2

APPLICATIONS

OF SATELLITE-BASED

OBSERVATIONS

During the discussions of the NOSS Science Working Group, a number of research areas were discussed for which satellite techniques offer either valuable complementary information to ship and buoy direct observing techniques, or represent the only feasible way to obtain the information needed. In addition, several other planning studies are under way in connection with the World Climate Research Program. The feasibility of at least one major ocean experiment is being examined internationally. Others are under discussion in the U.S.A. These suggestions are not idle suggestions; they relate closely to the research interests of individual scientists, based on what they would like to be involved in themselves should the opportunity arise.

The following descriptions of possible research activities are not ordered according to priority, but illustrate a range of important and challenging scientific applications. Many such research objectives could be met by a few satellite flight programs, and there are many ways in

which observing systems may be combined on any particular flight. No attempt is made here to discuss such matters.

2.1 Models

Phenomenological and analytical techniques and models play important roles in providing required detailed descriptions and also represent a highly useful and valid approach to the physical understanding of many important oceanic processes. The data series discussed in this report as necessary will be of great importance to further development and verification of all types of models.

Increasingly, numerical models of various types are becoming a powerful tool by which diverse observations of the ocean are tied together to form and test a coherent picture of the state of the ocean and the processes at work. There are nearly always insufficient data to determine uniquely all the fields needed. The simplest model is a diagnostic or universal one, which explores the range of possible interpretations subject to certain kinematic constraints (such as geostrophy, or conservation of mass, salt and temperature). A number of other model types have been developed which start from the equations of motion and are dynamically and thermodynamically consistent, though not necessarily correct descriptions of the real situation: some of these address the general circulation in the world ocean; there are eddy-resolving models of the circulation in specific ocean basins; some models aim at the transient response to changes in atmospheric forcing near the equator; there are one-dimensional models of the upper mixed layer and seasonal thermocline; and there are storm surge and tidal models for specific coastal waters. Each of these requires specific inputs and makes predictions which may be used for validation.

For example, a general circulation model normally assumes fields of surface wind stress, surface temperature, and net precipitation minus evaporation, and predicts time mean fields of subsurface temperature, salinity, and velocity, together with sea level and surface heat flux. Numerical experiments can determine the sensitivity to reasonable variations in model parameters such as eddy diffusivity, and then the available data are used to test the overall consistency. The more reliable the information that is available, the better. Specific types of measurements will have greater or lesser impact depending on the problem at hand. The surface stress, properly time-averaged over several weeks or longer, and sufficiently accurate to determine the distribution of its curl on the scale of the oceanic phenomenon under consideration, is nearly always a significant input; however, for mixed-layer models greater time resolution may be required, and near coastlines it is the stress itself rather than its spatial gradient which is most important. Time-averaged surface

temperature is usually also a required input; however, unless validation against surface heat fluxes is involved, the precision available from historical merchant ship data is normally adequate. Fluctuations in surface temperature may be a valuable validation tool in time dependent problems, and synoptic measurements sufficiently accurate to determine them are potentially important. The near-surface geostrophic velocity is another validation variable, intimately connected to gradients in sea level after tidal oscillations have been eliminated. It is the large-scale averages most naturally provided by measurements of sea level which are most valuable for all except local process studies.

For investigations of the large-scale time-dependent response of the near-surface waters to varying wind stress, a first approximation is to ignore the motion of the abyssal waters and to consider the movement of the layer above the thermocline as a vertically coherent unit. This is a central problem in the interannual variations of global climate, centered in the tropical ocean. The combination of wind stress and sea level alone would then provide a major test of the dynamics of the models; with the addition of sea-surface temperature to test the thermodynamics, remarkably complete intercomparison would be possible.

In most cases however, direct subsurface measurements are needed to provide an adequate range of validation variables to test model performance. Near-surface temperature and heat storage can be measured from surface drifters using satellite communications as an essential link. Subsurface velocities from constant-level floats can again provide the large-scale long-time averages so important for most studies. Satellite. communication links and position finding can greatly extend the possibilities here. On the other hand, there appears to be no substitute for ship-based profiles of temperature and salinity and radioactive tracers which provide essential information about the volumetric distribution of different water masses. Examination of the models designed for each of these specific problems yields insights into the information gained from observations of any given accuracy, sampling density, and overall coverage, which where possible are qualitatively reflected in what follows. More thorough examination of these issues is necessary before a complete picture of the potential impact of space techniques on oceanography can be constructed.

2.2 Generic Observations

There are a variety of observations which, if made continuously over several years and processed and reduced to formats designed for research use, would supply the needed coherent data sets for many specific research problems and for test of models, as discussed above. Moreover, such data sets would, if continued operationally, form the bases of extrac

tion of indices useful for long-term monitoring of variability. The existence of such coherent and long-term data sets is of paramount importance in atmospheric research. Oceanic research would be similarly benefited by establishment of regular observations.

Space-based techniques could contribute significantly to provision of long-term, internally consistent observational series. However, the satellite information would be even more valuable if reliable ties could be established to the conventional historical data bases. The establishment of such reliable connections, and the careful editing of the historical data would be far from a routine trivial task.

2.2.1 Wind Stress

The wind stress at the surface is one of the major driving forces of oceanic circulation. There are no systematic observations with which to test the performance of various models of ocean circulation and ocean response to the atmosphere. Ship observations of wind provide some coverage in regions served by commercial shipping; ship observations, however, are noisy (i.e., may contain undetectable errors) and uncalibrated (e.g., for ship effects) and must be processed carefully before use.

In the opinion of scientists who are trying to develop better models of the ocean circulation, one of the greatest needs, at present, is a coherent, calibrated long-term data set of surface stress or wind over at least the tropical zone, and preferably over the globe. The Seasat data processing effort and the experience with the validation program indicate what explicit measurements must be made in situ to facilitate the use of the basic observations. The Seasat data offer an enticing glimpse of future routine wind stress/wind velocity observations globally. But can satellite techniques really supply the information with enough ancillary data for its interpretation? Many special studies will be needed to improve the interpretation of scatterometer observations (i.e., to translate the radar backscatter cross section of capillary waves into stress/speed) and also to identify situations in which there might be other physical or biological factors contributing to the backscattered signal, i.e., to identify reliably the various surface effects that influence the backscatter, and to make adequate corrections.

For example, the return signal from a scatterometer depends on the presence of surface structures with scales in the centimeter range; the usefulness of the scatterometer in measuring wind speed depends upon variations in the intensity and density of these structures as a function of wind speed. Several different kinds of structures at these scales can be discerned and they may vary with wind stress in different ways; the scattered response lumps them all together.

One kind of structure involves groups or trains of capillary-gravity waves at these scales, generated directly by the wind stress and perhaps to