Fighting Climate change with AI

Experts speak. A Robotics professor and a researcher in Applied Mathematics from Johns Hopkins University discuss how AI can aid in the fight against climate change

When you read the title, you might wonder: How can something as virtual and abstract as AI assist in combating something as vast and tangible as climate change? Fortunately, the speakers have developed a clever framework that makes this understandable and logical by breaking it down into three clear areas: understanding the problem, predicting the direction, and assisting in the required transformation.

1. Understand the Problem

What do we know for certain? CO2 levels are rising, and the greenhouse effect is real. However, data collection is not very detailed. In the ocean, AI aids in operating autonomous underwater robots equipped with sensors. Remote control is impractical because radio waves do not work underwater, making AI a useful tool for enabling independent navigation.

In the atmosphere, we see the totality of greenhouse gases, but the exact sources of these emissions are not well understood. The emitting parties report this themselves, which does not provide complete transparency. The initiative Climate Trace assists in this area by combining satellite data and AI to measure gases worldwide. Looking at transportation, for example; satellite images and sensor data are used to map the congestion of roads. Combined with vehicle type registration, this method allows for accurate estimates per road segment. Visit https://climatetrace.org/; the data is open, and you can see emissions from many cities, factories, and airports.

2. Predict the Direction

Years of research have led to the development of accurate models of Earth’s climate system: our “digital twin.” These models enable the simulation of “what if” scenarios.

Within the climate system, 15 tipping points have been identified. One of these is the AMOC: a major deep-sea current that, among other things, is responsible for Europe’s relatively mild climate. This current is sensitive to changes such as the melting of polar ice. And how are these models useful? You can let AI play with them. As we know, AI is proficient at learning games (chess, Go, etc.). This capability is now being applied in a ’tipping point game’ between two AI systems. One system generates several scenarios with slightly different parameters (more or less melting ice, more or less effect of wind on sea waves), and the other system applies these to the model and evaluates them. For example, will the AMOC tip in this scenario?

Thus, the first player becomes adept at exploring the parameter space. What is important? And the second player becomes proficient at evaluating the scenarios. This way, we learn what could happen, and most importantly, we learn about the time scale: How quickly can this occur? And can one tipping point immediately cause another to tip?

3. Assist in the Transformation

Dealing with climate change involves two things: adaptation (adjusting to the new reality) and mitigation (countering the changes). AI can help with both.

Adaptation. Because AI can create models, it becomes possible to extend existing measurements. Take air pollution as an example. Climate change can lead to more drought and potentially fires, causing pollution to linger in the air; globally, 6.5 million people die annually from air pollution. There are measurements by NOAA of air quality, but they are too coarse in resolution. You want this information at the neighborhood level, but currently, it’s at a 12x12km scale, because computing at finer resolutions requires too much computing power. The solution is to emulate the models with machine learning, achieving higher resolution with far fewer computing resources.

Mitigation. An example from the energy sector. Sustainable energy from the sun and wind is not always available to the same extent. Switching and storing energy becomes necessary. But how much generated energy can you expect (very local weather forecasts) and how much consumption should you anticipate? In both cases, AI models are helpful in predicting supply and demand, enabling an optimal buffering strategy. Another way AI can assist is with inspections. In recent years, AI has significantly improved in perception: understanding images from cameras. This makes it very efficient to inspect wind turbines and schedule preventive maintenance at the right time.

A compelling story from two experts who, through this simple division and with clear examples, suddenly illuminate how AI can make a very useful contribution in this field.