As the TED AI conference unfolds in Vienna, Ian McCallum, IIASA Novel Data Ecosystems for Sustainability Research Group Leader, brings a fresh perspective on the pivotal role of artificial intelligence in Earth monitoring. He explores how AI can revolutionize our understanding of environmental changes and support sustainable practices in an era marked by significant transitions.
The study and exploitation of AI methods and tools at IIASA is increasingly important throughout the institute as we deal with ever larger amounts of data across our global research domains (i.e., energy, climate, population, environment). In particular we utilize vast quantities of social media data, household surveys, and satellite data, among others, all with rapidly increasing volumes. For example, the European Space Agency (ESA) receives on the order of 15 terabytes of daily data from the Sentinel satellites alone. Hence Large Language Models, Vision Language Models, and machine learning methods in general, are allowing us to handle these growing datasets, better analyze them to gain new insight and improved accuracy.
The IIASA Novel Data Ecosystems for Sustainability (NODES) Research Group in particular exploits efforts around linking citizen science and earth observation data with AI tools for the purposes of monitoring changes to the earth’s surface (i.e., tracking ecosystem health, commodities, deforestation, and more). We have developed the GeoWiki, which is one of the primary citizen science platforms available to track changes on the surface of the planet. We use crowdsourcing techniques applied to vast quantities of satellite imagery to create free and open databases for training and validating machine learning algorithms, which then classify the vast quantities of satellite imagery available each day over the planet. We are currently developing the next generation of this platform via the Evoland initiative.
These efforts are important because many of our models at IIASA, (e.g., GLOBIOM) require such baseline datasets to calibrate their statistics prior to forecasting, along with other global Integrated Assessment Models. In these efforts, we collaborate closely with ESA and Europe’s Copernicus program.
Several recent and ongoing initiatives:
https://esa-worldcereal.org/en
https://esa-worldcover.org/en
https://earthmonitor.org/
https://rapidai4eo.eu/
Recent NODES AI publications
- A Cloud-native Approach for Processing of Crowdsourced GNSS Observations and Machine Learning at Scale: A Case Study from the CAMALIOT Project.
- Automatic classification of land cover from LUCAS in-situ landscape photos using semantic segmentation and a Random Forest model - ScienceDirect.
- Mapping drivers of tropical forest loss with satellite image time series and machine learning
NODES will mobilize the tools of citizen and data science combined with Earth observations to monitor, analyze, and foster progress towards the UN Sustainable Development Goals (SDGs).
Note: This article gives the views of the author, and not the position of the IIASA blog, nor of the International Institute for Applied Systems Analysis.