Welcome to the Qinsun Instruments Co., LTD! Set to the home page | Collect this site
The service hotline


Related Articles

Product Photo

Contact Us

Qinsun Instruments Co., LTD!
Address:NO.258 Banting Road., Jiuting Town, Songjiang District, Shanghai

Your location: Home > Related Articles > Stanford’s AI can use satellite images to locate risk areas and provide assistance for the upcoming wildfire season

Stanford’s AI can use satellite images to locate risk areas and provide assistance for the upcoming wildfire season

Author:QINSUN Released in:2024-01 Click:34

The current method for testing the sensitivity of forests and shrubs to wildfires is to manually collect branches and leaves, and test their moisture content. This method is accurate and reliable, but it is clearly quite labor-intensive and difficult to scale. Fortunately, researchers have access to other sources of data. The Sentinel and Land satellites of the European Space Agency have accumulated a large amount of Earth's surface images, and after careful analysis, they can provide a second source for assessing wildfire risks.

This is not the first attempt to make such observations from orbit images, but previous work mainly relied on visual measurements of "extremely specific locations", which means that the analysis methods vary greatly depending on the location, making it difficult to scale up. The advanced technology utilized by the Stanford team is the Synthetic Aperture Radar (SAR) of the Sentinel satellite, which can penetrate forest shade and image the surface below.

"One of our major breakthroughs was studying a set of newer satellites that use much longer wavelengths, making the observation results much more sensitive to moisture in the depths of forest shade, directly representing fuel moisture content," said senior author and Stanford ecologist Alexandra Konings in a press release.

The team "fed back" these regularly collected new images since 2016 to a machine learning model along with manual measurements from the US Forestry Administration. This allows the model to "learn" which specific features in the image are associated with ground measurement data. Then, they tested the generated artificial intelligence to make predictions based on old data. It is accurate and accurate in predicting one of the common biological communities in the western United States, as well as one of the easily affected biological communities by wildfires in shrublands.

You can see the results of this project on this interactive map, which shows the model's drought prediction for different periods in the western United States. This is a validation of this method for firefighters - but the same model, after providing new data, can make predictions for the upcoming wildfire season, which can help relevant departments make more decisions in controlling combustion, hazardous areas, and safety warnings.

The research findings of scientists are published in the journal Remote Sensing of Environment.