Synthetic Aperture Radar Imagery Transcript

Slide 1

Welcome to the NOAA CoastWatch module on understanding Synthetic Aperture Radar (or SAR). I’m Chris Jackson and I’m part of the NOAA’s Center for Applications and Research. This module will present examples of how various oceanographic and atmospheric phenomena appear in imagery.

Slide 2

This class consists of three modules. Module 1 presented an overview of how a synthetic aperture radar operates. In module 2 we’ll talk about SAR data products and where you can get them. In this module, we will go over imagery examples, pointing out some interesting features that have been observed in various SAR images.

Slide 3

SAR observations provide useful information about a variety of phenomena and each one has a unique signature in the SAR imagery. Oceanic phenomena include ocean surface waves, current boundaries, surfactants, sea ice and internal waves. Atmospheric phenomena are detectable by SAR because they interact with the ocean surface. These include convection cells, rain cells and atmospheric gravity waves. Ships and ship wakes are readily visible. Land applications include identifying flooding and inundation or determination of land use. There are many additional applications.

Slide 4

Radar systems work by transmitting energy and recording the portion of the energy reflected back to the antenna. For smooth surfaces, the transmitted signal is mostly reflected away from the antenna. You can visualize this by thinking about shining a flashlight at an angle towards a mirror. Almost all the light will be reflected in the direction the beam is pointed rather than back to the flashlight. Rough surfaces allow a portion of the radar energy to be reflected back to the antenna. The reflected energy is called radar Backscatter is its measure is called the normalized radar cross section;

Slide 5

So let’s start the course by looking at a generic SAR image. At first glance it looks very similar to a black and white photograph but instead of recording reflected light, the variation in shading represents variations in the back scattered radar signal which is proportional to the surface roughness. The slide shows a Sentinel-1 SAR image acquired over the south shore of Labrador on 18 August 2023 in horizontal / horizontal (HH) polarization. In the upper left, the land is producing a high backscatter return.. You can see two distinct patterns of gray on the land and a series of irregular dark shapes (which are water bodies). In the lower left of the image the ocean surface is smooth and resulting in hardly any backscatter and appears nearly black. This means that the surface lacks roughness in the form of capillary waves that are generated by the wind. In the upper right, the ocean surface is being roughened by a wind of approximately five knots and so has a medium gray appearance. The lower right, the ocean surface has a mottled appearance, the results of the backscatter being modulated by some atmospheric convection.

Slide 6

The previous slide shows the ocean surface appearing as different shades of gray indicating a varying amount of radar backscatter or roughness. More roughness (at the scale of the radar wavelength) produces more backscatter and thus a brighter return. As noted on the prior slide the wind is the principal source of the roughness for the centimeter scale waves that produce ocean surface backscatter. The wind moves over the ocean surface interacting with the water generating small capillary waves. When the capillary waves reach the wavelength of the synthetic aperture radar (approximately 5 cm for Sentinel-1’s C-Band radar) Bragg-like resonance scattering takes place. This capillary wavefield is then modulated by the various oceanographic and atmospheric phenomena allowing their unique signatures to appear on the SAR image

Slide 7

This slide shows a near featureless Sentinel-1 image was collected on 27 January 2023 over the North Atlantic. And even though it appears featureless, even such non nondescript images can contain useful information….

Slide 8

…you just need to apply the right algorithm for extracting that information. In this case by applying a geophysical model function, which relates normalized radar cross-section to ocean surface wind speed, we can see there is a significant variation in the wind speed across the image from around 40 kts in the northwest corner, to around 20 knots in the southeast corner. The fine resolution of the SAR, In this case approximately 500 meters, allows you to see subtle variations in the wind field within this larger change across the image.

Slide 9

Here we have a SAR image acquired over Hawaii on 29 May 2022. The islands are clearly visible in light gray. The bright area over the ocean is an area of higher winds. The darker areas are the result of a portion of that wind being blocked by the island’s topography. The imprints of some storm cells are also visible to the northeast of the Big Island. Here localized winds associated with the storm cell modulate the capillary wave field to produce the cell’s imprint on the ocean surface. This image was collected from the Sentinel-1 C-band SAR in VV vertical / vertical polarization..

Slide 10 (Consider deleting - similar to slide 9)

Similar effects can be seen here in this Sentinel-1 image over the Windward Islands. A 15 knot wind out of the east is enough to set up the signatures of island wakes and wind funneling. In addition wind interaction with the topography has also produced a set of atmospheric lee waves in the island’s wake. These patterns can be seen to extend more than 100 km downwind of the islands

Slide 11 - Barbados

A different set of atmospheric signatures are visible in this Sentinel-1 image over the eastern Caribbean near Barbados. The wind is again around 15 kts out of the east but this time we see boundary layer roll signatures along with the imprints of gust fronts from localized storm cells. Cold pools and micro-cold pools (which are regions of colder, denser air produced when precipitation evaporates) are also visible near the storms.

Slide 12 - Alaska

As noted on the last two slides,the fine resolution of the SAR allows for the identification of localized wind features, in this case gap flows along southwest Alaska. “Gaps” in the mountains accelerate the winds (the effect is similar to when you put your thumb over the end of a garden hose). These localized high speeds can extend hundreds of kilometers from the land and are a hazard to mariners. In addition the mountains induce an oscillation in the air flow resulting in “lee waves” downwind of the mountains. These waves appear as alternating light and dark bands that are parallel with the topography (these light and dark bans correspond to where the crests and troughs of the waves touch the ocean surface)

Slide 13 - Goni

Next we have an example of a tropical cyclone, one of the most extreme wind events on the planet. Super Typhoon Goni roamed the Western Pacific in the fall of 2020 and on 30 October this 155 kt storm was captured nearly simultaneously by the Japanese Himawari-8 weather satellite and the Radarsat-2 SAR satellite. On the left is a visible light image from Himawari - which captures the reflection of sunlight from the cloud tops. The visible image shows transverse bands of cirrus clouds, a sloping eyewall, and spiral features circling around the clear eye (all indicative of an intense tropical cyclone). On the right, the Radarsat-2 SAR image recorded the backscatter return from the ocean surface and includes patterns of wind streaks, convection cells and rain bands. The SAR produces a unique view of Tropical Cyclones by recording their effects on the ocean surface and in fine detail. This allows for the extraction of information on the storm’s radial wind profile and determination of the location of the radius of maximum winds.

Slide 14

Another view from both the ocean surface and the top of the atmosphere, this time showing some open cell clouds over an area just north of Norway. The Modis True Color image on the right, was collected about 15 minutes before the Sentinel-1 SAR, and looks only hints at the impact of the atmosphere on the ocean surface, In the SAR local variations in wind speed and atmospheric density make these storm cells appear as large crater-like imprints on the SAR image.

Slide 15

From Norway to the Great Lakes, this Sentinel-1 SAR image shows the hexagonal pattern of fine scale Benard convection cells in Thunder Bay along the northern shore of Lake Superior. These cells form because the atmosphere is being “heated from below” by the warmer lake water setting up the convection pattern. Each cell is only a few kilometers across. The wind over the area at the time is very light at less than 3 m/s. Lake ice is also visible in the northeastern part (of the shallower) Black Bay.

Slide 16 Lake Superior 22 April 2022

A strong wind out of the east of around 20 kts produced an interesting set of atmospheric waves around the Keweenaw Peninsula along the South shore of Lake Superior in April 2022. The topography of the peninsula causes an oscillation in the air flow resulting in the lee wave pattern downwind (of the peninsula). But south of the peninsula is the pattern of a nonlinear atmospheric internal wave (aligned perpendicular) to the incoming wind. Both waves are types of gravity waves since the restoring force (which produces the oscillation) is gravity. But the lee wave represents a displacement of the whole air column that produces a standing wave pattern while the nonlinear internal wave forms along the boundary between layers in the atmosphere (called a pycnocline) and shows a pattern of decreasing wavelength from front to back. The internal wave will propagate away along the boundary once the wind speed drops below the wave’s propagation speed.

Slide 17

Individual rain cells are visible here in the southern Gulf of Mexico. Rain cells are characterized by isolated areas of higher winds (in near circular patterns ) and often contain very bright “feather-like” features which are believed to be scattering from atmospheric ice. Lower wind areas around the rain cells allow natural surfactants to appear. These streaks of natural oils appear dark because they damp the capillary wave resulting in a reduction is backscatter. Several small dark blobs along the western side of the image could be oil pollution related.

Slide 18

Also in the southern Gulf of Mexico is the Cantarell Field, one of the largest offshore oil fields in the world. This Sentinel-1 image contains a series of bright dots which are a combination of ships, oil rigs and oil platforms. Their metallic structures are efficient reflectors of radar energy. The dark features are surfactants, principally mineral oil, which damps the formation of capillary waves and results in a smoother ocean surface than otherwise would be present under the same wind conditions. At the time the winds were generally uniform out of the north at less than 5 m/s. The linear features crossing the dark signatures are ship tracks

Slide 19

The prior slides have shown how various atmospheric phenomena leave their signatures on the ocean surface. On this slide we can see that the SAR backscatter can be influenced by ocean currents and the ocean’s surface temperature. The north wall of the Gulf Stream is visible in this Sentinel-1 from April 2023 as the fast moving current creates a boundary with the surrounding ocean that results in a change in surface roughness. In addition the warmer water of the Gulf Stream (when combined with colder overlying air) causes some atmospheric instability resulting in a convection cell pattern visible along the eastern portion of the image.

Slide 20

Any ocean current is capable of producing a near surface boundary within the surrounding water. In this Sentinel-1 image from the Southern Gulf of Mexico several current interfaces (sometimes called Suloys) are visible

Slide 21

Ocean surface swell is caused by storm systems or by wind blowing consistently across the ocean surface over a period of hours to days. The resulting swell waves with wavelengths on the order of a few hundred meters to around a kilometer, modulate the capillary wave field and produce a pattern in SAR imagery resembling the grooves on a record album. In this Sentinel-1 image over Point Reyes California, the swell waves that are propagating towards the south east can be seen refracting around the promontory as they make their way into Drakes Bay .

Slide 22 (In progress) Internal Waves Cape Cod

The ocean can contain “layers” of different density - caused by difference in either temperature (a thermocline) or salinity (a halocline). Waves can propagate along these sub surface interfaces. And although these waves are contained within the water column, they produce currents that set up convergent and divergent zones on the surface thereby altering the distribution of the capillary waves and allowing them to become visible. These “internal waves” are dominated by nonlinear physics allowing them to propagate for days and hundreds of kilometers.

Slide 23

Two images over the mouth of the Columbia River in the US Pacific Northwest. In the left image a river plume is visible where lighter (fresher) water from the river has been pushed into the ocean creating a boundary with the surrounding sea water. In addition the pulse of water has displaced the oceans thermocline and generated an internal wave packet.
In the right image a strong swell wave pattern is visible propagating towards the coast. But look closely and you can see the SAR imagery becomes a little blurry immediately adjacent to the land. This is the result of wave breaking which randomizes the phase of the returned radar signal and prevents a clear image from being produced in this area.

Slide 24 - Sea Ice Greenland

From the US West Coast we move to southeastern Greenland. Here an ocean swell field is visible but this time the swell waves are propagating northward. The sea ice in this area is thin enough that the swell remains clearly visible under the ice. Between this area of thin ice and the coast is a region of thicker (and rougher) first year ice which appears brighter. The first year ice is composed of individual flows which can be distinguished with the fine resolution SAR.

Slide 25 - Canadian Coast Old Ice

This Sentinel-1 image from March 2023 shows the multi-year ice pack along part of the Canadian archipelago near the Beaufort Sea. The long dark bands in the ice are leads, gaps in the ice opened up by the ice’s motion which allows water to become exposed. The water is smooth, and therefore dark compared to the rough surface of the ice. Narrow bright bands are also visible throughout the scene. These are pressure ridges where the ice motion has pushed the ice together resulting in areas of enhanced roughness. The multi-year pack is composed of individual flows giving the ice a somewhat mottled appearance

Slide 26 Beaufort Sea 23 Oct 2023

From old ice near the coast to the ice edge early in a new season, our next view of sea ice comes from a Sentinel-1 image acquired over the Beaufort sea in October 2022. The image shows several ice types including new ice, young ice and first year ice. New ice is exceptionally smooth and appears nearly black since it reflects away almost all the incident radar energy. The Young ice contains a variety of patterns because it is thin and is easily influenced by both the ocean currents and the wind. The remainder of the ice signatures are from first year ice which show a more complex pattern of individual flows and brakes.

Slide 27

Just two months earlier the sea ice pattern on this same patch of the Beaufort Sea looks very different as the prior year’s ice cover has melted and thinned over the summer months. The ice is being broken up from the combined effects of winds, currents and wave action leading to a very dispersed ice edge.

Slide 28

Back to the Great Lakes - this time Lake Erie which regularly has ice cover during the winter. In this image from January 2022 a variety of ice types are visible, from rough (bright) ice in the western end, which changes over to an area composed of (darker) individual ice floes (with smoother surfaces) which then changes over to a region of recent freeze up that shows a collection of large cracks (or ice leads) .

Slide 29

This slide shows Lake Erie a day later - where the skies were clear and the MODIS Spectroradimeter aboard NASA’s Aqua spacecraft was able to capture a true color image of the lake ice. The optical image shows snow covered ice at the western end and many of the same leads visible in the SAR.

Slide 30

If we zoom into the western side of the Lake Erie SAR image, we can see an interesting collection of long thin bright linear features in the ice. These are the tracks of ships that have traversed the pack, breaking the ice with the water re-freezing after their passage.

Slide 31

As presented in this course, SAR imagery over the ocean captures the unique signatures of a wide range of both atmospheric and oceanographic phenomena. These signatures can be complex at times making the imagery difficult to interpret. Sometimes outside information is required to better understand the in situ conditions and the potential sources of the observed signatures. This Sentinel-1 image from February 2023 is just such an example. So whats going on? In the days prior, further to the northeast, high winds flowing south out of the Gulf of Tehuantepec produced an upwelling event in the ocean. This intrusion of cooler water can be seen in sea surface temperature imagery. The interaction of the cool water and the surrounding ocean set up a series of current boundaries and created the observed ocean vortices . In addition a weak atmospheric front is present in the area visible along the eastern edge of the scene. The various oceanographic and atmospheric conditions combined to produce a complex pattern of returns.

Slide 32

Here is the same Sentinel-1 image with the location of specific features labeled.

Slide 33

Some additional resources for exploring SAR imagery are listed here. The links to NRCS png images included on many of the sides are from the SAR Ocean Surface Winds websites developed and maintained by NOAA STAR. Back in 2004 NOAA published its Synthetic Aperture Radar Marine Users Manual which has more details on the types of information that can be extracted from the SAR imagery over the ocean. And while its publication date precludes it from containing information from the most recent SAR system (S1 and RS2 and RCM) but it remains pertinent since underlying physics behind the phenomena and their manifestation in SAR imagery remains unchanged by time. Many interesting SAR related case studies are available at the CIMISS satellite blog and ESA’s Ocean Virtual Laboratory is an excellent tool for looking at many different types of satellite imagery including Sentinel-1.

Slide 34

Sentinel-1 imagery are available from the Coastwatch Data Portal. Go to the L1/L2 Spatial Search Tab and then click the radio button for S1A or S1B NRCS. The imagery are available in NetCDF format alongside the PNG images. The CW archive goes back to the fall of 2018 and covers the US EEZ (including Alaska Hawaii) as well as the Caribbean and part of the North Atlantic