How Much Do You Know About IBM's Computing and AI for Climate and Sustainability...

IBM Research's Computing and AI for Climate and Sustainability

• IBM Research's Climate and Sustainability team operates in 12 laboratories around the world.

• Climate change is already happening, with most parts of the globe experiencing drastic warming trends.

• Climate change leads to weather extremes that have physical, financial, social, and political impacts, with rising pressures from investors, consumers, and constituents.

• Increased greenhouse gas emissions drive rapid climate change, with remaining carbon budgets and pathways to mitigate and adapt to climate change requiring multifaceted approaches.

• IBM Research is focused on developing core technologies to support a spectrum of climate mitigation and adaptation applications, including nature-based carbon sequestration, renewable energy, carbon performance, and climate risk and impact.

• IBM Research's geospatial discovery network accelerates finding climate solutions by overcoming challenges due to the vast logical and physical distribution of climate information.

• The network of geospatial discovery nodes for computational systems enhances geospatial information representation, in-situ information discovery, and new architectures for distributed workflow optimization.

• Foundation models, as the basis for custom application UX, are pre-trained on 100s of TBs of unlabeled language data and fine-tuned on a class of tasks, allowing scalable access to data, features, and models.

• Geospatial data relevant for sustainability include satellite and aerial imagery, weather measurements and forecasts, and multispectral images from different processes.

• IBM Research's nature-based solutions include carbon sequestration, such as mangroves that are among the most carbon-rich forests in the tropics, and lidar processing, such as Global Ecosystem Dynamics Investigation (GEDI) that estimates canopy height and above-ground sequestered carbon.

• IBM Research uses financial transactions and natural language processing (NLP) to estimate scope 3 emissions by leveraging spend data as a proxy for the quantity of goods/services purchased and embodied carbon, and domain adaptation of large language models to map transaction descriptions into International Standard Industry Classification (ISIC) industries.

• IBM Research's World Wide IoT Weather Monitoring Network creates a blended forecast model via machine-learned, situation-dependent multi-model blending, applying functional analysis of variance to understand first, second, third, …order errors and model accuracy that can depend strongly on “weather situation” category.IBM's Climate and Sustainability Research Innovations

• IBM's Climate and Sustainability team operates in 12 laboratories globally.

• Climate change is already happening and driving weather extremes with massive impacts.

• Responding to climate change requires multifaceted approaches.

• IBM's core technologies support a spectrum of climate mitigation and adaptation applications.

• IBM's Geospatial Discovery Network accelerates finding climate solutions.

• IBM's Geospatial Foundation Models provide a basis for custom applications.

• IBM uses satellite image analysis, LIDAR processing, and GEDI to measure above ground sequestered carbon.

• IBM uses large language models and financial transactions for scope 3 emission estimation.

• IBM's IoT weather monitoring network creates a blended forecast model for weather forecasting.

• The blended model uses machine-learnt, situation-dependent multi-model blending, resulting in over 30% error reduction for 30-day ahead forecasting.

• IBM uses a multi-faceted approach for methane detection, including satellites, drones, and IoT sensors.

• IBM's data fusion technology indexes methane information at scale with PAIRS and uses kernel and physics-based super-resolution techniques for inverse modeling and leak localization.