A New AI-Generated Map Can Fight Climate Change

Date: September 01, 2023

Satlas, a new Artificial Intelligence engine, creates high-resolution images of the earth from the low pixel satellite photos to analyze climate change patterns.

Climate change is as real as Artificial Intelligence. We all considered AI another superhuman concept until it started showing its wonders in real life. One such wonder is being tested in the Allen Institute for AI laboratories, where the team has created an engine that converts Sentinel-2 satellite images of Earth into AI-powered high-resolution images for accurate study of climate change patterns.

The accuracy of these AI-generated pictures is still questionable as they are in the toddler-learning phase. However, the advancement of this technology can help global leaders to create stronger policies that fight climate change.

For now, Satlas focuses on renewable energy projects worldwide, except in the extreme parts of Antarctica and far-off ocean areas. It has already been fed with the existing renewable energy systems worldwide, including 36000 wind turbines, 7000 offshore platforms, 4000 solar farms, and 3000 tree cover canopy percentages. This was the basic framework for Satlas’ AI learning. 

Super Resolution

The satellite images of Sentinel-2 are of low quality and fail to clearly identify micro-patches of urban buildings and vehicles. The AI uses its learning from tonnes of data already fed into the system to generate a super-resolution image that clearly makes up a real-life mimicking image of the same. The data is continually fed every month to the satellite to increase the Satlas' accuracy over time. 

The below image is one of the first looks at the Earth through super-resolution generation by the AI from the actual satellite footage.

Like any other generative AI model for image generation, Satlas is also prone to hallucination. “You can either call it hallucination or poor accuracy, but it was drawing buildings in funny ways,” said Ani Kembhavi, senior director of computer vision at the Allen Institute. “Maybe the building is rectangular, and the model might think it is trapezoidal or something.”

The scientists working on Satlas have larger-than-life plans for it. It includes the generation of global crop area, analysis of how much it should increase, and finding the ideal place to increase it while minimizing environmental harm. Optimizing human intervention will be one of the key initiatives Satlas can help climate change bodies strengthen over time. If the advancement of AI leads to a pro-human and pro-earth ideology, the world can actually reverse climate change and create a better future for the next generation, using Artificial Intelligence at its core.