jagoosw / getgfs Goto Github PK

It seems that when I run the code several times, there are both error and data in the output which means the function is not strong enough to get the data every time. Is is the problem of the website?

Thanks for your time.

While trying to query future forecasts, I encounter an error. I do not understand where else I could manipulate the variable or date time to get a value for a future date other than in the get request. I tried to go to the discussions as referenced in the documentation but encountered a 404 not found. I cannot find anything in the documentation that explains how to achieve this. Attached below is the error encountered. Any help would be greatly appreciated! Thank you.

Hi,

I have encountered a bug in the library which occurs when fetching a range of coordinates instead of just a point.

By running:

>>> f = getgfs.Forecast("1p00")
>>> res = f.get(["gustsfc"], "20220509 5:30", "[-90:90]", "'[0:360]'")

We get

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Input In [49], in <module>
----> 1 res = f.get(["gustsfc"], "20220509 5:30", "[-90:90]", "'[0:360]'")

File ~/dev/env/lib/python3.9/site-packages/getgfs/getgfs.py:103, in Forecast.get(self, variables, date_time, lat, lon)
    100 forecast_date, forecast_time, query_time = self.datetime_to_forecast(date_time)
    102 # Get latitude
--> 103 lat = self.value_input_to_index("lat", lat)
    105 # Get longitude
    106 lon = self.value_input_to_index("lon", lon)

File ~/dev/env/lib/python3.9/site-packages/getgfs/getgfs.py:259, in Forecast.value_input_to_index(self, coord, inpt)
    257 if isinstance(inpt, str):
    258     if inpt[0] == "[" and inpt[-1] == "]" and ":" in inpt:
--> 259         val_1 = float(re.findall(r"\[(.*?):", inpt))
    260         val_2 = float(re.findall(r"\:(.*?)]", inpt))
    261         val_min = self.value_to_index(coord, min(val_1, val_2))

TypeError: float() argument must be a string or a number, not 'list'

The reason for the bug is that re.findall returns a list, not a string, and therefore the casting to float fails.
By changing lines 259 and 260 of getgfs/getgfs.py to:

val_1 = float(re.findall(r"\[(.*?):", inpt)[0])
val_2 = float(re.findall(r"\:(.*?)]", inpt)[0])

the issue is solved.

I'm trying to retrieve the next day forecast. There is any way of getting all values for once or only requesting hour per hour?

Hi, I just installed this library and tried the example but I am getting below error:

import getgfs
f=getgfs.Forecast("0p25")

Error:

SSLError: HTTPSConnectionPool(host='nomads.ncep.noaa.gov', port=443): Max retries exceeded with url: /dods/gfs_0p25/gfs20220824/gfs_0p25_00z.das (Caused by SSLError(SSLError(1, '[SSL: UNSAFE_LEGACY_RENEGOTIATION_DISABLED] unsafe legacy renegotiation disabled (_ssl.c:997)')))

There is no information regarding an extra setup, do I need to register somewhere? Get an API key? or something similar to that or the library should work out of the box with the public URL?

This is just a note on documentation you might want to add somewhere as it tripped me up. I'm used to working with longitude on a -180:180 basis. If you put in a negative longitude, the program doesn't throw any issues, it just sets it to zero. I didn't realize this until I had a different, forward looking issue, and I noticed the longitude was off.

It could be helpful to note this somewhere in the docs, and maybe plug the simple solution that the conversion can be found using the modulo, as seen at the bottom of the screenshot.

-100 Longitude == 100W == -100 % 360 == 260; Doesn't break for positive values. (10 % 360 = 10E still)

As far as I understand, this repository uses the ASCII service of the OpenDAP server to download the data. To get a NetCDF instead of a CSV file (this seems to be on the roadmap), an optional dependency NetCDF4 could be added. Then the NetCDF file can also be downloaded directly from the OpenDAP server using, for instance, xarray (see an example here). However I found this to be quite unreliable, throwing this error quite frequently (even when everything seems OK):

RuntimeError: NetCDF: Malformed or inaccessible DAP2 DATADDS or DAP4 DAP response

For that reason, I would consider using the service GFS filter. This service is able to filter and download the GFS GRIB files directly. I wrote an example notebook if you want to take a look. The only problem is that you need to add the dependency cfgrib, which in turn depends on ecCodes, that cannot be installed with pip. Reading the description, avoid using ecCodes seems to be one of the main motivations of this library. However, I think you can get around ecCodes not working on Windows by installing it with conda.

In any case, I would consider adding either one of both of the previous ways with the optional dependencies NetCDF4 or cfgrib, since I think it is much easier than constructing the NetCDF file "by hand" from the ASCII response.

In the readthecods todos (https://getgfs.readthedocs.io/en/latest/#todo), you put Add historical forecasts from https://www.ncei.noaa.gov/thredds/dodsC/model-gfs-004-files-old/202003/20200328/gfs_4_20200328_1800_384.grb2.das

Are you still on this? Because I'm very interested by this functionnality.

Thx,
N.M.