Reprinted from backscatter - Newsletter of the Alliance for Marine Remote Sensing, Vol. 7. No. 1, Feb., 1996.
Merging Over-the-Horizon Radar with Satellite Oceanographic Data
- T. M. Georges, NOAA Environmental Technology Laboratory, Boulder, CO
- J. A. Harlan, Cooperative Institute for Research in Environmental Sciences, University of Colorado/NOAA, Boulder, CO
- Paul Chang, NOAA National Environmental Satellite, Data and Information Service, Camp Springs, MD
The ocean-remote-sensing capabilities of over-the-horizon (OTH) radar and satellite ocean sensors are obviously complementary. OTH radar looks at fixed ocean areas on demand, whereas satellites cover the globe in swaths dictated by orbital dynamics and sensor field-of-view. Furthermore, each of these instruments measures different ocean properties with varying reliability. The prospects of extracting improved products, specifically surface wind fields, by merging satellite and OTH radar data, prompted some tests, whose early results we describe here.
Both active (radar) and passive (radiometer) microwave sensors can be used to determine ocean surface wind speed, and active microwave instruments are also used to derive wind direction, though resolving directional ambiguities has been an ongoing issue. Additionally, recent airborne radiometer systems have demonstrated a capability of determining wind speed and direction using polarimetric and multi-look measurement techniques. Development and refinement of instrumentation and algorithms for ocean surface wind retrieval, and particularly wind direction, is an ongoing process for both active and passive sensors.
OTH radars measure surface wind directions with a two-fold ambiguity that is often resolvable by combining incidental surface observations with meteorological insight [Harlan et al., 1994; Young et al., 1996]. Wind speeds can also be measured in principle, but in practice, ionospheric distortions often severely limit coverage in space and time. Therefore, it seems reasonable to combine wind directions derived from OTH radar with wind speeds provided by spaceborne radiometry and scatterometry.
During the 1994 hurricane season, we used the Air Force OTH-B radar system in Maine to map surface wind direction in the tropical Atlantic, for evaluation by the National Hurricane Center [Georgeset al., 1995]. We combined these wind directions with the surface wind speeds measured in the same region by the ERS-1 scatterometer and by the Special Sensor Microwave Imager (SSM/I) on the DMSP satellites. The merged products were made available daily on the World Wide Web. Figure 1 displays an example of ERS-1 ocean-surface wind speed (encoded in grey scale) and direction (shown as arrows) at 10-m height, as computed by the European Space Agency (ESA) and provided in their fast delivery product. The empirically-derived algorithm used by ESA to relate normalized radar cross-section to wind speed and direction is referred to as CMOD4 [Offiler, 1994]. The ascending and descending swaths shown represent 24 hours of coverage in the North Atlantic.
Figure 2 shows the same ERS-1 wind speeds encoded in grey scale, but with the wind directions replaced with those measured by the OTH-B in about one hour on the same day. The height to which OTH wind-direction measurements refer is the effective height of the boundary-layer winds that drive decametric ocean waves, within which direction does not change significantly. In this case, the OTH-B directional ambiguity was resolved by simply selecting easterly (rather than westerly) surface flow over the entire region. Within the area mapped by the OTH-B, the spaces between the satellite swaths are filled in with sufficient continuity that the synoptic flow pattern can be discerned. In some parts of Fig. 1, the wind directions given by the ESA fast delivery algorithm are inconsistent, particularly in regions where wind speeds are low.
Figure 3 shows ocean surface wind speed at 19.5-m height calculated from the SSM/I brightness temperature on the same day as Figs. 1 and 2. The SSM/I brightness temperatures used are calculated by the Navy at the Fleet Numerical Meteorology and Oceanography Center (FNMOC). The wind speed algorithm used was developed by Goodberlet et al. [1989]. Again, the OTH-B wind direction field is superimposed. The gaps within the satellite swaths are most likely due to sensor outages in regions of high liquid water content, which obscures the ocean surface. The wind speeds mapped by the active and passive satellite sensors are in reasonable agreement, where there is overlap.
When ambiguities are correctly resolved, scatterometer and OTH wind directions are also in reasonable agreement, although there is some evidence of a bias. Recently, Schollaert et al. [1996] compared OTH-B wind directions measured for 41 days in the tropical and subtropical Atlantic with the Freilich/Dunbar (FD) maximum-likelihood ERS-1 winds in an effort to evaluate the performance of the FD ambiguity removal algorithm. When 1,143 data pairs were analyzed, and the closer OTH direction to ERS-1 was selected, the results of the comparison show that there was a mean directional offset, or bias, of approximately 10°, with the OTH value higher than, i.e., clockwise of, the ERS-1 value. The standard deviation was about 20°. It remains to be determined whether this bias has a geophysical explanation or points to errors in one or both wind-direction algorithms. One physical possibility is that long ocean waves, traveling at an angle with the surface wind, shift the direction of maximum wind stress (and the dominant direction of the short waves seen by the scatterometer) away from the surface wind direction. A consistent bias in the mean long-wave direction compared with the mean surface-wind direction in the region studied could explain the OTH-scatterometer direction bias. Wave climatology for the Atlantic is being studied to examine this hypothesis.
Multiple technologies for mapping ocean-surface winds are developing at such a rapid pace that it is premature to judge which will emerge in operational form. It is already clear, however, that existing experimental ground- and space-based techniques complement each other in space and time coverage and resolution, as well as in the reliability of speed and direction measurements. Further studies of merged products should lead to methods for reconstructing surface winds over the ocean with coverage exceeding that now available over the land.
Georges, T.M., J.A. Harlan, L.R. Meyer, and C.A. Grunden, Ocean surface wind directions measured by the Air Force over-the-horizon radar during the 1994 hurricane season, NOAA Tech. Memo. ERL ETL-246, 93 pp., 1995.
Goodberlet, M.A., C.T. Swift, and J.C. Wilkerson, Remote sensing of ocean surface winds with the Special Sensor Microwave/Imager, J. Geophys. Res. 94, pp. 14, 547-14, 555, 1989.
Harlan, J.A., and T.M. Georges, An empirical relation between ocean-surface wind direction and the Bragg-line ratio of HF radar sea echo spectra, J. Geophys. Res. 99, pp. 7971-7978, 1994.
Offiler, D., The calibration of ERS-1 satellite scatterometer winds, J. Atmos. Ocean. Technol. 11, pp. 1002-1017, 1994.
Schollaert, S.E., P. Cornillon, and J.A. Harlan, A Comparison between Over-the-Horizon radar and ERS-1 Scatterometer Wind Directions, Proc. 1995 ADEOS/ NSCAT Science Working Team Meeting, Kyoto. November 29 - December 1, 1995.
Young, G.S., J.A. Harlan, and T.M. Georges, Application of over-the-horizon radar observations to synoptic and mesoanalysis over the Atlantic, submitted to Weather and Forecasting, 1996.
[Table of Contents]
[Previous]
[Next]
