NIWO still crashes - Githubissues

NEONScience / NCAR-NEON

Repository for developing code related to NCAR-NEON workshop

GNU Affero General Public License v3.0

6 stars 13 forks source link

NIWO still crashes #38

Open wwieder opened 3 years ago

wwieder commented 3 years ago

@danicalombardozzi noticed that NIWO still crashes with the following error:

NEON_NIWO_test1: Ran through 2018-10; exited with: ERROR: CH4 Conservation Error in CH4Mod during diffusionERROR in ch4Mod.F90 at line 4008

This may still be related to #23, but my best guess is:

Odd temperature profiles (below) or;
Low air pressure measured at NIWO? I know the site is high, but this seems kind of crazy!

Especially when compared to results from another site like HARV.

ddurden commented 3 years ago

@wwieder , I'll look into the temp issues more. The pressure seems to be right. I checked the LTER saddle site, and the values are in the same ballpark.

wwieder commented 3 years ago

@jedwards and @negins found this is still an issue. I think it's because precipitation has missing values from 14 Nov 2018 through the end of the month. All the precip data from Dec 2018 and Jan 2019 are also missing. @ddurden is this something you can correct?

wwieder commented 3 years ago

I guess this should have been easy to diagnose from this plot https://neon-ncar.s3.data.neonscience.org/NEON/atm/cdeps/v1/NIWO/NIWO_missing.pdf

danicalombardozzi commented 3 years ago

Thanks for tracking this down @wwieder! I seem to remember a conversation with @ddurden during which he mentioned it was a challenge to find data to use for gap filling at NIWO (if I remember correctly), particularly since there is not a secondary sensor. I don't remember if we came up with a plan or not. It seems like perhaps we could see if data are available from the nearby Fluxnet tower.

wwieder commented 3 years ago

I've taken data from the Saddle (roughly same elevation as NEON and TVan towers), which is recorded daily, and then distributed it at the same time that precipitation falls at the ameriflux tower at C1 (recorded every 30 minutes). The code to do this is here https://github.com/wwieder/NWT_CLM/blob/master/prepare_forcings_for_clm.R

There are issues with winter precip measurements at the saddle, however, which is less than ideal. We can also take data from D1, but these are higher elevation, and also reported daily...

Both Saddle and D1 data from the NWT LTER will also have a longer data latency than NEON, but are fine through 2020 so we can at least get started.

ddurden commented 3 years ago

@wwieder , I've been work on adapting the code to deal with these gaps at sites without replicate sensors. The results still aren't great though, I'll have a look at your code. Sounds like the winter periods are going to be difficult though.

wwieder commented 3 years ago

yeah, winter is not ideal at niwot for a bunch of reasons!

The code I sent applies a correction factor to snow measured at the saddle. Others have recommended using the D1 precip data at Niwot instead, which is higher elevation, but doesn't have the same bycatch issues as the saddle gauge.

wwieder commented 2 years ago

@ddurden can we use outside data from the LTER site to fill these NEON gaps at NIWO? Issues here will be published data are at a daily resolution and not updated in near-real time, but they would at least let us run.

ddurden commented 2 years ago

@wwieder , we should be able to use the outside data. I've been waiting on the yearly data release before rerunning the code.