Problem with ECMWF hybrid levels

Any issues with the actual running of the WRF.

Problem with ECMWF hybrid levels

Postby a_kowaleski » Tue Jun 07, 2016 8:36 pm

I am attempting to run WRF (version 3.8) using ECMWF model level output (Grib1 files).

I run into a problem when I run calc_ecmwf_p.exe. When I run this, using the ecmwf_coeffs file, I get the following error message:

WARNING: Either TT or SPECHUMD not found. No RH will be computed.

My input Grib1 files contain both temperature and specific humidity on sigma levels; however, something is being lost in ungrib between the Grib1 files and the intermediate files. The intermediate files do not contain specific humidity (or relative humidity) values.

If anyone can help me figure out how to get the intermediate files to contain specific humidity values (which I think will solve the problem in calc_ecmwf_p.exe), I will be most appreciative.

For what it's worth, I do not think using pressure-level data is an option in this case (I only have model-level data available)

Alex
a_kowaleski
 
Posts: 2
Joined: Tue Jun 07, 2016 8:12 pm

Re: Problem with ECMWF hybrid levels

Postby alainaketh » Sun Apr 16, 2017 10:22 pm

So I have four nested domains (54, 18, 6 and 2 km resolution), simulation lasts 36 hours, 12 of which are the model spin-up time. Domains 3 and 4 (6 and 2 km resolution respectively) don't have a cumulus scheme.

I've spent waaaay too much time trying to run my case study with data from ECMWF and here's what I've found:

First of all, data from ECMWF was downloaded from two different sites:
1) http://apps.ecmwf.int/datasets/data/era ... vtype=sfc/
2) https://rda.ucar.edu/datasets/ds627.0/

From the first one a esearcher with token and access to MARS downloaded for me pressure levels and surface levels, and for the second one I downloaded pressure and surface levels, and also model and surface levels.

For the three sets of information, I ran as described in my post: http://forum.wrfforum.com/viewtopic.php?f=22&t=1001&start=40#p26253, where I used Vtable.ECMWF, ran ungrib twice, and didn't use calc_coeffs. Yeah, might sound like I did everything wrong but I actually modified the Vtable to include SPECHUM and the model actually ended all simulations. The problem was that it yielded way lower values than expected.

Then I tried something different:
Using model levels and surface levels obtained from (2), I:
1. Ran ungrib for model levels using Vtable.ERA-i.ml (prefix for intermediate files was MODEL)
2. Ran ungrib for surface levels using Vtable.ECMWF_sigma (prefix for intermediate files was SFC)
3. Created file ecmwf_coeffs created using 61 levels as suggested in http://conorsweeneyucd.blogspot.co.uk/2015/01/how-to-run-wrf-driven-by-ecmwf-data.html
4. Ran calc_ecmwf_p.exe and yes, I got the error "No TT or SPECHUM found. No RH will be computed",
however, intermediate files 'PRES' are created.
6. Ran metgrid with MODEL, SFC and PRES.

The simulation, again, ended without problems but it still yielded really low values. I decided to have a look at the input data and compare them to NCEP data (https://rda.ucar.edu/datasets/ds083.3/#description)for the same domain, same time-step, same period of time to be analysed...

- NCEP dataset for my specific case study contains SPECHUM in two levels and RH in at least seven.
- Regarding information downloaded from (1), ECMWF pressure levels contains SPECHUM in only one level (100) and RH in one level as well (100). 100 represents Upper air according to http://www2.mmm.ucar.edu/wrf/users/docs/user_guide/users_guide_chap3.html#_Creating_and_Editing.
- Regarding information downloaded from (2), ECMWF UV model levels dataset doesn't contain neither of those variables, although SC model levels contains SPECHUM in one level (109, which is..?).

I believe the lack of SPECHUM and RH in more levels is the reason behind the warning. And just to be sure, I also analised the AVN data to run the Katrina tutorial and I found that SPECHUM is contained in two levels, and RH is found in at least 8.

Long story short: I gave up and used NCEP data. Needless to say, POD, FAR and RMSE describe a good model performance.
Hydrology Researcher, University of Bristol
alainaketh
 
Posts: 36
Joined: Fri Jun 26, 2015 12:30 pm

Re: Problem with ECMWF hybrid levels

Postby jbasconcillo » Thu Aug 29, 2019 8:54 pm

Have you solved this error?
jbasconcillo
 
Posts: 20
Joined: Thu May 16, 2019 8:56 am


Return to Runtime Problems

Who is online

Users browsing this forum: No registered users and 7 guests

cron