Leveraging the opportunities offered by AI for global climate change is both feasible and desirable. Still, it involves a sacrifice in view of a very significant gain. AI can help improve and expand the current understanding of climate change. AI is increasingly part of a package of reactions essential to combatting the climate crisis effectively by delivering much greener, more sustainable, and more effective solutions.

Despite scientific consensus about the basic facts of climate change, many aspects of the environmental crisis remain uncertain. This includes explaining past and present events and observations and accurately predicting future outcomes. AI's ability to process enormous amounts of non-structured, multi-dimensional data using sophisticated optimisation techniques is already facilitating the understanding of high-dimensional climate datasets and forecasting future trends. 

AI techniques have been used to forecast global mean temperature changes; predict climatic and oceanic phenomena, cloud systems, and tropical instability waves; better understand aspects of the weather system—like rainfall, generally and in specific locales, such as Malaysia—and their knock-on consequences, like water demand. In many cases, AI techniques can help to improve or expedite existing forecasting and prediction systems, for example, by automatically labelling climate modelling data, improving approximations for simulating the atmosphere and separating signals from noise in climate observations.

NASA's foundation models

As extreme weather events around the world increase due to climate change, the need for further research into our warming planet has risen. In NASA, climate research means conducting studies of these events and empowering outside researchers to do the same. The AI efforts spearheaded by the NASA offer a powerful tool to accomplish climate goals.

In 2023, NASA partnered with IBM Research to create an AI geospatial model. The model provides a base for various AI-powered studies to tackle environmental challenges. In keeping with open science principles, the model is freely available to access. 

In a NASA report, Manil Maskey, the data science lead at NASA's Office of the Chief Science Data Officer (OCSDO), stated that the foundation models only know what things are represented in the data. In his opinion, the models are like Swiss Army knives—they can be used for multiple things. After its creation, the model can be trained on a small amount of data to perform a specific task. NASA's Interagency Implementation and Advanced Concept Team (IMPACT) have demonstrated the geospatial foundation model's capabilities by fine-tuning it to detect burn scars, delineate flood water and classify crop and other land use categories.

NASA and IBM

NASA and IBM partnered to create the computational resources required for the initial foundation model. NASA brought the data and scientific knowledge, while IBM brought computing power and AI algorithm optimisation expertise. Following the success of their geospatial foundation model, NASA and IBM research are continuing their partnership to create a new similar model for weather and climate studies.

This time, the dataset will be a massive collection of atmospheric reanalysis data from 1980 to the present day. Covering all aspects of Earth science would take several foundation models trained on different datasets. Moving forward, NASA and IBM's geospatial and climate foundation models are expected to enable leaps in Earth science like never before.

NASA Blog

The AI gambit: leveraging artificial intelligence to combat climate change—opportunities, challenges, and recommendations