The U.S. has suffered from devastating wildfires over the last few years as global temperatures rise and weather patterns change, making the otherwise natural phenomenon especially unpredictable and severe. To help out, Stanford researchers have found a way to track and predict dry, at-risk areas using machine learning and satellite imagery.
Currently the way forests and scrublands are tested for susceptibility to wildfires is by manually collecting branches and foliage and testing their water content. It’s accurate and reliable, but obviously also quite labor intensive and difficult to scale.
Fortunately, other sources of data have recently become available. The European Space Agency’s Sentinel and Landsat satellites have amassed a trove of imagery of the Earth’s surface that, when carefully analyzed, could provide a secondary source for assessing wildfire risk — and one no one has to risk getting splinters for.
This isn’t the first attempt to make this kind of observation from orbital imagery, but previous efforts relied heavily on visual measurements that are “extremely site-specific,” meaning the analysis method differs greatly depending on the location. No splinters, but still hard to scale. The advance leveraged by the Stanford team is the Sentinel satellites’ “synthetic aperture radar,” which can pierce the forest canopy and image the surface below.
“One of our big breakthroughs was to look at a newer set of satellites that are using much longer wavelengths, which allows the observations to be sensitive to water much deeper into the forest canopy and be directly representative of the fuel moisture content,” said senior author of the paper, Stanford ecoydrologist Alexandra Konings, in a news release.
The team fed this new imagery, collected regularly since 2016, to a machine learning model along with the manual measurements made by the U.S. Forest Service. This lets the model “learn” what particular features of the imagery correlate with the ground-truth measurements.
They then tested the resulting AI agent (the term is employed loosely) by having it make predictions based on old data for which they already knew the answers. It was accurate, but most so in scrublands, one of the most common biomes of the American west and also one of the most susceptible to wildfires.
You can see the results of the project in this interactive map showing the model’s prediction of dryness at different periods all over the western part of the country. That’s not so much for firefighters as a validation of the approach — but the same model, given up to date data, can make predictions about the upcoming wildfire season that could help the authorities make more informed decisions about controlled burns, danger areas, and safety warnings.
The researchers’ work was published in the journal Remote Sensing of Environment.
SpaceX has received authorization from the Federal Aviation Administration (FAA) to fly suborbital missions with its Starship prototype spacecraft, paving the way for test flights at its Boca Chica, Texas site. SpaceX has been hard at work readying its latest Starship prototype for low-altitude, short duration controlled flight tests, and conducted another static engine fire test of the fourth iteration of its in-development spacecraft earlier today.
Officially, the FAA has granted SpaceX permission to conduct what it terms “reusable launch vehicle” missions, which essentially means that the Starship prototype is now cleared to take-off from, and land back at, the launch site SpaceX operates in Boca Chica. The Elon Musk-led space company has already conducted similar tests, but previously used its ‘Starhopper’ early prototype, which was smaller than the planned production Starship, and much more rudimentary in design. It was basically used to prove out the capabilities of the Raptor engine that SpaceX will use to propel Starship, and only for a short hop test using one of those engines.
Since that flight last year, SpaceX has developed multiple iterations of a full-scale prototype of Starship, but thus far they haven’t gotten back to the point where they’re actively flying any of those. In fact, multiple iterations of the Starship prototype have succumbed during pressure testing – though SN4, the version currently being prepared for a test flight, has passed not only pressure tests, but also static test fires of its lone Raptor engine.
The plan now is to fly this one for a short ‘hop’ flight similar to the one conducted by Starhopper, with a maximum altitude of around 500 feet. Should that prove successful, the next version will be loaded with more Raptor engines, and attempt a high altitude test launch. SpaceX is quickly building newer version of Starship in succession even as it proceeds with testing the completed prototypes, in order to hopefully shorten the total timespan of its development.
There’s something of a clock that SpaceX is working against: It was one of three companies that received a contract award from NASA to develop and build a human lander for the agency’s Artemis program to return to the Moon. NASA aims to make that return trip happen by 2024, and while the contract doesn’t necessarily require that each provided have a lander ready in that timeframe, it’s definitely a goal, if only for bragging rights among the three contract awardees.
A team of researchers from the West Virginia University (WVU) Rockefeller Neuroscience Institute (RNI), along with WVU’s Medicine department and staff from Oura Health have developed a platform they say can be used to anticipate the onset of COVID-19 symptoms in otherwise healthy people up to three days in advance. This can help with screening of pre-symptomatic individuals, the researchers suggest, enabling earlier testing and potentially reducing the exposure risk among frontline healthcare and essential workers.
The sudsy involved using biometric data gathered by the Oura Ring, a consumer wearable that looks like a normal metallic ring, but that includes sensors to monitor a number of physiological metrics, including body temperature, sleep patterns, activity, heart rate and more. RNI and WVU Medical researchers combined this data with physiological, cognitive and behaviroral biometric info from around 600 healthcare workers and first responders.
Participants in the study wore the Oura Ring, and provided additional data that was then used to develop AI-based models to anticipate the onset of symptoms before they physically manifested. While these are early results from a phase one study, and yet to be peer-reviewed, the researchers say that their results showed a 90 percent accuracy rate on predicting the occurrence of symptoms including fever, coughing, difficulty breathing, fatigue and more, all of which could indicate that someone has contracted COVID-19. While that doesn’t mean that individuals have the disease, a flag from the platform could mean they seek testing up to three days before symptoms appear, which in turn would mean three fewer days potentially exposing others around them to infection.
Next up, the study hopes to expand to cover as many as 10,000 participants across a number of different institutions in multiple states, with other academic partners on board to support the expansion. The study was fully funded by the RNI and their supporters, with Oura joining strictly in a facilitating capacity and to assist with hardware for deployment.
Many projects have been undertaken to see whether predictive models could help anticipate COVID-19 onset prior to the expression of symptoms, or in individuals who present as mostly or entirely asymptomatic based on general observation. This early result from RNI suggests that it is indeed possible, and that hardware already available to the general public could play an important role in making it possible.