This page documents frequently asked questions about the IMI.
There are two ways to access the IMI
See our Narrative description and how to reference the IMI page for details.
Running the IMI on AWS incurs fees for using AWS compute resources. The typical cost for a 500x500 km2 domain (e.g., the Permian Basin) for 1 month is about $20.
Cost scales with duration and domain size. Costs for larger domains can be effectively mitigated using the smart clustering capability available through the IMI.
The IMI is documented at imi.readthedocs.io.
The IMI Preview is a feature for evaluating an IMI configuration without actually running an inversion. With the IMI Preview you will:
The IMI Preview has no significant costs, and we strongly recommend using it to ensure that the proposed IMI configuration will lead to a successful inversion.
Yes. Users can continuously monitor emissions for a region of interest using the IMI Kalman filter feature.
Yes. Users can override the default IMI prior emission inventories with their own by following these instructions.
No. The IMI gives you total emissions at up to 12x12 km2 grid resolution, including all sources. However, it scrapes point source observations from other databases to improve the prior estimate (optional) and to put the total emission results in context.
Yes. The IMI uses prior sectoral information to allocate the posterior emission estimates to specific sectors.
Yes. Although the satellite observations are mainly over land, the IMI has the option to use glint observations over water. Offshore emissions can also be quantified from observations of the plume advected over land by onshore flow.
The IMI is a completely different product. It provides total gridded continuous emissions, not snapshot emissions from specific point sources as from point source data providers.
The IMI information is most useful for emission reporting, understanding contributions from different sectors, monitoring emission trends, and quantifying long-term averages.
The IMI can now incorporate several of these point source datasets into its clustering algorithm to ensure native resolution at those locations.