Satellite remote sensing extraction of debris flow areas

Debris flows are torrents formed when heavy rain or floods saturate and dilute loose, sandy soil in mountainous areas. They are widely distributed in regions with unique topography and landforms around the world, and their destructive power is more extensive and severe than that of individual collapses, landslides, or floods. Debris flows pose a threat to human life and property through their destructive and burying mechanisms.

In the past, field geological disaster research or investigation was difficult to conduct in sparsely populated areas with high mountains and dangerous roads. However, remote sensing survey methods have the advantages of wide field of view, fast data acquisition, continuous observation and no terrain restrictions, and have been widely used in the field of debris flow disaster investigation in recent years.

△Debris flow remote sensing monitoring

Debris flow extraction can be performed on superpixels obtained by the SLIC method using relevant spectral indices to extract debris flow regions.

First, the image is segmented using the SLIC method. This involves determining the number of segmented regions and evenly distributing seed points within the image based on these regions. Next, seed points are reselected within the neighborhood of each assigned seed point, and the gradient values ​​of all pixels in that neighborhood are calculated. The pixel with the smallest gradient is then reassigned as a seed point to avoid it being set as a contour boundary point. A clustering label is assigned to each pixel based on a clustering metric. Since each pixel is searched by multiple seed points, each pixel has a distance to its surrounding seed points. The seed point with the minimum distance is selected as the cluster center for that pixel. The region centers are then recalculated, and clustering is iteratively optimized based on the metric until the error converges.

（a）K=100

（b）K=300

（c）K=500

△Region segmentation based on SLIC method

After the segmentation regions are generated, the vegetation index NDVI, the improved water index PNDWI, and the spectral average value of each band are calculated for each segmented region. Based on the visual interpretation area of ​​the debris flow region on the image, the range of values ​​for the vegetation index, water index, and spectral values ​​of each band in the debris flow region is determined, and a threshold is set to monitor the debris flow region in the study area.

The area of ​​debris flow information in Sanyanyu, Zhouqu County, Gannan Tibetan Autonomous Prefecture, Gansu Province, obtained by inversion based on this method is shown in the figure below.