, 1998a)—the unexpected processes or linkages that appear in studies of climate change. Box 1.1 elaborates further on the distinctions between weather and climate.
Nonlinear processes and the increasing number of interacting processes make it impossible to define a priori all of the types and scales of observations that need to be made. Accordingly, there should be balance between the focused research missions where the scientific underpinnings are well known and the wide-open, broadly based observations of some operational missions. The Pathways report overview (NRC, 1998a) discusses a scientific framework to support an observing strategy. This framework builds on the first decade of the U.S. Global Change Research Program (USGCRP) and identifies several areas of science and observations where a renewed focus and a rebalancing of priorities are required.
A noteworthy inference that can be drawn from the Pathways report is that establishing a robust understanding of the linkages between large-scale global processes and smaller-scale regional processes is an enormous challenge. For example, changes in ecosystem structure may be linked to changes in the patterns of climate variability, which in turn have feedbacks on the climate system. Moreover, public policy will respond to such regional-scale impacts rather than to broad-scale global change in mean Earth system properties. The Earth Observing System (EOS) and National Polar-orbiting Operational Environmental Satellite System (NPOESS) missions need to accommodate this scientific framework and balance the often conflicting primary missions of operational and research systems, the needs for continuity and innovation, and the needs for process studies and long-term monitoring.
The characteristic scales of climate variability demand long time series in order to determine the critical processes as well as to separate natural variability from anthropogenic influences. Unlike weather forecasting, the interval between stimulus and response can be several years to centuries. With a high level of background variability, subtle changes in Earth's climate system can be difficult to detect. This problem is further complicated by the changes in instrument technology or sampling strategies that may occur during the period of observations.
The task of assembling a record of total solar irradiance (Willson and Hudson, 1991) illustrates the challenges facing the development of long-term consistent time series. Developing this record required a rigorous calibration and sensor characterization program and an observation approach that ensured sufficient temporal overlap (as well as sensor validation) to achieve accurate cross-calibration between successive sensors. Another notable example is the record of upper-troposphere temperature (NRC, 2000).
In the science community, long-term data sets are sometimes perceived as being the result of unchanging data collection activities that are not at the forefront of innovative research. It is difficult to base a scientific career on such an activity. Nevertheless, the atmospheric CO2 record started by C.D. Keeling at Mauna Loa shows the importance of such long-term records and how the scientific value of such time series increases as the record