CMAQ support

[jonathancallahan]

Several things need to happen to support CMAQ output:

• [x] cmaq_toCommonFormat()
• [x] cmaq_load() (including support for level = #
• [ ] cmaq_modelInfo (or should this be part of bluesky_modelInfo
• [x] bluesky_load() support for level = #
• [ ] See how easily bluesky_~() functions can support cmaq -- refactor to have only bluesky_~() functions except for cmaq_toCommonFormat() and possibly include even that functionality.

[tabrasel]

I have several questions to get started:

• What exactly is the relationship and difference between between BlueSky and CMAQ, and how can I tell when a netCDF is produced by one or the other?
• Where can CMAQ files typically be found in the https://haze.airfire.org/bluesky-daily/output/standard/ database? Will there be a cmaq-daily/output/standard/ database at some point?
• What is the level = # you are referring to? Should that specify which layer of a netcdf to load for a multilayered model run?
• What is the intended purpose of a xxxx_modelInfo() function?
• bluesky_toCommonFormat() seems to work just fine for the cmaq file you pointed out: hysplit_conc.nc. Either no extra work is necessary, or hysplit_conc.nc itself is a BlueSky file?

[jonathancallahan]

Here's an old Slack message from Susan:

I put a sample regrid file out of the CMAQ bluesky run (CMAQ_PM25.2020167.DRId02.regrid.nc), and the v2 version of it (CMAQ_PM25.2020167.DRId02.regrid_v2.nc) at: https://haze.airfire.org/webaccess/susan/regrid_file/

---

1. CMAQ and BlueSky are different types of models for predicting smoke dispersion. I believe CMAQ has full "atmospheric chemistry" physics whereas BlueSky does not. Both models are coupled with Meteorological forecast models such as "PNW-1.33km" or "CANSAC-4km" to push the smoke around. I'm going to have to get more information for you on how to tell the difference.
2. Eventually, they should be available in the archive directories but for now we'll probably only be dealing with files in a local directory.
3. I think the level will be found in the netcdf file at the URL above.
4. bluesky_modelInfo is a dataframe created in AirFireModeling.R that keeps track of model grids. It is currently used in bluesky_findModels() but might be useful elsewhere in the future.
5. I pointed you at the wrong file. Please download and try the file mentioned above.

[tabrasel]

Status:

• cmaq_toCommonFormat(): Already completed. There are only 2 lines different between this and bluesky_toCommonFormat(), so they can be combined very easily.
• cmaq_load(): Functions identically to bluesky_load() except that it can only load local files (so no modelName, modelRun, modelMode, or baseUrl params). Can probably be combined once the CMAQ model runs are organized in a database and are downloadable.
• cmaq_modelInfo: This is its own dataframe separate from the bluesky_modelInfo one. Until more CMAQ files are made available, I only have an entry for the file from https://haze.airfire.org/webaccess/susan/regrid_file/.
• bluesky_load(): Now supports the level parameter. As far as I have seen though, BlueSky models only have a single level.
• Refactor to only have BlueSky: This should be very easy given how similar the BlueSky and CMAQ functions are so far. I'm a little confused though: If CMAQ and BlueSky are different models, does it make sense to name all the functions after just BlueSky? Is that sort of the dominant one, or will we come up with an umbrella term that covers both?

[jonathancallahan]

I asked Susan O'Neill and and Robert Solomon to comment on the differences between "BlueSky" and "CMAQ" but I think they are fundamentally different in nature.

So I think you are right in suggesting that we should have an entirely separate suite of functions even if they share code.

In the past I have tried to write code at a higher level of abstraction too early, only to have to split functionality out again to handle all of the persnickety differences between different monitors, models, etc.

So lets just have a separate suite of functions for CMAQ.