Detection Limits of the GHGSat Constellation for Carbon Dioxide and Methane

GHGSat launched its first satellite dedicated to carbon dioxide monitoring in October 2023, joining GHGSat’s ten satellite methane-sensing  constellation. All the satellites are designed to measure and attribute emissions at the facility level, leveraging their ~30 meter-scale spatial resolution. Understanding the detection threshold of both CO2 and CH4 satellites is crucial, not only as a fundamental performance metric, but also for interpreting observations where no emissions are detected (null observations). This understanding is especially important when combining observations from multiple sites or times. 

For methane-sensitive satellites, GHGSat has built a large dataset of controlled releases, including both self-organized and third-party single-blind studies. Analysis of this dataset shows a detection threshold of 102 kg/hr, with a 50% probability of detection (PoD) at a wind speed of 3 m/s. One shortcoming is that controlled releases repeatedly measure the same sites at varying source rates, and the number of controlled release sites is limited. These sites might not represent the full spectrum of measurement conditions encountered by the constellation around the world. To address this issue, we adapt the non-linear PoD model developed by Conrad et al.[1] with the goal of providing site- and time-specific detection thresholds.

We will also present an update on the performance of GHGSat’s first CO2-sensitive satellite. We will highlight the difference between CH4 and CO2 point source detection such as co-emission of CO2 with aerosols, multiple release points at a given facility and the high elevation of the release points. We will also present a preliminary analysis of detection threshold by comparing detection events and continuous emission monitoring data available from some power plants. 

[1] Conrad, B. M., Tyner, D. R. & Johnson, M. R. Robust probabilities of detection and quantification uncertainty for aerial methane detection: examples for three airborne technologies. Remote Sens. Environ. 288, 113499 (2023).