WindSat - Remote Sensing of Ocean Surface Winds

P.W. Gaiser
Remote Sensing Division

The wind vector affects a broad range of naval missions, including strategic ship movement and positioning, aircraft carrier operations, aircraft deployment, effective weapons use, underway replenishment, and littoral operations. Furthermore, accurate wind vector data aid in short-term weather forecasting, the issuing of timely weather warnings, and the gathering of general climatological data. WindSat is a satellite-based multifrequency polarimetric microwave radiometer developed by the Naval Research Laboratory for the U.S. Navy and the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Integrated Program Office (IPO). It is designed to demonstrate the capability of polarimetric microwave radiometry to measure the ocean surface wind vector from space. The sensor provides risk reduction for the development of the Conical Microwave Imager Sounder (CMIS), which is planned to provide wind vector data operationally starting in 2010. WindSat is the primary payload on the Air Force Coriolis satellite, which was launched on 6 January 2003. It is in an 840-km circular Sun-synchronous orbit. It is currently undergoing rigorous calibration and validation to verify mission success.

Navy Need For Wind Information

WindSat strives to answer the battlespace environment question, "What's the wind speed and direction around the carrier strike group?" Winds over the ocean affect nearly every aspect of naval operations, including carrier operations, mission planning for precision guided munitions, surf forecasting for expeditionary forces, and avoidance of nuclear, biological, and chemical clouds. The global ocean surface wind vector (speed and direction) provides essential information for short-term weather forecasts and warnings, nowcasting, and climatology and oceanography studies in both the civilian and military sector. This can lead to improved accuracy in tropical cyclone forecasting and improved ship routing. Carrier post-cruise reports following deployments in the Adriatic Sea stated that the most critical piece of meteorological data was the wind direction. The same reports indicated that wind direction was the least-understood parameter in their environment. Despite this critical need, the Navy has not been able to obtain global wind direction information from space. Space-borne passive microwave sensors, such as the Special Sensor Microwave Imager (SSM/I), operationally provide environmental data such as tropospheric water vapor mass, cloud liquid water mass, sea ice age and concentration, and ocean surface wind speed.1 One parameter that has not been provided by microwave radiometers is wind direction. However, recent work and advance in polarimetric radiometry suggest that it may be possible to measure the complete ocean surface wind vector (speed and direction) from space-borne microwave radiometer (such as SSM/I) if the instrument is modified to include measurement of the full Stokes vector (the current generation of SSM/I sensors measure only the first two components of the four-element vector).2,3

WindSat is a satellite-based multifrequency polarimetric microwave radiometer developed by the Naval Research Laboratory Remote Sensing Division and the Naval Center for Space Technology for the U.S. Navy and the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Integrated Program Office (IPO). WindSat is designed to demonstrate the viability of using polarimetric microwave radiometry to measure the ocean surface wind vector from space. It is the primary payload on the Air Force Coriolis satellite, which is sponsored jointly by the DoD Space Test Program (STP) and the Navy (SPAWAR PMW-155). WindSat shares this mission with the Solar Mass Ejection Imager (SMEI) developed by the Air Force Research Laboratory (AFRL). A spacecraft developed by Spectrum-Astro of Gilbert, Arizona, supports both payloads. The WindSat/Coriolis mission was launched on a Titan II rocket from Vandenberg Air Force Base on 6 January 2003. In addition to potentially providing the Navy with badly needed ocean surface wind vector measurements, WindSat provides risk reduction data that the NPOESS will use in the development of the Conical Microwave Imager Sounder (CMIS), which would provide wind vector measurements operationally beginning in the 2009 time frame, building on the pathfinding work of WindSat.

Polarimetric Radiometry Background

Microwave radiometry is a well-established technology for remote sensing of the environment. Radiometers such as WindSat measure the microwave emission from the field of view (FOV) of its antenna. Figure 1 illustrates that the received energy is a combination of energy emitted from the surface (ocean), radiation from the atmosphere, and energy from the sky reflected off the surface. The measured quantity is known as the brightness temperature. For the surface emission, it is related to the physical temperature by

TB,p = ep(q,j)TPhys

where TB,p is the brightness temperature in polarization p, e is the scene emissivity, Tphys is the physical temperature of the scene, and (θ,φ) represent the viewing geometry. By definition, a perfectly absorbing and emitting blackbody has an emissivity of one. Therefore, at thermal equilibrium it has a brightness temperature equal to its physical temperature. All other scenes have an emissivity less than one. The emissivity depends not only on the geometry and polarization, but also on the physical properties of the medium. For sea water, in particular, the emissivity is a function of the water temperature, salinity, and the roughness of the medium's surface.

Figure 1 Image
Conceptual description of principles of microwave radiometer measurements.

It has long been known that the microwave emission from the ocean surface depends on the wind speed at the surface. As the winds increase, the seas become rougher and the microwave emission increases. However, the wind-driven waves on the ocean surface are not isotropic; their distribution varies with wind direction. Therefore, the intensity of the emission depends not only on the wave structure, but also on the orientation of the wind-driven waves.

WindSat is the first space-borne polarimetric microwave radiometer. As a polarimetric radiometer, WindSat measures not only the principal polarizations (vertical and horizontal), but also the cross-correlation of the vertical and horizontal polarizations. The cross-correlation terms represent the third and fourth parameters of the modified Stokes vector, defined as

Equation  Image

In this definition, Tv, Th, T45, T-45, Tlc and Trc represent brightness temperatures (radiances) at vertical, horizontal, plus 45°, minus 45°, left-hand circular, and right-hand circular polarizations, respectively. The Stokes vector provides a full characterization of the electromagnetic signature of the ocean surface and the independent information needed to uniquely determine the wind direction.

WindSat Instrument Description

Figure 2 shows the WindSat payload. The radiometer operates in discrete bands at 6.8, 10.7, 18.7, 23.8, and 37.0 GHz. Table 1 provides key design and performance parameters of the system. The 10.7, 18.7, and 37.0 GHz channels are fully polarimetric. The 6.8 channel is dual-polarization (vertical and horizontal), and is more sensitive to sea surface temperature (SST) than to winds. Thus it is used to remove measurement noise due to variations in SST. Similarly, the 23.8 channel has dual-polarization. This frequency responds strongly to water vapor in the atmosphere, which attenuates the signal from the ocean surface. Thus, 23.8 channel data mitigates the effects of the water vapor.

Figure 2 Image

WindSat payload in the thermal/vacuum chamber. The cold sky reflector has been removed in this picture.

WindSat uses a 1.8-m offset reflector antenna fed by 11 dual-polarized feed horns. The antenna beams view the Earth at incidence angles ranging from 50 to 55°. Table 1 shows the nominal beamwidth and resulting surface spatial resolution of each band. The Coriolis satellite orbits Earth at an altitude of 840 km in a Sun-synchronous orbit. The satellite completes just over 14 orbits per day. The orbit and antenna geometry result in a forward-looking swath of approximately 1000 km and an aft-looking swath of about 400 km. The fully integrated WindSat payload stands 10 ft tall and weighs approximately 675 lb.

Table 1 Image

WindSat Radiometer Measurements

After being powered on on 24 January 2003, the WindSat radiometer achieved thermal stability in less than 36 h, at which time the receiver gains also stabilized. The gains and receiver offsets were adjusted via ground commands to optimize the dynamic range of the receivers. The WindSat calibration and validation (cal/val) team then began rigorous analysis of the system performance and debugging of the ground data processing system (GDPS). One of the first products from this process was radiometric imagery for all 22 channels. As an example, Fig. 3 shows a brightness temperature image of three consecutive descending passes for the 18.7 GHz +45°-polarization channel collected on 31 March 2003. Several features stand out in the false-color image. The visible structures of bright bands and other regions of locally high brightness temperatures over the ocean are typical of atmospheric variations in water vapor, clouds, and rain. In the North Atlantic, the brightness temperatures in the Labrador Sea and Hudson Bay (colored yellow/green in the image) are higher than the ambient ocean and are typical of sea ice signatures, whereas the cooler brightness temperatures over Greenland are representative of a permanent ice sheet. Over land, there are also noticeable differences in brightness temperatures, which are due to differences in the physical temperature of the terrain, such as between North America (early spring) and equatorial South America. Lastly, in North America, the brightness temperature decreases with increasing latitude over northern Canada, which is consistent with signatures of snow cover in those regions for this time of year.

Figure 3 Image

WindSat image showing three consecutive descending passes of 18.7, +45 (18P) channel.

Figure 4 highlights a portion of this same pass over Mexico and Central America, showing simultaneous images at horizontal polarization for each of the five WindSat frequencies. In this series of images, one can see how the various frequencies respond differently to the same oceanic and atmospheric conditions, most noticeably in the increased sensitivity to clouds and water vapor (which appear as bright features) in the higher frequency channels.

Figure 4 Image

Composite image showing horizontal polarization at each of five WindSat frequency bands. Area is Yucatan peninsula of Mexico.

To illustrate the differences among polarizations at the same frequency, Fig. 5 depicts the 37 GHz imagery at vertical, horizontal, +45° linear, and left-circular polarizations. (For these color scales, differences between left and right circular polarizations are too subtle to be illustrative; the same is true for +45° and -45° linear polarizations, so only one of each pair is shown here.) When comparing these images, it is clear that over the ocean the magnitudes of the +45° and left-circular brightness temperatures lie between the values of the vertical and horizontal polarizations as is expected, since these polarizations are weighted combinations of vertical and horizontal.

Figure 5 Image
Composite image showing 37 GHz brightness temperatures at vertical, horizontal, +45°, and left-circular polarizations.

Wind Data from WindSat

As part of the WindSat calibration and validation, we have compiled a large set of WindSat data matched with modeled wind fields over the ocean. These wind fields come from the Global Data Assimilation System (GDAS) generated by the NOAA National Center for Environmental Prediction. The GDAS wind fields are generated every 6 h. The matchups are limited to WindSat data collected within 1 h of the analysis. Figure 6 shows WindSat third and fourth Stokes parameters plotted against wind direction. The different colors represent different wind speeds. The signals can be modeled as sinusoidal functions. The third has significant first and second harmonics, whereas the fourth is dominated by the second harmonic. It is also interesting to note that the third and fourth Stokes parameters are odd functions with respect to the wind direction. Vertical and horizontal polarizations also behave sinusoidally, but they are even functions relative to wind direction. These signatures agree well with model simulations and data collected from airborne polarimetric radiometers.4

Figure 6 shows that the peak-to-peak variations change with the wind direction. However, even at the higher winds, the peak-to-peak signal is less than 4 K. Typical 37 GHz vertical polarization brightness temperatures over the ocean are approximately 200 K. Thus, the wind direction signal is a very small. Therefore, the WindSat design emphasized sensitivity and stability in the calibration.

Figure 6 Image
Sensitivity of 10.7 GHz third (a) and fourth (b) Stokes parameters to wind direction. The colors represent different wind speed ranges. Wind vector truth data supplied by the NOAA GDAS system. Figure courtesy of NOAA/NESDIS/ORA.

As a more tangible way to look at the wind direction signal, consider Fig. 7, which depicts an image of Hurricane Isabel from the WindSat 18.7 GHz third Stokes parameter channel. The wind circulates around the storm in a counter-clockwise fashion. Comparing the signal in the image with the plots in part (b) of the figure, one can see how the signal changes as the wind direction shifts around the perimeter of the storm. Hurricane Isabel was a Category 5 storm when these data were collected. The intense rain and high levels of clouds and water vapor attenuate the peak-to-peak response of the WindSat signals. Collecting data at multiple frequencies provides necessary information to account for the effects of the atmosphere and the sea surface temperature.

Figure 7 Image
Hurricane Isabel on 14 September 2003, as seen by WindSat 18.7 GHz third Stokes parameter. Notice how the signal changes around the circulation pattern of the storm.

The objective of the WindSat program is to retrieve the ocean surface wind vector from the WindSat data. The NRL WindSat team is developing algorithms and software that take advantage of the characteristics described earlier. The algorithms must work globally and derive the wind speed and direction to within 2 m/s and 20°, respectively, over a range of 3 to 25 m/s winds. After successful calibration and validation, all algorithms will be transitioned to the Navy for operational use of WindSat data.

Working in collaboration with other institutions such as the Jet Propulsion Laboratory (JPL) and the National Oceanographic and Atmospheric Administration (NOAA), preliminary wind vector retrievals have been developed. Figure 8 shows an example of these early retrievals in the southern Atlantic where (a) is the retrieved wind vector and (b) is the GDAS wind field for that time. There is very good agreement between the two images. Notice in particular the increasing wind speed from north to south. Similar circulation patterns can be seen in the low wind region at the northern edge of the image. While this is only a preliminary retrieval, it demonstrates the vast potential of the WindSat mission.

(a)Preliminary wind vector retrieval compared with (b) GDAS analytical wind field. The image is over the southern Atlantic and is based on empirical model developed using WindSat/GDAS matchups. Figure courtesy of NOAA/NESDIS/ORA.


As the mission, tactics, and platforms of the Navy continue to evolve, the need for improved battlespace environment intelligence grows. The WindSat mission is designed to satisfy this requirement by providing timely ocean surface wind vector measurements to the warfighter. The WindSat payload is the first space-borne polarimetric microwave radiometer. Early results demonstrate the capability to retrieve the wind speed and direction over the ocean. As the calibration and validation phase of WindSat nears completion, NRL will continue to develop the WindSat capability to get needed information into the hands of the users.

The mission of WindSat does not end there. WindSat data, algorithms, and designs are playing an important role in the development of the NPOESS CMIS, which is the future operational source of environmental data such as ocean winds. Furthermore, the WindSat data are being studied and exploited, not only at NRL but also at many research institutions, to develop new applications for polarimetric microwave radiometer data.


I acknowledge the contributions of the many who helped in the development and exploitation of WindSat. These include David Spencer, Michael Mook, Jerry Golba, Jeff Cleveland, Bill Purdy, and Pattie Klein of the Naval Center for Space Technology. Within the Remote Sensing Division, I thank Elizabeth Twarog, Karen St. Germain, Gene Poe, Don Richardson, Richard Cember, Larry Choy, Craig Smith, Mike Bettenhausen, Al Uliana, and Beverly Gardiner. Lastly, I thank my collaborators at the NOAA/NESDIS Office of Research and Applications: Paul Chang, Nai-Yu Wang, and Tim Mavor, with special thanks to Larry Connor and Zorana Jelenak for building the matchup database and developing the preliminary wind direction retrieval.

[Sponsored by SPAWAR and NPOESS]

1J.P. Hollinger, J.L. Peirce, and G.A. Poe, "SSM/I instrument evaluation," IEEE Trans. Geosci. Remote Sens. 28(5), 781-790, (1990).
2S.H. Yueh, W.J. Wilson, F.K. Li, S.Y. Nghiem, and W.B. Ricketts, "Polarimetric Measurements of the Sea Surface Brightness Temperature Using an Aircraft K-band Radiometer," IEEE Trans. Geosci. and Remote Sens. 33(1) 85-92, (1995).
3P. Chang, P. Gaiser, K. St. Germain, and L. Li, "Multi-Frequency Polarimetric Microwave Ocean Wind Direction Retrievals," Proceedings of the International Geoscience and Remote Sensing Symposium 1997, Singapore, 1997.
4K.M. St. Germain, G.A. Poe, and P.W. Gaiser, "Polarimetric Emission Model of the Sea at Microwave Frequencies and Comparison with Measurements," Prog. Electromag. Res. 37, 2-32 (2002).