ghost
commented
1 year ago I wrote an issue on the previous gitlab :
We want to generate the profiles of the emissions.
Requirements
Profiles are simply scaling factors to apply to the emission map.
Profiles must have dimensions:
- time (for temporal)
- level (for vertical) Depending on the case, the scaling factors can depend once only on time and once only on level, or we could have a dual coordinate. This means instead of having two vectors and selecting one value in each vector (time and level), we have a 2d array and we select one value at the time/level coordinate.
Profiles can have additional dimensions for which they vary:
- substance
- category (of the inventory)
- grid cell
Data adding
This explains conceptually how the data will/can/must be added.
Vertical
Each source in the data can have a vertical value assigned to it. This can be very easy in the gdfs to set a single value to each source.
For gridded emissions we need to assign a profile to the data.
Temporal
Each source must have a profile. we must assume that the sum of the values of the profile equals to one. considering a yearly profile. However we might have daily, weekly, monthly profiles as well. These ones are not covering the full domain alone. This means that their value don't sum to 1, but vary around 1.
Missing data
In the case of missing data it might be impossible to generate some kind of output. In that case we want to make sure that we have a good way to specify the missing data.
Suggested solution
We need to adopt many parts of the code the this profiles and we don't want to change the existing framework too much.
In the inventory class, we want to add the following
Vertical data for the sources (gdfs)
gdfs should contain a 3d points or geometry such that we could extract the vertical data from them.
profiles attribute
This should be a xarray dataset containing all the profiles. A convention should be decided for how to treat exactly these things, but the use of multiple dimensions (with names following the conventions) should make it determinable what kind of scaling should be used for generating data higher than the 2-D maps.
general operations notes
When doing inventory operations, one should be careful how the profiles are implemented.
One should make sure that the results come as expected using testing on basic test inventories.
merging inventories
When putting two inventories together, it is possible to keep the original profiles of the two inventories, or to combine them. Combining profiles should be done using a weighted method scaling based on the emission values of the inventories. If now we merge two inventories with different dimensions it should be possible to either downscale the dimensions of the profiles or to upscale them.
Test case
Roofs for zurich mapluft
For the heating sector we could use some of the height of the buildings to create the vertical profiles.
TNO profiles
TNO has profiles already shipped with it. We can use these to test reading to emiproc format and then exporting.
lionel42
Adding vertical profiles and temporal profiles for the emissions