A Bristol-NIWA partnership to unlock MethaneSAT observations

We seek to establish a partnership between the University of Bristol (UoB) and the Aotearoa/New Zealand National Institute of Water and Atmospheric Research (NIWA) to deliver low-latency, high-resolution methane emissions evaluation using MethaneSAT observations. Our proposed project brings together our groups’ complementary expertise to solve a major bottleneck in our ability to provide rapid, policy relevant methane emissions estimates from this unique new dataset. It will bring valuable new capability to the UK science community and new outputs to non-academic beneficiaries who rely on robust, high-resolution methane flux estimates for tracking progress on climate change mitigation activities. MethaneSAT, launched in 2024, is now making observations of methane (CH4) at unprecedented resolution and precision (~1 km2 and 2-3 ppb) in 200 km2 scenes over key emitting regions. The instrument has been designed to enable near-real-time tracking of the world’s major oil, gas and agricultural sources and therefore to provide rapid feedback to industry and policymakers on the efficacy of mitigation actions. However, the very large data volumes now being delivered by MethaneSAT cause major bottlenecks in the atmospheric modelling and inference methods that are needed to interpret the data. Our partnership aims to overcome this limitation by bringing together UoB’s globally unique artificial intelligence (AI)-accelerated emissions evaluation system and NIWA’s MethaneSAT data interpretation capability. Through a Google PhD studentship and UKRI funding, our team at UoB have created a deep learning model of atmospheric transport for greenhouse gas flux inference, the Graph-based Atmospheric Transport Emulation System (GATES, Fillola et al., 2023). This system can calculate satellite methane measurement “footprints”, the key component of emissions inference frameworks, almost three orders of magnitude more quickly than 3D atmospheric models. The system will be further developed and deployed for global lower-resolution global mapping satellites over the next three years through a NERC Pushing the Frontiers grant. In this seed corn project, we aim to leverage this activity to develop, in parallel, the capability for GATES to simulate MethaneSAT observations, opening the possibility of simultaneously evaluating distributed area fluxes and large point sources. The proposed partnership with NIWA is critical for this project, because, with the Environmental Defence Fund and Harvard University, they are key members of the MethaneSAT science team and are leading on the evaluation of agricultural emissions. They have expert knowledge of the use of MethaneSAT observations for flux inference, which is not available in the UK, and they have developed a substantial set of model simulations that can be used to train GATES. The proposed research plan was co-developed during a 3-week sabbatical of the PI at NIWA in January 2025. These preliminary activities, complementary expertise, and existing model training set will allow us to scale up our project immediately. The primary aim of this project is to make accurate observation-based CH4 flux estimates at scales relevant for tracking climate action in the UK and New Zealand. We aim to do this at a fraction of the computational cost of previous methods. This new capability will attract wide interest from greenhouse gas researchers and stakeholders around the world, opening several avenues for broader investigations into global emissions and sustainable longer-term funding.