Technology

Photo by USGS on Unsplash

Technology

Photo by USGS on Unsplash

SkyTruth uses satellite imagery, data science, custom software development, and technical capacity building to advance environmental protection across the world. And from the beginning, we’ve given our technology and services away for free to the conservation community.

Technology for impact

When SkyTruth was founded in 2001, we had to purchase satellite data from civilian satellites like Landsat and SPOT, and we relied on manual inspection to understand our changing world, one satellite image at a time.

Now, we use free data from a host of environmental monitoring satellites operated by NASA and the European Space Agency (ESA), as well as data we buy from commercial satellite providers. And we analyze that data using machine learning, computer vision, and other automated methods that help us monitor the world’s environment at a global scale.

Satellites: environmental protectors

Satellite imagery has revolutionized the whole way we analyze things; it’s transformed the way the Earth is pictured.

Technology spotlights

SkyTruth’s Cerulean application uses Sentinel-1 satellite data from ESA, an image processing pipeline we built in Google Cloud Platform, computer vision, and a custom user interface to help users identify and eliminate marine oil pollution. There are two key parts to Cerulean: oil slick detection, and source detection.

Slick detection applies a computer vision model to data from ESA’s Sentinel-1 satellite. Sentinel-1 collects synthetic aperture radar (SAR) images, which can penetrate clouds and distinguish oil slicks on open water. Our model automatically scans Sentinel-1 data, outlines potential oil slicks, and assigns a probability to each detection.

When slicks are detected, the next step is identifying which vessels or offshore oil platforms the slicks came from. To identify vessels, we use vessel location data from the Automatic Identification System (AIS), which is a location broadcast system used by large ocean vessels for safety purposes. We use AIS data to compare the timing and trajectories of vessels and oil slicks, which helps narrow down the list of vessels potentially associated with a given slick. Similarly, we use offshore oil infrastructure locations from Global Fishing Watch to identify platforms that coincide with oil slicks, enabling us to compare platform locations with the origin point of oil slicks.

The results of these models are available via a programmatic API and a custom frontend application that lets users map, evaluate, and find the source of oil slicks across the ocean.

Inambari uses satellite data from ESA, a machine learning model that we designed to detect mining-related deforestation, and a custom web mapping application called SkyTruth Alerts to aid monitoring and enforcement of unregulated mining in the Peruvian Amazon.

Our machine learning model detects new mining activity in Sentinel-1 and Sentinel-2 satellite data from the European Space Agency. The model runs automatically once per month using Google’s Cloud Computing and Earth Engine platforms, which allow us to process massive amounts of imagery to pinpoint new mining activity.

Those mining detections appear in a custom map application that shows users the progression of deforestation over time, cloud free radar and multispectral satellite imagery for visual inspection of new mining detections, protected area boundaries, and other critical base layers that enable our users to detect and verify new mining activity in Peru.

Check out our Inambari page to read more about the project, our partners, and its impacts.

To understand the legacy of mountaintop removal mining (MTM) in Appalachia, SkyTruth uses many of its core tools – advanced remote sensing methods, ecological modeling, and a 50-year satellite archive – to map the progression of coal mining as well as the landscape’s recovery.

MTM is a form of surface mining where mountain summits are literally blasted away, with the pulverized rock deposited in nearby valleys. SkyTruth created the first comprehensive dataset showing the annual progression of MTM in Central Appalachia, dating back to 1985.

To do this, we use satellite imagery from the Landsat program, which has collected imagery of the earth’s surface for over 50 years. Our model detects persistent losses of vegetation in coal mining regions of Appalachia using NDVI and other satellite-derived measures of vegetation greenness. We also use time series of digital elevation data from satellite and shuttle missions to identify and quantify where mining has permanently changed the region’s topography.

We’ve also begun to use satellite data and statistical modeling to monitor ecosystem recovery on former mining sites. Using time series of NDVI data from satellites, we can compare revegetation rates on former mine sites with nearby, healthy forests, enabling us to track and assess the progress of current recovery efforts in the region. Read more about this work and the methods behind it on our blog.

SkyTruth Alerts is a free environmental mapping and monitoring platform, enabling users to view satellite imagery, map various environmental data layers and pollution events, and receive email alerts for emerging environmental threats and events. What’s under the hood of our mapping platform?

SkyTruth Alerts uses the Google Maps API to support its dynamic mapping capabilities, and the application itself is hosted on Google’s App Engine, which is a scalable cloud service for hosting web applications. 

The data available in Alerts comes from many sources. The satellite imagery is streamed from Google Earth Engine in the form of raster tiles. The various map layers come from a variety of sources – some we download and host in our own PostgreSQL databases that can stream data to the map, some we pull directly from various third party APIs hosted by data providers like Global Forest Watch, some come from streaming vector tiles generated from Google Earth Engine, and others – like our spill alerts – we scrape from state and federal databases to make them more accessible to the public.

We also provide an email alerting service, where users can specify an area of interest and an environmental incident to get automatically alerted to new developments. 

This is just a sample of the capabilities and technology behind Alerts. We’d love it if you checked it out and told us what you think!