10:45 AM - 12:15 PM
[ACC25-P11] Visualization of snow distribution by means of SAR and strongly-reflecting ground objects.
Keywords:SAR, Snow depth
In the Tohoku, Joetsu and Hokuriku regions in Japan, local communities have long been adapted to heavy snowfall in winter. As global warming progresses, the annual amount of snowfall will decrease, but extreme snowfall events will occur at higher rates. In addition to traffic disruption, this can lead to snow and ice disasters such as avalanches and disturbance on electrical facilities. The heterogeneity of annual snowfall might require complex management of water resources in dams and rivers during snowmelt.
To solve these issues, the spatial distribution of snow depth is needed. A potentially-effective tool for that is space bourne Synthetic Aperture Radar (SAR). The backscatter coefficient, which indicates the strength of microwaves scattered back in the direction of the satellite, varies greatly depending on the depth of snow cover and the amount of moisture. The authors' research has revealed that C-band SAR microwaves respond to snow cover at depths of 25 cm to 50 cm, where corner reflector (CR) reflected microwave above 0 dB in snow-less conditions to below -5 dB after snow accumulation.
There are innumerable such structures on the ground that reflect microwaves strongly. By finding such strong-reflecting objects (SROs) in advance and capturing changes in backscattering coefficients due to snow accumulation, it will be possible to determine snow accumulation with high sensitivity over a wide area. The purpose of this study is to investigate the effectiveness of using SROs to determine the snow depth over a wide area by spatially analyzing the decrease in backscattering intensity (difference from those on Nobember).
The data in this study were observed by Sentinel-1, a C-band SAR operated by the European Space Agency, in ScanSAR (IW: Interferometric Wide) mode in 2017-2023. The observation area covered almost the entire eastern Japan (Hokkaido to Tokyo) with relative orbit number of 46. The local incidence angle in the coverage is about 30-45°.
First, pixels are extracted as SROs under a condition that the average value of the speckle noise-removed backscatter coefficient observed in November was greater than 0 dB. The difference between the speckle-noise-removed backscatter coefficient in winter seasons and the average pixel value in November (assuming that the paddy fields were not irrigated before snowfall) was calculated. The backscatter coefficient in each SRO was analyzed spatially by applying multiple filter processing to determine how much the backscatter coefficient decreases during the winter season.
The analysis revealed that SROs that responded highly to snow cover were found over a wide area from Hokkaido to the Kanto region. On the other hand, there were many SROs that did not respond to snow cover, possibly depending on land cover conditions. The future issue is how to appropriately select SROs that correspond to snow cover changes, transform them from point distributions to area distributions, and visualize them in map form. We will present some examples of how to develop maps that take land cover into account and are appropriate for each use.
To solve these issues, the spatial distribution of snow depth is needed. A potentially-effective tool for that is space bourne Synthetic Aperture Radar (SAR). The backscatter coefficient, which indicates the strength of microwaves scattered back in the direction of the satellite, varies greatly depending on the depth of snow cover and the amount of moisture. The authors' research has revealed that C-band SAR microwaves respond to snow cover at depths of 25 cm to 50 cm, where corner reflector (CR) reflected microwave above 0 dB in snow-less conditions to below -5 dB after snow accumulation.
There are innumerable such structures on the ground that reflect microwaves strongly. By finding such strong-reflecting objects (SROs) in advance and capturing changes in backscattering coefficients due to snow accumulation, it will be possible to determine snow accumulation with high sensitivity over a wide area. The purpose of this study is to investigate the effectiveness of using SROs to determine the snow depth over a wide area by spatially analyzing the decrease in backscattering intensity (difference from those on Nobember).
The data in this study were observed by Sentinel-1, a C-band SAR operated by the European Space Agency, in ScanSAR (IW: Interferometric Wide) mode in 2017-2023. The observation area covered almost the entire eastern Japan (Hokkaido to Tokyo) with relative orbit number of 46. The local incidence angle in the coverage is about 30-45°.
First, pixels are extracted as SROs under a condition that the average value of the speckle noise-removed backscatter coefficient observed in November was greater than 0 dB. The difference between the speckle-noise-removed backscatter coefficient in winter seasons and the average pixel value in November (assuming that the paddy fields were not irrigated before snowfall) was calculated. The backscatter coefficient in each SRO was analyzed spatially by applying multiple filter processing to determine how much the backscatter coefficient decreases during the winter season.
The analysis revealed that SROs that responded highly to snow cover were found over a wide area from Hokkaido to the Kanto region. On the other hand, there were many SROs that did not respond to snow cover, possibly depending on land cover conditions. The future issue is how to appropriately select SROs that correspond to snow cover changes, transform them from point distributions to area distributions, and visualize them in map form. We will present some examples of how to develop maps that take land cover into account and are appropriate for each use.