Original article was published by Felipe Tambasco on Artificial Intelligence on Medium
How to use Artificial Intelligence to fight global warming?
Climate change is an undeniable threat to humanity. Scientific evidence has been mounting for decades on the warming temperatures that carry severe consequences for all life on earth. At an accelerating rate, the warming of the globe’s atmosphere and oceans spells disaster for otherwise balanced biological processes. Anthropogenic activities which burn fossil fuels and biomass release carbon dioxide (CO2) and other greenhouse gases into the atmosphere; trapped within and raising the temperature on spaceship earth. The consequences of this increase in atmospheric temperatures has resulted in altered climate events — climate being the long-term, expected weather for a geographic region. Consequences from human activities, such as: deforestation for agriculture; ocean acidification from increased CO2; and habitat destruction from expanding cities intersects with extreme weather and warming climates to create widespread, and deadly, events. In the face of this great existential challenge, Earth Sciences is a vital component of tackling this pressing issue.
Thanks to the growing collection of satellite data, data science can also play a role in solving some of the world’s most critical problems; such as forest fires, greenhouse gas emissions, and melting arctic sea ice. Emerging data science practices in the domain of earth sciences are called upon in a field becoming known as Big Earth Data, so that the complex and dynamic interconnectedness of our world can become more palatable for easier decision-making. Exploiting satellite data for prediction, understanding and intervention is what Big Earth Data is all about. Of course, none of this is possible without the right equipment.
The accessibility of satellite datasets is not sparse for aspiring earth data scientists. There are various portals online that offer data products, packaging together visible, infrared, multispectral and hyperspectral data types. The use of this data allows researchers to analyze patterns in land usage, habitat growth (or destruction), temperature changes, as well as characteristics of vegetation (density, coverage, health etc.). Satellite data, acquired through remote sensing techniques, allows for the mapping of visible and invisible phenomenon to track changes in the environment.
Combining this plethora of data with emerging artificial intelligence practices can be the secret to getting an edge of climate change, which many feel is a losing battle for humanity at the moment.
A notable example of publicly available satellite data is the Copernicus Open Access Hub. This portal provides access to image, ozone, and air quality data from the Copernicus Sentinel missions, a European earth observation programme. Registration for these data products is free, and can be accessed through downloads or API.
Artificial Intelligence Applications
Advances in neural networks and image processing has led to high quality computer vision solutions. The ability to train algorithms to recognize the occurrences of patterns and anomalies in image data creates a powerful tool that can be used for good.
Forest Fire Mitigation
Overstory.ai is a Netherlands-based company whose objective is to harness the power of AI and satellite data to mitigate climate change and comprehensively monitor all natural resources on Earth in real-time. They provide insights to governments, utility companies, and NGOs. Services like their vegetation management are arguably essential in a time when the globe’s forests are being exploited or destroyed by out-of-control forest fires. Dry, hot conditions exacerbated by global warming and the inability to secure funding for proper forest protection and management have resulted in hellish conditions for nations like the United States, Australia, Canada, and Brazil and more as stands of vegetation ignite. The world has seen just how bad poor management of forests and increasing temperatures can be, with the mass wildfires that have spread across Australia, Brazil and the United States in 2020.
Overstory.ai uses satellite data and machine learning to provide predictive insights and identify hotspots for intervention to protect biodiversity. Their Forest Intelligence Platform is powered by artificial intelligence that takes into consideration forest size and health, tree species and biomass. Climatic variables are also used to inform on wind directions, air and ground temperature, and drought levels. All this is possible with the direct usage of satellite data.
Overstory.ai is also not the only organization trying to use AI for wildfire applications. Other organizations such as Swedish start-up, Spacept; and Sante Fe-based Descartes Labs also harness artificial intelligence and satellite imagery to monitor for wildfire risks. The latter released their wildfire detection tool in July 2018 which uses RGB visible light and infrared light to detect and track the progression of wildfires in real time. The company’s fire detection system is able to alert firefighters and regional authorities of spreading wildfires, and even provide latitude and longitude coordinates.
This next case study covers a coalition of nonprofit organizations, tech companies, and climate change leaders including: Rocky Mountain Institute, Earthrise Alliance, Blue Sky Analytics, and former U.S. Vice President Al Gore. Climate TRACE, backed by Google, is using artificial intelligence and satellite image processing to monitor worldwide greenhouse gas emissions. The coalition is expected to release a tool that tracks human-caused emissions to their sources using data obtained from the European Space Agency’s Sentinel satellites. The release date for this innovative and ground-breaking tool has not been determined but the team has summer 2021 in their sights.
In a 2019 research paper by Rolnick et al., a collaboration of cross-industry experts and organizations (including Google AI, DeepMind, and Stanford University) provide extensive and thorough research into how machine learning opportunities exist in 13 different climate change solution domains. From city-planning and transportation networks, to climate prediction and geoengineering this paper goes into detail on how artificial intelligence is not only a possibility for fighting the effects of climate change, it’s becoming a necessity. In particular, utilizing the petabytes of data that are generated by satellites for monitoring vegetation and GHG emissions in real-time is becoming more and more apparent as a viable solution for enforcing climate change commitments; mitigating unsustainable land use practices; and identifying GHG ‘hotspots’. Remote sensing, combined with AI solutions like the one Climate TRACE is developing can go a long way for humanity’s fight against the dark days ahead.