-
Notifications
You must be signed in to change notification settings - Fork 0
Description of inputs
Once OpenGHG inversions is installed we need to prepare the required inputs to the model.
The first thing we need to get ready is the measurements of the gas of interest. It is these measurements that will inform the inversion framework.
The measurements should be prepared using OpenGHG. This will ensure that they are in the correct format to be read by OpenGHG inversion. Once the measurement data has been sourced, it can be prepared in the correct fashion using the tutorial here.
If unfamiliar with the terms footprints, boundary conditions and a priori flux, it may be best to first read the conceptual introduction to OpenGHG inversions.
These inputs will be standardised using OpenGHG for use in OpenGHG inversions. The standardised input data is added to an object store. General information on standardising spatial data using OpenGHG can be found here.
Footprints must be generated using some form of atmospheric transport model (usually a Lagrangian Particle Dispersion Model, LPDM). OpenGHG and OpenGHG inversions do not facilitate the generation of the footprints themselves but allows the standardisation of footprints into a common format. If you are do not currently have footprints available, some commonly used LPDMs that have been used are NAME, FLEXPART and STILT.
Information about standardising footprints for OpenGHG and OpenGHG inversion is available here.
Your own output file will have to contain certain variables which will follow a certain format for OpenGHG to successfully place them in an object store.
Boundary conditions contain a priori estimates of the atmospheric mole fraction at the edges of the spatial domain. These are currently estimated as part of the inversion framework.
The boundary conditions are curtains, spanning the vertical and horizontal boundaries of the domain. They can be taken from other assimilation models (e.g., the CAMS product from ECMWF) or some other prediction model or method.
The boundary conditions must also be standardised using OpenGHG for use in OpenGHG inversions. The documentation on OpenGHG is currently not complete on structuring boundary condition files to be standardised and placed in an object store. The ACRG at the University of Bristol currently have in house code that can create some of these files, and may be useful to use and adapt for your own purposes.
The a priori flux, or a priori emissions, are generally our best estimate of emissions in space before the inversion has been carried out. The initial flux file must follow the formatting recognised by OpenGHG, which will be standardised and added to an object store.
The original data can come from a variety of places, e.g. the EDGAR inventory, which contains spatially resolved emissions for many UNFCCC reported gases.
For a given domain, a country mask is needed (in netcdf format). This country mask defines the countries (or regions) that emissions will be quantified for and will be included in the output from OpenGHG inversions. If a country mask is not available it may be easiest to 1) contact someone in ACRG at the University of Bristol to see if it already exists and can be shared and, failing that, 2) use preexisting code from the ACRG to generate the country mask.
The country mask can be used between all inversions that you do.
A boundary conditions basis function is needed to tell the inversion how to treat the atmospheric mole fraction at the boundaries of the domain. For example, it may be split into only the four cardinal directions, or also into vertical chunks, or as some form of gradient.
Again, it may be easiest to use the preexisting code from the ACRG to create these files. It is likely that only one file will be needed for all inversions that you do.