11:00 AM - 11:15 AM
[STT39-02] Effect of Snow Cover on L-band Interferometric SAR
—Existence of a Paradox Inversely Correlated with Snow Depth—
Keywords:Interferometric SAR, L-band, snow cover
Introduction
GSI has been detecting and monitoring crustal and ground deformation nationwide by interferometric SAR using the ALOS-2 satellite equipped with L-band SAR. While snow-covered areas often show decorrelation at all, they may also show good coherence, which may provide information on snow depth and snow conditions.
This presentation introduces one aspect of the complexity of snow cover effects in L-band interferometric SAR by providing examples of SAR interferograms that are inversely correlated with snow cover depth due to snow melt after snowfall.
Example of the Isawa Plain
Figure (left) shows an ALOS-2 SAR interferogram of the Isawa Plain, Iwate Prefecture from February 3 to March 17, 2022. The Isawa Plain is a vast fan-shaped area stretching from west to east, with a height difference of more than 200 meters. The phase values of the SAR interferograms show a gradual gradation along the slope, with displacement away from the satellite to the east, and if it is vertical displacement, the east side is more subsided, and the amount is more than 10 cm.
According to the nearby AMeDAS data, there was more than 10 cm of snow on February 3 in the primary image and no snow on March 17 in the secondary image, and the temperature on February 3 was about 0 °C and sunshine a few hours before the image was taken.
The snow melted between February 3 and March 17, which would imply that the snow cover was deeper on February 3 on the eastern side, the end of the fan. However, the topography and seasonal winds indicate that the snow cover should be deeper on the west side, which is the opposite of the SAR interferogram.
Example of the Tokachi Plain
Figure (right) shows an ALOS-2 SAR interferogram of the Tokachi Plain, Hokkaido (November 19, 2016-February 25, 2017). In this image, the southern side of the plain is more than 150 m higher than the northern side. Along that slope, the phase values of the SAR interferogram show a gradual gradation, with the south side more displaced away from the satellite, and if it is a vertical displacement, the south side is more subsided, and the amount is more than 10 cm. There is no snow cover in the primary image in November, and more than several tens of centimeters of snow cover in the secondary image in February, so the phenomenon is the same as that of the Isawa Plain. The weather in the secondary image on February 25 was clear from the previous day and the temperature was about -1 °C.
As shown in the right side of the figure, the higher the elevation, the deeper the snow cover is, the SAR interferogram and the snow cover depth are opposite.
What is Snow Cover Paradox?
For L-band microwaves, dry snow (ice grains + air) at low temperatures and immediately after snowfall is "transparent," and microwaves penetrate the snow cover and are observed by SAR as reflections from the underlying soil. However, if the volume scattering of microwaves occurs in wet snow due to melting (ice grains + melted water + air) caused by sunshine, etc., the apparent increase in the reflective surface of microwaves can be explained. In both examples, there was sunshine after the snowfall and the temperature was not too low. If the lower elevation side is warmer than the higher elevation side, the snow melts and turns into wet snow, and microwaves are reflected by volume scattering closer to the surface according to the degree of melting, the phenomenon can be explained.
Note that the plains shown in Figure are relatively uniformly covered with rice paddies and fields, and thus SAR interferograms showing good coherence and uniform gradients are obtained due to uniform melting progress under certain meteorological conditions. However, differences in vegetation and other factors can also cause very different InSAR phase changes, and the complexity of the effect of snow cover on L-band InSAR is not straightforward.
GSI has been detecting and monitoring crustal and ground deformation nationwide by interferometric SAR using the ALOS-2 satellite equipped with L-band SAR. While snow-covered areas often show decorrelation at all, they may also show good coherence, which may provide information on snow depth and snow conditions.
This presentation introduces one aspect of the complexity of snow cover effects in L-band interferometric SAR by providing examples of SAR interferograms that are inversely correlated with snow cover depth due to snow melt after snowfall.
Example of the Isawa Plain
Figure (left) shows an ALOS-2 SAR interferogram of the Isawa Plain, Iwate Prefecture from February 3 to March 17, 2022. The Isawa Plain is a vast fan-shaped area stretching from west to east, with a height difference of more than 200 meters. The phase values of the SAR interferograms show a gradual gradation along the slope, with displacement away from the satellite to the east, and if it is vertical displacement, the east side is more subsided, and the amount is more than 10 cm.
According to the nearby AMeDAS data, there was more than 10 cm of snow on February 3 in the primary image and no snow on March 17 in the secondary image, and the temperature on February 3 was about 0 °C and sunshine a few hours before the image was taken.
The snow melted between February 3 and March 17, which would imply that the snow cover was deeper on February 3 on the eastern side, the end of the fan. However, the topography and seasonal winds indicate that the snow cover should be deeper on the west side, which is the opposite of the SAR interferogram.
Example of the Tokachi Plain
Figure (right) shows an ALOS-2 SAR interferogram of the Tokachi Plain, Hokkaido (November 19, 2016-February 25, 2017). In this image, the southern side of the plain is more than 150 m higher than the northern side. Along that slope, the phase values of the SAR interferogram show a gradual gradation, with the south side more displaced away from the satellite, and if it is a vertical displacement, the south side is more subsided, and the amount is more than 10 cm. There is no snow cover in the primary image in November, and more than several tens of centimeters of snow cover in the secondary image in February, so the phenomenon is the same as that of the Isawa Plain. The weather in the secondary image on February 25 was clear from the previous day and the temperature was about -1 °C.
As shown in the right side of the figure, the higher the elevation, the deeper the snow cover is, the SAR interferogram and the snow cover depth are opposite.
What is Snow Cover Paradox?
For L-band microwaves, dry snow (ice grains + air) at low temperatures and immediately after snowfall is "transparent," and microwaves penetrate the snow cover and are observed by SAR as reflections from the underlying soil. However, if the volume scattering of microwaves occurs in wet snow due to melting (ice grains + melted water + air) caused by sunshine, etc., the apparent increase in the reflective surface of microwaves can be explained. In both examples, there was sunshine after the snowfall and the temperature was not too low. If the lower elevation side is warmer than the higher elevation side, the snow melts and turns into wet snow, and microwaves are reflected by volume scattering closer to the surface according to the degree of melting, the phenomenon can be explained.
Note that the plains shown in Figure are relatively uniformly covered with rice paddies and fields, and thus SAR interferograms showing good coherence and uniform gradients are obtained due to uniform melting progress under certain meteorological conditions. However, differences in vegetation and other factors can also cause very different InSAR phase changes, and the complexity of the effect of snow cover on L-band InSAR is not straightforward.