Methane leaks from oil and gas operations happen irregularly and are hard to measure accurately. Different tools, like satellites or on-the-ground sensors, see different parts of the picture because they vary in how often they measure, how sensitive they are, and how much area they cover. Current methods often are not able to combine all these sources of data, especially when signals overlap or are related.

We developed a new statistical method to bring these different types of measurements together in a reliable way. Our model simulates how methane leaks turn on and off over time and how large the leaks are. It also accounts for the fact that sensors might miss small leaks or lump together multiple sources. By testing this method in simulated scenarios, we show under which conditions it can estimate emissions with less uncertainty or more accurately than existing methods.

This new approach can help support better regulations, and guide how different sensors are used together.