Sentinel-1 Interferometric Coherence Image Services

This map displays the COH12 VV data for the summer season (June/July/August) in France.

Follow the prompts to explore the map. You can also pan and zoom to take a closer look if you see something of interest. Click on the map to view pixel values and more information about the source rasters.

Note that the cities show up clearly, dotted around the landscape. Urban areas typically have high coherence, as the built structures comprising cities tend to be stable, effective scatterers.

Paris is a very large bright spot on the map.

The agricultural regions outside the city display a field-by-field patchwork of coherence values, but generally have much lower coherence than the urban areas.

During the growing season, agricultural lands undergo constant and significant change that results in decorrelation of the phase from one acquisition to the next. Changes in crop growth stages and activities around tilling and harvesting can all contribute to this decorrelation.

This urban-rural difference is much less pronounced during the winter months in areas with strong seasonality.

The agricultural fields are still patchy, but have much higher coherence in the winter than they did in the summer. The settlements amidst the fields are much less obvious in the winter season.

Toggle between the winter (December/January/February) and summer (June/July/August) seasons to compare the coherence values across the country.

The coherence values of the rural areas are not as different from the urban areas during the winter, when many of the fields are either planted with slow-growing winter grain or cover crops, or left fallow.

There is a persistent dark spot in the middle of the country. This dark area becomes even more obvious during the winter season.

This patch with consistently low coherence values is located just south of Orleans.

The Loire Valley is densely vegetated, with extensive forests, orchards and vineyards.

Areas with dense and complex vegetation typically have low coherence. This can be problematic if you're interested in generating InSAR products in densely vegetated regions, particularly when using SAR sensors with relatively short wavelengths.

Notice how low the Sentinel-1 coherence values are in the Amazon rainforest. Click on the map to view coherence values for the darkest areas in northern Brazil.

Sentinel-1 has a wavelength of about 5.6 cm, which allows some penetration of dense forest canopy, but the signal will rarely make it through to the forest floor. Forest canopy tends to have poor correlation due to changes in vegetative structure and wind-driven movement over time. This makes it very difficult to use Seninel-1 for InSAR in these areas.

SAR datasets collected with L-band sensors (~25 cm wavelength), such as the  upcoming NISAR mission , may provide better results in densely forested regions.

In South America, the coherence values are even lower in the December/January/February season (southern hemisphere summer) than in June/July/August. The Amazon, however, has consistently low coherence values in all seasons.

Mountain ranges are not particularly easy to pick out based on summer coherence values.

Some ranges may have higher coherence than surrounding areas, especially if they are rocky or sparsely vegetated. Others may exhibit similar or lower coherence compared to surrounding areas.

It's a different story in the winter, however.

Mountain ranges, especially those with very high elevation, generally have more variable snow and ice conditions from one image to the next than the surrounding areas during the winter. This results in low coherence values, clearly illustrated by very stark contrasts in the winter coherence maps.

This map displays the AMP data for VV and VH polarizations.

Follow the prompts to explore the map. You can also pan and zoom to take a closer look if you see something of interest. Click on the map to view pixel values and more information about the source rasters.

Calm surface water generally has very low backscatter. Water effectively reflects the SAR signal, but because the signal arrives at an oblique angle to the surface, the signal bounces off in the other direction. Very little of that sent signal makes it back to the sensor to be measured.

Lac Saint-Jean is a fairly large lake in Québec. It covers more than 400 square miles (1,000 km ² ), and is ice-free for the months included in the summer season (June/July/August) of the GSSICB dataset. The open water of this lake and other smaller lakes, ponds, streams and rivers in the area, stands out clearly as very dark features on the landscape.

On windy days, the surface of the water becomes more rough, and VV backscatter is much higher. In some conditions, it can be difficult to tell the difference between lakes and the surrounding land when looking at a single VV image. Because the GSSICB dataset averages together three months of data, however, even sizeable lakes such as Lac Saint-Jean appear to have uniformly low backscatter.

Very large lakes, such as the Great Lakes in North America, are more likely to have wave action throughout the open water season, so surface conditions are much less uniform than on smaller lakes. Even averaging the seasonal VV backscatter values together will not completely smooth out the signal caused by variable wind conditions, as seen here for Lake Ontario.

The VH polarization is much less sensitive to surface roughness, so does not exhibit the wave-driven variability seen in the VV data. The VH backscatter returns are generally much lower than VV returns overall, however, which makes them more susceptible to noise, such as the burst-level artifacts visible here. Even with the noise, the VH values over the water are still generally much lower than what is seen on land in this area, making it an effective way to identify water pixels.

Click on some of the pixels to view the dB values over the lake in the VH dataset, and compare them to the VV values in the previous view. The VH values (even the brightest ones) are all less than -27 dB, while the VV values range from about -15 to -20 dB.

Great Bear Lake is located in northern Canada, and is generally free of ice from mid-July through September.

From a distance, the summer VV values (which includes data from June, July and August) appear to be uniformly low over the lake.

If you zoom in further, however, and the dynamic range adjustment in the web map stretches the values of the dark pixels along the grayscale, you can see the impact of including acquisitions from when the lake was not yet ice-free.

Zoom out one level and pan around the lake to see the different patterns imprinted on the June/July/August dataset by the inclusion of acquisitions with ice cover.

Again, even though the lake exhibits some variability in the backscatter values, the open water pixels are still much darker than those of the surrounding uplands.

In winter, Great Bear Lake freezes over. The ice is much more effective at scattering the signal back towards the sensor, especially when the ice surface is rough. A lake this large takes some time to freeze over, and the impacts of wind, currents, and human activities can leave their mark. Even though this dataset displays mean values, you can see ice features that are persistent enough through the season to be visible here.

Use the zoom buttons and pan around the lake to explore the ice.

There is one community on Great Bear Lake: Délı̨nę.

In the winter, this community is linked to others along the Mackenzie River by winter roads, including an ice road across the southwestern tip of Great Bear Lake.

While there is an official ice road, other corridors can also be established by activity such as snowmobile travel.

Compare the AMP VV datasets from winter (December/January/February) to spring (March/April/May).

Some of the travel routes across the ice remain the same, while some new routes are established and some old routes are abandoned or adjusted.

Zoom out one level and pan around the lake to see common travel corridors.

Let's return to Paris and take a quick look at the optical imagery northeast of the city.

Notice the urban, agricultural and forested areas in this region. Now add the VV backscatter imagery.

As with coherence values, amplitude values tend to be relatively high over urban areas, particularly in the VV polarization. Agricultural fields show up as a patchwork driven by different conditions from one field to the next.

There are some very dark pixels in the image, including those over the Seine River winding through Paris, and a smattering of lakes and other waterbodies dotting the landscape. There are also two distinct linear dark patches just outside the city.

Compare the backscatter imagery to the optical basemap.

The two dark patches are the runway areas of the Charles de Gaulle Airport (CDG). The tarmac is flat and smooth, and the grounds are flat and lack trees or shrubby vegetation, resulting in very low backscatter. The dark area to the southwest of CDG is another airport.

From a distance, these dark areas may appear to be water, but if you compare the backscatter from the airports to those of lakes, the values are often much lower for the lakes.

In this view, click on the dark patch identified as the Aérodrome des Mureaux (a small airfield) and compare the backscatter values to those from the lakes Étang de la Grosse Pierre and Étang du Gallardon.

Turn off the backscatter layer to compare the features to the optical basemap.

The backscatter patchwork pattern of agricultural landscapes changes from one season to the next.

Adjacent fields can have very different backscatter signatures in the summer, depending on the type of crop being grown and irrigation practices. Some fields may be left fallow or be planted with a cover crop, which also impacts their backscatter signature.

The pattern in the spring can be different, especially if some fields are planted earlier than others, but are similar to the summer patchwork in terms of variability.

Fall and winter conditions are a different story. There is much less variability from one field to the next than there was in the spring and summer.

For backscatter in VV polarization, the changes in the appearance of the fields are driven largely by soil moisture and surface roughness. The fields are much more uniform in the fall and winter seasons than they are in the spring and summer.

The patterns in VH are a bit different. Click through the seasons to look at the differences.

The patchwork of fields is most obvious in the summer season for the VH backscatter. While the VV polarization had similar degrees of patchiness in the spring and summer, the VH backscatter from the fields is much lower overall in the spring than in any other season.

For the VH polarization, the main drivers of difference tend to be complex vegetation, though soil moisture may also play a role. Many of the fields in the spring VH image may have been prepared and even planted, but might not yet have much complex vegetation, resulting in the lower VH backscatter.

In general, the winter season has the highest VH backscatter values in the agricultural areas of this region. This may be due to a combination of slow-growing winter grain crops and the widespread use of cover crops after harvest throughout France.

To wrap up our exploration of backscatter characteristics, let's take a look at some desert areas.

The Sahara and Libyan deserts cover large areas in northeastern Africa. Deserts, like surface water, have extremely low backscatter, but for completely different reasons.

Water has such high dielectric properties that the SAR signal can't penetrate at all, and just bounces off and away from the sensor. Dry sand in the desert has extremely low dielectric properties, and can absorb the incoming signal so well that very little of it is scattered back to its source.

This can be problematic when using backscatter to identify water in arid regions.

In this image, we can see Great Bitter Lake, which is connected to the Red and Mediterranean Seas by the Suez canal. East of the lake is an extremely arid part of the Sinai peninsula. In the mean backscatter products, this sandy expanse actually has even lower backscatter than the lake.

The mean backscatter products are impacted by the presence of ships on the lake, which exhibit high backscatter against the dark background of the lake. This results in a clear indication of the route used by ships navigating the Suez Canal.

To the east, the wavy structures of the sand dunes are visible, as are roads and industrial developments.

Zoom out and pan to explore the dune structure.