StormWatch Experiment


Theresa Brandt
Atholton High School

            

(Click on image to see larger view.)

Wondering what theses images are???  They're both of the Atlantic Ocean off the eastern coast of the US.....
The pictures were made by AVHRR and SAR space-borne satellites.

The purpose of this website is to create a description of the Synthetic Aperture Radar (SAR) experiment and related topics. Topics to be mentioned include:

 
 1. A general description of the operation of the SAR (Synthetic Aperture Radar),
 2. A brief description of the Advanced Very High Resolution Radiometer (AVHRR),
 3. A description of the differences and similarities between AVHRR and SAR,
 4. A summary of the relevant main points of past experiments,
 5. A summary of the present (StormWatch 97-98) experiment, including its:
 a) Purpose,
 b) Hypotheses,
 c) Assumptions,
 d) Materials/Data Sources,
 e) Procedures,
 f) Calculations,
 g) Analyses, and
 h) Conclusions, and
 6. References.

 

Part 1: A General Description of the Operation of the SAR.

 
(link to major source of information for topic: http://southport.jpl.nasa.gov/ then go to science and applications and then imaging radar reports)
 
The SAR, otherwise known as Synthetic Aperture Radar, actively views and creates images using radar. This active imaging can be most easily imagined as the taking of a flash picture where the illumination of the scene comes from radio waves emitted by the satellite, rather than from light waves emitted by the flash. The radio waves reflect or scatter off the Earth's surface back to the antenna. As waves in a pool of water reflect from objects placed in the tank, so too do the radio waves emitted by the antenna reflect or scatter diffusely off the Earth's (or any) surface. The diffusely scattered waves that return to the radar are called backscatter.

In the simplest case, the antenna and the target area on the surface have no relative motion. Thus, a diagram of a wave transmitted and received directly above a flat surface, or perpendicular to the point, would look like this: (see diagram1a).

diagram1a
The entire wave would be reflected back to the antenna. At any angle other than one perpendicular to the satellite as described above, the entire wave would bounce off the surface and go off into space (see diagram1b, ID).
diagram1b
In either case, the wave does not scatter diffusely because the surface has no roughness (on the scale of the radar wavelength). Were the surface to have roughness however, the radio waves would scatter diffusely. The amount of backscatter received by the antenna depends upon the roughness of the surface at the point of illumination. The rougher the target, the greater the diffuse scattering; the greater the diffuse scattering, the more backscatter received; the more backscatter received, the brighter the image. This works at any angle because some backscatter will always reach the antenna if the surface is rough (see diagram2a, ID).
diagram2a
By measuring the time the wave takes to return to the antenna and knowing that the velocity of the wave is c, the speed of light, the distance from the point to the antenna is easily computed. This process finds the range of the target, the y value in an x-y coordinate system. Range resolution, or the amount of detail in the y direction, is limited by the ability to measure the return-trip travel time of backscattered energy. The azimuth resolution, or the amount of detail in the x direction, depends on the size of the radar antenna. The angular resolution of an antenna is l/D, where l is the radar wavelength and D is the size of an antenna. In the case of a spaceborne antenna that is 12 meters long and a radar wavelength of 0.20 meters off nadir (the positive y axis), the angular resolution is 0.20 / 12 or 0.016 radians. At an orbital altitude of 800 km looking 20 off nadir, this angular resolution corresponds to an azimuth ground resolution of 14 km. This means that a SAR image taken from space would show details in the x direction as small as 14 km. Thus, azimuth resolution is limited by antenna size. Yet SARs today achieve good resolutions of 25 m from space, which would normally require an antenna about 7 kilometers long. The solution to the relatively small antenna with such detail lies in the motion of the SAR in space.

In the previous examples, the relative velocity of the radar and the target is zero; they are not moving with respect to each other. Examining some examples of single-object motion, where the relative velocity of the Earth is zero and the satellite some non-zero number, illustrates two equally valid ways the azimuth, or the x value, can be found, either by understanding the "Synthetic" part of SAR or by understanding the effect of the Doppler shift on the backscatter as the satellite travels through space.

The movement of the SAR creates a "synthetically" large aperture. As the satellite moves, the antenna emits pulses at regular intervals, receiving in the between times. Returning to the flash photograph example, just as the aperture of a camera changes its depth of field and focus, so too does the aperture of a radar change its ability to perceive the surface beneath it. In the radar's case, the longer the antenna, or the greater the aperture, the sharper azimuth resolution. A longer antenna is synthesized or imitated by receiving and recording backscatter data at multiple points. Thus, the SAR is a Synthetic Aperture Radar with azimuth resolution limited now only by the synthetic aperture size.

Another, equivalent way to appreciate SAR's ability to achieve such high azimuth resolution is to understand how the Doppler shifts involved affect the backscatter. The backscatter received by the antenna as it travels has experienced a Doppler shift (for a description of the Doppler shift, see Appendix B). As the satellite travels past the target, the time is measured, the wavelength adjusted for Doppler shift, and the range and azimuth, the y and x values respectively, of the point found. When the SAR is far from the target, the Doppler shift in the frequency of the wave is large. As the SAR approaches the target, the Doppler shift decreases to zero: where the along track position of the SAR equals the azimuth position of the target or the SAR is directly over the target. The amount of Doppler shift determines the azimuth or the x value on an x-y coordinate system. Thus, enough is known that, by taking the Doppler shift into account, it becomes possible to find and record the actual data necessary for later reconstruction of the image.

A radar image from a satellite is composed of many pixels which themselves are a compilation of backscattering data sent serially from the radar to a ground receiving station and then recorded by digital computers. The serial form of the data is processed by the computers into a finished image. The brightness of a certain area of the image depends upon the amount of backscatter the antenna received. This amount depends upon the roughness of the surface elements illuminated by the radio waves. A rougher surface will cause a greater reflection of the radar waves back to the radar. This means that the antenna can receive more backscatter as it moves along its path. Backscatter variation is caused by the size of the scattering objects in the target area, the moisture content of the target area, the polarization of the emitted pulses, the inclination towards the radar, and the observation angles of different wavelengths. The size of the object is a major factor. Objects the size of the emitted wavelength or larger reflect more backscatter while objects smaller than the size of the wavelength reflect less backscatter. The fewer the backscattered waves received by the antenna, the darker the resulting image. The opposite is also true, resulting in a brighter image. Below is a list of various common objects matched with their amount of backscatter (from very rough to very smooth) and their brightness (from black, as minimum backscattering received, to white, as maximum backscattering received).
 
1. Vegetation: moderately rough, gray to light gray.
2. Buildings: moderately smooth, dark gray. Some buildings appear very bright (light gray to white) because they are lined up in such a way that the pulse bounces off street and then the building, finally returning directly back to the antenna. This is called a double bounce.
3. Roads: smooth, dark, usually black.
 
Electrical properties also cause variation in backscattering. The wetter or more highly conductive an object is, the brighter it appears in the image and the opposite. A smooth body of water, the exception to this, acts as flat surface, reflecting transmitted waves away from the antenna and thus appearing dark and smooth on the radar image.

Top of Page

 


Part 2: A Brief Description of AVHRR and How It Works.

 
(link to major source of information for topic: http://fermi.jhuapl.edu/avhrr/primer/primer_html.html)
 
Advanced Very High Resolution Radiometer (AVHRR) passively measures the amount of spectral radiation emitted by bodies of water. An AVHRR is passive because it only receives radiation from the scene and not from the instrument as a radar does. The AVHRR measures the amount of radiation, emitted from the surface of a body of water to derive the surface temperature of the water. Differences in temperature can reveal currents, eddies, and other aspects of the ocean.

Scientists use the equations of black body physics when interpreting the amount of spectral radiation emitted by the sun, reflected off the earth, and received by AVHRR. The theories of black body radiation, derived from those of molecular motion, show that all substances above absolute zero, the point at which all molecular motion stops, emit radiation. A body which absorbs all incoming radiation is called a black body. While no existing material is an absolute black body (nothing absorbs absolutely all incoming radiation), some bodies, like water, absorb enough that they may reliably be considered black bodies.

In using the AVHRR, one must realize that spectral radiance depends mainly upon the electromagnetic wavelength and absolute temperature of a body. Wien in 1893 and Planck in 1901 developed the basis for the equation used today with variables of wavelength and temperature. Since no real object yet discovered is a perfect black body, the emissivity of any real object is described as the ratio of the amount of radiation actually emitted to that of a black body at the same temperature and wavelength. Since the wavelength of spectral radiance is known and AVHRR measures the actual emission, scientists can then estimate the surface temperature of the body relatively easily.

As with any data-gathering process, AVHRR has its limitations. The intervening atmosphere, indeed, anything between the AVHRR and its target absorbs some of the radiation emitted by the body, as well as emitting its own radiation both directly to the satellite sensor and indirectly to the surface of the target and then to the satellite sensor. Clouds block all infrared radiation from the surface and reflect a large amount of their own radiation to the sensor. This is referred to as cloud contamination. Another form of contamination, called solar contamination, comes from reflected solar radiation. While all cloud- contaminated regions must be discarded, regions affected by solar contamination may be corrected for by mathematically removing the exterior influences. Scientists use the passive measurement of the subject's radiation at different wavelengths to determine the amount of contamination that they need to correct. (For a more detailed description of AVHRR channels and correction factors see Appendix A.)

 
Some software is available for compiling the raw data into a final map of surface temperatures. Most of the software involves the following steps:
 
1. Create a master map.
2. Obtain the initial pass data.
3. Ingest and calibrate the data.
4. Navigate the data onto the master map.
5. Ignore data from large angles.
6. Test for clouds.
7. Compute the temperature.
8. Register the final image.

In addition to images from single satellite passes, composite images incorporate approximately three and sometimes seven consecutive days' data. A composite image shows much greater detail than a single day's data would because such things as cloud contamination on the first day are filled in by the following days' data. Among others, currents are generally more apparent in composite images than in single images. Such currents as the Gulf Stream off the North American Atlantic coast are colorfully illumined as warm, columnar regions with eddies swirling off their edges. The AVHRR is also useful for examining currents such as the North Atlantic Current, the last remnant of the once strongly delineated (temperature-contrasted) Gulf Stream. As the North Atlantic Current flows eastward and becomes diffuse, although still retaining much of the heat present in the Gulf Stream. This heat retention both allows the current to be followed for a time with AVHRR, and significantly moderates the climate of northern Europe.

Top of Page

 


Part 3: A Description of the differences and similarities of SAR and AVHRR.

 
SAR and AVHRR are similar in that they both use electromagnetic wave properties to measure ocean characteristics such as currents. Otherwise, the two are relatively distinct. The SAR actively measures electromagnetic radio waves while AVHRR passively measures electromagnetic visible or nearly visible waves. Under some conditions, the SAR can penetrate some meters of vegetation and soil, showing old contours, and thus making it more useful than AVHRR for some classes of land imaging. However, AVHRR does measure ocean temperature which the SAR cannot. An AVHRR also creates quality images of surface ocean currents when their temperature is distinct form the surrounding ocean. The SAR is useful for creating topography, or altitude maps and for locating ocean currents even in the presence of clouds. Thus, SAR and AVHRR both use the properties of electromagnetic waves to measure currents and other ocean phenomena. An AVHRR passively measures temperature, a very specific, measurable quantity, while SAR actively measures the roughness of a surface even through cloud cover and darkness. So, while both a radar and a radiometer are useful for the study of oceans, the SAR does so in an active and much more generic sense than the specific, passive measurement of AVHRR, and each provides information about the ocean surface that the other cannot.

Top of Page

 


Part 4: A Summary of the Relevant Points of Past Experiments.

 
(link to major sources of information for topic: http://fermi.jhuapl.edu/sar/scratches/overview.html and http://fermi.jhuapl.edu/sar/stormwatch/stormwatch.html)

The ERS-1, 1992:

The ERS-1 (European Remote Sensing satellite SAR) collected data over the ocean from Cape Hatteras to Cape Cod to investigate the sensitivity of SAR to the wind and wave fields, the stability of the marine atmospheric boundary layer, and other geophysical parameters. Specifically, scientists hoped that the SAR could complement the AVHRR's current monitoring ability by taking data readings through cloud cover, darkness, and rain which AVHRR can not. Limited by a meager sample time and insufficient coincident supplementary data (only one pair of AVHRR images was sufficiently cloud-free so as to merit correlating to the SAR images), the ERS-1 SAR was yet able to portray many of the aspects, called signatures, of wind and waves, as well as various man-made phenomenon.

The SAR can show these seemingly widely different signatures because it records the backscattering of small waves, which are affected by many different phenomena. These small waves may be modified in any number of ways, some of which scientists identified in this experiment. First, atmospheric and oceanic fronts alter the small waves either by changing them with the surface wind velocity (atmospheric fronts) or with the current gradients (oceanic fronts). Both natural and man-made surfactants (surface-active substances such as oils), as well as longer surface waves, affect the small waves' backscatter.

If cold air is beneath warm air, the atmosphere is stable. If cold air is atop warm air, the atmosphere is said to be unstable, since warm air tends to rise. The plane in which the water and air meet is known as the marine atmospheric boundary layer. Changes in the marine atmospheric boundary layer stability, when an atmosphere is unstable, alter the way winds cause small waves' backscattering, thus altering the radar image. Changes in the marine atmospheric boundary layer stability also affect wind speed and its signature in a SAR image. For example, if cold water is beneath cold air, the atmosphere is stable. A wind blowing through the cold air will lose momentum to the cold air. On a molecular level, since wind is the transfer of energy between air molecules through collisions, more of the warm air's energy will be lost to exciting the cold air's molecules. Cold molecules, with less kinetic energy than those of the warmer air, will require more energy to excite to the same speed as the warm air. If warm air is atop cold water, the atmosphere is said to be unstable, since warm air tends to rise. The effect on wind speed is opposite that of a stable system. The wind will gain momentum from the rising warm air. Molecularly, the warm air present will require less of the wind's energy to excite the atoms. The closer the two bodies of air are in temperature, the easier it becomes for the wind to pass through. Since wind affects the signature of an object, especially the ocean, on the SAR image, scientists found the knowledge of these relationships and the subsequent interpretation of the estimate of wind speed from the SAR image quite useful.

The wide variety of signatures SAR produces necessitates more research to define and identify each signature. In some cases, the definition and identification process is easy; in others, it becomes increasingly difficult. Scientists also realized that the vast range of phenomena which affect a SAR image would require well controlled and comprehensive experiments to identify each signature separately. A truly well controlled and comprehensive experiment will probably never happen since the environment the SAR observes is far from well controlled (we can not control the weather, for one thing) and a truly comprehensive experiment would cost an exceedingly large amount of money. Yet scientists continue with hope and further study, continuing to learn more.

Specific to wind and the development of the StormWatch 97-98 experiment, ERS-1 was used to examine the wind field by rescaling the images to a common backscatter calibration curve, allowing wind speed estimates over all the passes. Having discerned that a monotonic relationship existed between wind magnitude and SAR image intensity, scientists believe that, with some prior knowledge of wind direction, a relationship between the two can be found that will create accurate wind fields using only one SAR.
 

The RADARSAT, 1996-7:
The Canadian RADARSAT SAR, like the ERS- 1 SAR it followed, collects data from Cape Hatteras to Cape Cod. However, the wider RADARSAT SAR swath (500 km versus only 100 km from ERS-1) effectively increases adjacent spatial coverage to once every three days. By developing techniques to temporally composite AVHRR imagery (several temporally consecutive images' data merged together) scientists are able to obtain relatively uncontaminated AVHRR images collocated to the SAR images. Both the AVHRR images and the added wind and wave model forecasts provided every six hours greatly increased the corresponding supplementary data, allowing scientists better empirical comparisons than were possible using ERS-1. Scientists continued to study the relationships between backscatter and the various phenomena which affect a SAR image. Further signatures were identified, and the influential relationship between wind field and the SAR image was realized. Thus, RADARSAT allowed scientists to continue their study of SAR and its resultant images.
 
Top of Page


Part 5: A Summary of the Experiment

 
(link to major source of information for topic: http://fermi.jhuapl.edu/sar/stormwatch/stormwatch.html)
 
 
Purpose: "To explore the potential of the SAR to yield high resolution wind field estimates in the coastal zones."

Hypotheses: "[T]hat SAR will ... provide the extra information necessary to construct high resolution (order of 1 km) estimates of the wind field all the way to the coast, and even into inland bays and estuaries." (http://fermi.jhuapl.edu/sar/stormwatch/stromwatch.html).

Assumptions: That correcting for the marine atmospheric boundary layer stability is possible. That other factors will have a minimal influence on the image. That a connection between the wind velocity and that shown in a SAR image exists.

Materials / Data Sources: SAR, Geostationary Operational Environmental Satellite (GOES) water vapor data, Fleet Numerical Oceanographic Meteorological Center (FNMOC) wind direction fields, ERS-2 SAR wind vectors, FNMOC WAM waves and NOPGAPS wind direction fields and analysis and SST sea surface temperature, AVHRR marine atmospheric boundary layer (MABL) stability estimate, Navy Operational Regional Atmospheric Prediction System (NORAPS) air temperature analysis, and Special Sensor Microwave Imager (SSMI) wind fields.

Procedures: The SAR makes six cycles with ten passes per cycle and twenty-four days in each cycle. Each swath has dimensions of 450 km by 1350 km. The resolution is 100 m with a pixel size of 50 m. Incidence angles range from twenty to forty-six degrees. Eventually, winter weather will bring extreme wind speeds which will be useful in determining the relative boundaries of the derived algorithms for wind speed. Concurrent data gathered from the following sources will be used in verifying data extracted from the SAR images.
GOES, FNMOC, ERS-2, AVHRR, FNMOC WAM, NORGAPS, and SST, NORAPS, and SSMI.
After correcting for the marine atmospheric boundary layer stability, the SAR image will be further filtered and matched with coinciding data from the above mentioned sources and a correlation will be found. Once found, an algorithm for eliminating contaminants and refining the wind field image will be generated. Eventually, after StormWatch 98-99 to test and extend the findings of StormWatch 97-98, an operational product might be produced which correctly estimates coastal wind field maps with a minimum of error in real-time situations.

Analyses: TBA

Conclusions: TBA

 
Top of Page

Home
Created by: Theresa Brandt
Last up-dated: 3-4-98