run hysplit *ERROR* metinp: too many missing times - 1

[andries]

Dear All,

Currently I try to use ERA-Interim data to run hysplit. The conversion of ERA-Interim data (latlon format, 720x360) to ARL format goes well, however, when I run Hysplit, I encounter the following error in the simulation log:

Model started ...
HYSPLIT4 - Initialization
USE: svn info | tail -n3
to fill in these three records
with the version information!
Calculation Started ... please be patient
STOP 900

and the following error in the "MESSAGE" file

.............................................................................

WARNING metinp: missing data - HGTS
Time: 9 11 25 12 -1
Level: 1 Variable: 1
WARNING metinp: missing data - TEMP
Time: 9 11 25 12 -1
Level: 1 Variable: 2
WARNING metinp: missing data - UWND
Time: 9 11 25 12 -1
Level: 1 Variable: 3
WARNING metinp: missing data - VWND
Time: 9 11 25 12 -1
Level: 1 Variable: 4
WARNING metinp: missing data - WWND
Time: 9 11 25 12 -1
Level: 1 Variable: 5
WARNING metinp: missing data - RELH
Time: 9 11 25 12 -1
Level: 1 Variable: 6
WARNING metinp: missing data - HGTS
Time: 9 11 25 12 -1
Level: 2 Variable: 1

..............................................................................
and the last two lines of the "MESSAGE" file
..............................................................................

*ERROR* metinp: too many missing times - 1
Current version no missing time permitted

..............................................................................

At first I expected the problem to be related to the time format (the minutes, mm) in the ERA-Interim or ARL files, or the time setting in the hysplit setup. However, after playing with different setups of Hysplit I suspect that the error is related to a memory problem and incomplete ARL files after the conversion from ERA-Interim. Hysplit runs fine for the first 5 days of November (1-5 November), however, when it runs after 5 November the error occurs. For example, when I run 72hrs forward trajectories from 4 November, the first few time steps go well, after which the error occurs. Now I tried the same with ERA-Interim data at a lower resolution (360x180 instead of the 720x360 format). Here the first 19 days work well, but when I run hysplit for a day after 20 November, the same error occurs again. For example, when I run 72hrs forward trajectories from 19 November, the first few time steps go well, after which the error occurs.

Under the following link you can find the ARL data and MESSAGES files for both the ERA-Interim resolutions when I run in hysplit, +72hrs from 4 and 19 November, respectively.
https://www.copy.com/s/sQ9eTYvw7c8Q/Hysplit

Any help about how to solve this problem, or how to deal with it would be most welcome.
Thank you very much.
Andries

[ariel.stein]

Can you please describe how you are creating these files?
Also, please run chk_index and chk_file from the exec directory to check the content of the files you generated.

[andries]

The different ERA-Interim (N128 gaussian grid) variables are combined into the three input files for HYSPLIT via:

cdo merge ifile-var1 ifile-var2(etc,etc) ofile

Then the ERA-Interim files are converted from N128 gaussian grid format to latlon format (let say resolution 360x180) via the following command:

cdo remapbil,r360x180 ifile ofile

Then the three files (output-3d-lonlat-360x180.grib, output-2d-lonlat-360x180, ei_an_geopotsurf_gg128_200911-lonlat-360x180) are converted to ARL format via:

/exec/./grib2arl -k38 -p0 -ioutput-3d-lonlat-360x180.grib -soutput-2d-lonlat-360x180 -cei_an_geopotsurf_gg128_200911-lonlat-360x180

Finally the ofile DATA.ARL is used as the meteorological file when running hysplit. Depending on the resolution of the ERA-Interim data, hysplit can only be run for limited days in the month as described before.


I suppose "chk_index" and "chk_file" should be run for the file "DATA.ARL"? Both checks seem to be fine...

Under the following link https://copy.com/sQ9eTYvw7c8Q I placed the files which I used, including the "MESSAGE" files obtained when running Hysplit for the 720x360 and 360x180 resolution of ERA-Interim data, +72 foreward trajectories from 4 and 19 November 2009, respectively.

[ariel.stein]

It looks like you are missing data for specific dates. Please run chk_index in each of the ARL formatted files (DATA.ARL) you are using and send me the screenshot
Thanks.
Ariel

[andries]

Under the link https://copy.com/sQ9eTYvw7c8Q I placed the files

chk-index-DATA-720x360-g3-k38-p0.txt

and

chk-index-DATA-360x180-g3-k38-p0.txt

which contain the output of ./chk_index over the files DATA.ARL based on the ERA-Interim data with the 720x360 and 360x180 resolution, respectively. After day 5 (720x360 resolution) and day 19(360x180 resolution), the respective files contain only 0 values! I do not understand what goes wrong during the conversion from grib to arl....

[ariel.stein]

Not all the data are 0s. Apparently there are some missing data in the grib files.
Can you please extract and convert every time period in a separate file and then concatenate the files for each day?
That will make the analysis easier.

[andries]

I checked the grib files with the "wgrib" program and I also read in the files with NCL. The files seem fine, there are no missing values and no "0" values. I suppose something must go wrong when the grib files are converted into the ARL format.

I splitted the monthly grib files into daily files and copied the ones for day 5 and day 25 under the link

https://copy.com/sQ9eTYvw7c8Q

[ariel.stein]

Can you send me the ARL formatted files for those 2 days?

[andries]

sorry for the delayed response, I am on holidays these days. I will send you the ARL formatted files as soon as I am back to work, by the end of August.

[andries]

back to work:) I uploaded the ARL formatted files for day 5 and 25 separately under the link https://copy.com/sQ9eTYvw7c8Q under the sub folders

day05

and

"day25"

Hope these are useful. I presume there will be no problems with these files since they contain data for only 1 day; the problem as described in the aforementioned post occurs in this resolution (360x180) when the ARL file contains data of more than +/-19 days. If there is anything else you would need, please, let me know.