Missing / Failed Daymet Subgrid Sets

[jnicklas1]

I'm using the TRESDDS server and it works great for almost all requests. However, there are a few that persistently fail and I can't understand why. For instance, I'm taking a 3km grid of the whole dataset in 2020 and 2021 (using open in bash, which just opens a webpage with this exact url), and the following times out (502 error):

https://thredds.daac.ornl.gov/thredds/ncss/ornldaac/2129/daymet_v4_daily_na_tmin_2021.nc?var=lat&var=lon&var=tmin&north=83.0163&west=-157.0759&east=-6.1376&south=6.0761&horizStride=3&time_start=2021-10-01T12%3A00%3A00Z&time_end=2021-12-31T12%3A00%3A00Z&timeStride=1&accept=netcdf4

But the 2019 request works somehow. This is also true for a vapor pressure horizontal stride request in 1993. I would prefer all of the data to be generated in the same way, although I know I can use NCO for subsetting the full dataset (which I can obtain directly).

[ORNL - yaxingwei]

Hi @j@jnicklas1,

Thanks a lot for the feedback. We are investigating this problem and will find out whether it's due to data structure issue or the THREDDS server issue. We will keep you updated soon.

Best,
Yaxing

[ORNL - yaxingwei]

Hi @jnicklas1,

We did some investigation and found some minor differences between files before 2020 and files in/after 2020 for the Daymet data. For your subsetting request to get the 2021 tmin data, did you always get 502 error? Did that request ever work for you sometimes? Could you please help us to do a few more tests, like increasing the horizStride or the timeStride and choosing a shorter time period for subsetting? We'd like to further investigate this to assess how much impact these minor differences have on THREDDS services. Thanks!

Best,
Yaxing