Page 512 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

Fig. 5. Slope map after thresholding

3.2 Preprocessing
.

The noise in the input images was reduced using a mean filter [2 * 2] to reduce SPECKLE noise in SAR data [18]. When
downloading the landsat8 image, the image with the least amount of cloud coverage for the period 12122116 to 123122112 is
used, and a MEDIAN filter is used to reduce image noise.

.
3.3 Processes
Post-flood SAR image was divided into pre-flood SAR image, and a threshold was applied to the result of this division to detect
flood-induced changes in the region.
The threshold should be taken from the NDWI image obtained by calculations on two bands B3 (green), B5 (swir) (Eq. 1) [14 ,21]
to identify areas with still water and rivers. Fig. 6 shows the NDWI image obtained from Landsat8 data.

(green−swir)
NDWI= (1)
(green+swir)

Fig. 6. NDWI image

.
3.3 postprocessing:
Applying the threshold to the NDWI image obtained from LANDSAT8 should identify areas with stagnant still and snow at the
tops of the peak.

The SRTM data is used to create the slope map, which is then subjected to a 19 percent threshold. Areas obtained from this
operation are removed from SENTINEL-1 images [21]. Following these steps, a majority filter [22] is applied to the final images
to remove any remaining single pixels.

3 . Results
Fig. 2 shows the NDWI images befor and after thresholding.

5