-
Notifications
You must be signed in to change notification settings - Fork 38
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add the MERRA2 model support #35
Comments
Here is some code from pyaps - might help get started. nc4 might be most relevant. |
An example of an actual download is here, although looking to leverage OpenDAP for retrieval of the specific layers and some slice&dice operations, not a complete file download. MERRA-2 datathere are pressure and model outputs. We should aim to use the model level files.
OpenDAP for slice and dice of dataWe made an initial test with OpenDAP but were in contact with GSFC to resolve some issue. The idea is to contruct a URL which has all the subsetting included. Here is relavant documentation:
Below i capture the example we worked on for MERRA-2 with Ray:
The problem experienced when getting the temperature:
Feedback from the GSFC colleagues was the following link: |
Also there is a lot of similarity between MERRA-2 and the GMAO models, both leverage OpenDAP and also have same filename, format and layers. So we should be able to re-use it. That say we are working closely with GMAO team, so we could try to get that to work with openDAP first, as now we have their support in case of questions. |
@dbekaert could you ask the MERRA-2 team for the updated url? Jeremy and I both encountered the 302 found error, saying the "the document has moved", when using the OpenDAP to retrieve the data. |
@leiyangleon might just be a typo in the documentation. See the parent folder of the URL: https://goldsmr5.gesdisc.eosdis.nasa.gov/opendap/MERRA2/M2I6NPANA.5.12.4/1980/01/ Also, the URL changes depending on the year its being accessed; so one would need to construct it based on the user request. See the above link that @piyushrpt provided or look within the code of TRAIN to get an idea on URLS. |
@dbekaert the above URL link is the same as the one @jlmaurer and I tried. It won't work. I also looked at the URLs that you and Piyush mentioned above (in Pyaps) and it is indeed as you mentioned year-month-date-hr dependent. So I need to get the right format of these. Well I am guessing for example, year=1980, month=01, date=19800101 and hr=00. However, it would be really helpful to have a documentation or updated URL database. Could you direct me to either of these resources? Otherwise I have no idea/experience what I am manipulating and expecting. |
Hey @leiyangleon I just went to the URL and copied this link an example file: |
We reversed engineered the link to know what it its. The baseurl can change depending on the year. See the Pyaps or TRAIN code for it. |
@leiyangleon looked a bit in more detail on the internet.
I tried it now:
Can you add documentation to the RAIDER README.md as well? |
@jlmaurer: When a user submits a call and asks for a region of interest, do you automatically expand the model bbox with extra nodes prior to calling the download code or does @leiyangleon need to account for that himself? @leiyangleon the example provided above does not slice and dice yet. Next step would be to slice and dice based on the bbox, the utc time etc using opendap; and then just load the complete array. |
@dbekaert Thanks for spotting the problem. That is probably the reason why I cannot get access to the specific data values. Sounds good. Will update the docs then. |
@dbekaert @jlmaurer problem solved after adding the application @dbekaert thanks for the notes. Yes, we will slice the array based on the bounding box and utc time, etc. |
@dbekaert @jlmaurer I checked the above-mentioned pressure-level (inst6_3d_ana_Np, M2I6NPANA 5.12.4) and model-level (inst6_3d_ana_Nv, M2I6NVANA 5.12.4) data in the MERRA2 docs. Please see attached the two figures extracted from the docs. For the pressure level, surface pressure and Geopotential height are provided, which can be used to calculate the pressure and height; for model level, only pressure thickness is provided for each model layer without height information. I further checked the docs, which says all of the 72 model layers come with fixed 1-D vector of top pressure of each layer, and the pressure can be calculated with this top pressure plus the layer pressure thickness ( Also, even though we can find the pressure for model-level, we still need the 1-D or 3-D height, which is not available for model-level data but does exist in pressure-level. By referring to the docs again, it says all of these model-level data are defined on a terrain-following hybrid sigma-p coordinate or hybrid-sigma, see below So I need your input about how we are gonna proceed, e.g.
|
@leiyangleon can you replace the model to point to inst3_3d_asm_Nv instead of inst6_3d_ana_Nv. I will verify with the atmospheric scientist to see if this is the best one to use. |
@dbekaert Yes, I can use that instead. That one looks almost identical to the GMAO model. |
I am closing this issue because MERRA-2 mode support has be completed as here. |
Closing as addressed by #112 |
No description provided.
The text was updated successfully, but these errors were encountered: