This sensitivity of the SAR surface signature to such a wide variety of physical processes is a mixed blessing. To effectively harvest such a rich data set, one must be able to recognize and separate each signature. Often, the signatures can be distinguished by the spatial scales over which they occur, by the local depth of the modulation they produce, or by their shapes and patterns. But ambiguities remain, since we have only a partial understanding of what makes each signature unique. This ignorance is to some extent a result of the vast areas over which the SAR imagery is collected, the high spatial and temporal volatility of some of the physical processes which lead to the signatures, and the logistic problems of conducting well-controlled and comprehensive experiments. Nevertheless, even at this stage, many of the signatures can be categorized with certainty, and many others with high probability. Furthermore, as more of these images are pondered by the larger science community, we can expect our understanding to increase.

The initial motivation for this compendium was to examine the possibility of SAR for global current monitoring, particularly as an adjunct to the NOAA Advanced Very High Resolution Radiometer (AVHRR), which is often severely limited by cloud cover. The SAR signal penetrates not only cloud cover, but also darkness and moderately heavy rain, although heavy rain can affect the character of the surface signature. To systematically examine the possibility of current monitoring, ERS-1 SAR ascending passes were collected at three-day intervals for a 35 day period (with one exception) off the US East Coast during April and May 1992, in conjunction with AVHRR imagery of the same region and time interval. The 35 day repeat orbit of ERS-1 allowed the creation of a large, 800km square mosaic by merging a set of 12 ascending passes of typical length 800 km.

Unfortunately, but as expected, the supporting AVHRR images were usually cloud covered (429K), and of little value in daily tracking the thermal boundaries of the Gulf Stream and its associated eddies. However, two nearly cloud-free images of the region were collected close to the beginning and end of the 35 day interval. This single pair is a poignant reminder of the limitations of AVHRR imagery, and is also a clear indicator of the dynamics of ocean circulation that might be revealed with SAR imagery.

The ERS-1 SAR images have been corrected for both the spacecraft antenna pattern, and the relative fall-off of radar backscatter with increasing off-nadir angle. Small (1% to 5%) radiometric mismatches between adjacent frames have been removed, so that the relative backscatter within a pass is probably correct to within a few percent. The entire set of 12 passes has been rescaled to a common backscatter calibration curve. When combined with both the local wind climatology and one of the commonly accepted radar wind algorithms, an estimate of absolute backscatter is possible.

This common backscatter calibration curve allows wind speed to be estimated over all the passes. For a fixed wind direction, the intensity of a SAR image (in the absence of contaminants due e.g. to current features) is a monotonic function of the wind magnitude. With a few basic assumptions, including some a priori knowledge of the wind direction, one should therefore be able to relate the intensity of the SAR image to the surface wind magnitude. This example shows an attempt to illustrate this relationship between the backscatter and wind speed.

The remainder of the compendium consists of 87 sets of SAR images. The set , when viewed as a whole, clearly illustrates that SAR ocean images are literally filled with atmospheric and oceanic signatures of all scales. Not a single frame lacks some sign of activity, and while many of the signs can be easily associated with a specific cause, many cannot. So there is much speculation, and much tentativeness in the captions. Some will someday be proven wrong. This risk is both inevitable and welcome since, as noted above, a well-controlled comprehensive experiment to explain each "scratch, scrape, and scar" is logistically impossible, given the spatial scope of the imagery.

This compendium therefore is not offered to explain, but more to provoke. In that spirit, it should not be taken as polished science, but simply as an observational record of some especially interesting natural processes. If it is even moderately successful, it will help provide an improved and more accurate perspective on the potential value of spaceborne SAR over the ocean.