Flood Monitoring According to Sentinel-1 Radar Imagery
By Igor Garkusha
California has been in a drought for over 5 years. This year’s rainy season unexpectedly brought California the strong showers and thunderstorms it desperately needed. As a result of this sudden wetness on the dried-out state, many parts of California were flooded and over saturated.
Monitoring flooded areas and tracking the water levels in reservoirs are very important, due to their potential impacts. Tracking this can be easier using remote sensing, i.e. space images made by both optical and radar scanners. Unfortunately, in bad weather conditions, the monitoring process can be impeded by dense clouds as they greatly reduce the value of the optical images. In such case, radar imagery becomes more useful than optical. Radar works by sending out a signal and it reflects back to the satellite from the Earth, so it does not depend on cloudiness or lighting.
To evaluate a flooded area, we used radar imagery from the European Space Agency’s Sentinel-1A/B GRDH (Ground Range Detected High Resolution), which provides data intervals of 6–12 days. The imagery analyzed was from February 2nd, 2017 and February 24th, 2017 (Figure 1).
On these images, the increase in the number of water bodies can be seen with the naked eye.
On February 10th, the Anderson Reservoir near Morgan Hill was reported to be filling up too quickly. Water began overflowing into the nearby Coyote Creek when the amount of rain filling the reservoir became greater than the amount able to be released through the outlet.
On Figure 2, an enlarged fragment of the lake and the boundaries of water bodies can be seen.
The water bodies and waterlogged soils (shown in blue) are identified by the calibrated radar data of the VV-polarization channel. As the cut-off border, the value σ < -17 dB has been taken. On the second image, it is clearly visible that the north part of the lake has greatly increased in size.
An enlarged fragment of central California is shown on Figure 3.
The number of flooded agricultural land has increased significantly from the rain as well. In parts of Manteca and Modesto, California, the approximate area of water bodies has changed from 1012 km2 to 1130 km2.
In conclusion, we have found that radar imagery is quite effective for solving a variety of tasks. In this example, with Sentinel-1, we can locate bodies of water and track their changes, despite the bad weather conditions. This will allow better decision making and tracking potential hazards quickly so people are prepared.
Sentinel-1 imagery is a free open data available for viewing, analysis and downloading in LandViewer.
