Bilinear interpolation for ParaGrid - #1102
Open
einola wants to merge 17 commits into
Open
Conversation
I ended up implementing a new class, called ParaGridInputs. This may change later. An object of this class holds information about the input data, as well as the raw data from before and after the current time step. Calling 'update' loads new raw data, if needed, rotates vectors on the original grid, and then interpolates the resulting fields onto the model grid. Calling 'getField' returns a temporal interpolation between the data from before and after the current time step. As it stands now, no interpolation or vector rotation is implemented, but it seems like the pre-processed data is read in correctly.
This includes implementing weights calculation for the two lat/lon representation posibilities; a Mercator projection (lat and lon are two vectors --- untested!) and a general projection with lat and lon as 2D fields. Lots of refactoring and fixing as well.
I do this, so that we can read in variables from different files as part of the same data set (as is the case with ERA5, for example).
This required some fixing of setWeights1D and refactoring that affects the rest of the class. Nothing is well tested so far, beyond looking at the output fields.
And some cleaning up. I now use FloatType for everything except time.
A small python class to "flood" a dataset's land points with near-by values. This uses a Delaunay triangulation to calculate the values over land, meaning that only the convex hull can be filled. This is probably sensible anyway - we don't want to extrapolate. There's also a function to fill land points with zeros. The class depends on there being a land mask available and unique coordinates for the rows and columns. Re-use of the Delaunay triangulation is ChatGPT code. This is particularly useful for ocean forcing data, but may also be used to flood atmospheric forcing (aka "drowning").
Now that we read in "pure" ERA5 data, we must expect Kelvins, and convert.
Some netCDF files (notably the ERA5 files) save the values as a short, plus a scale_factor and offset. We now support that through NetCDFUtils::readNetCDFVar.
There's something not right with the way they're handled ... but they were zeroed out earlier, so this belongs with a different issue.
I don't know why the static_assert wasn't triggered before, though.
A simple script taking advantage of the flooding.py class to flood the TOPAZ4 files. This is a prerequsite for running the new code, as the mask created by make_init_geographic.py may not overlap with the TOPAZ land mask.
This is to make sure that the new interpolation done in flooding.py fits with the land mask, i.e. that we don't get ocean cells in the land mask where there are NaNs in the forcing. Also, some minor fixes.
Ruff wants me to reformat things.
It works correctly now, but I'm not sure if I need to be this careful. Shouldn't it be enough to just wrap to a fixed bound (like [-180 180]).
We can reasonably assume that it's only ever the latitude axis that is flipped.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bilinear interpolation for ParaGrid
Fixes subs 1 and 2 of #1101
Task List
Change Description
I've implemented a new class for reading in forcing data, called ParaGridInputs. The name will change, and this class will be integrated into the IParaGridIO interface later. The idea here is to submit the initial working example to a different branch, so the final PR won't be too large.
Key changes:
Test Description
No tests so far, beyond checking that the interpolated fields look ok on output. The vectors still point in the wrong directions, so don't expect dynamics to run correctly for very long.
Documentation Impact
Documentation about using different I/O interfaces needs to come later.
Other Details
Switching between different I/O interfaces and vector rotation comes later.