CFSR data

[Martina]

Hi all,
I'm happy that the forum is finally online again, however I'm still struggling with using CFSR data in hysplit.
I have managed in the meanwhile to download grib2 data from http://rda.ucar.edu and to convert it to grib1. Now I think I should be able to use grib2arl to finally get it ready for hysplit. However, I do not really understand the grib-files that I'm getting: there is always a whole set of files. I have looked with wgrib/wgrib2 at the inventory of these files and I can't really make sens out of it, each file seams to contain all variables. Currently when I use grib2arl I'm ending with a segmentation fault, and I think that's because I'm using the wrong options not knowing what my grib files actually are. I was googeling already quite a bit and reading through the CFSR websites, but couldn't really find information that would allowed me to advance.

I would be happy if someone could help me with this last conversion step or indicate where to find more hints about the filestructure of this CFSR data!

Thanks a lot,
Martina

[Martina]

Hi all,

I have advanced a bit now and do understand better the CFSR data (there is - a bit hidden - acutally a list explaining the filename conventions on the CFSR website).

However, I do not yet get grib2arl working:
Can I have all my data in one grb file? Or does it need to be split in the 3d fields, 2d fields and the invariant as it is stated in some examples with ERA data? In the userguide however it says "GRIB files may contain ..." in different files, so I assume I can leave them in one file?

So far when I run grib2arl I get the error "Unsupported meteorological center: 0". wgrib2 tells me "centre=National Center for Atmospheric Research (NCAR), Boulder, CO". This error is probably independent of the question if I have to split the files or not, but I have no idea how to deal with that! Does that mean I can't use that data at all? Where do then the few datasets available for ftp dowload from the hysplit website come from?

And it tells me also "First time period found: 0 0 0 0" what for me sounds wrong, however I do not know how this "time period" is defined.

Any help would be much appreciated,
thanks,
Martina

[Martina]

Hi all (and thanks to Yaqiang),

here a short update:
I had found some errors in what I did with grd2arl, now the data is correctly read (for example the meterological centre is now 60), however - what I found in the meanwhile also in the source code - this center is not supported by grd2arl.
I also have tried GFS2ARL which does not support my grids either (even when trying different ones proposed on the CFSR website).

However, the good news, MeteoInfo seams to do the job (after some modifications in Jython script), at least I get an arl-file which I can reopen with MeteoInfo and where also chk_file from hysplit seems to be happy with. Thanks.

However, I'm a bit unsure now, which variables I actually have to use and to what they correspond in the lines

Code: Select all

gvars = [...],
avars=[...],
arlDI.LevelVarList.add(...).

Did I get it right that gvars and avars have to correspond to what is in my grib file (what I can check with MeteoInfo) and that arlDI.LevelVarList.add() contains then the list of variables which I actually need in hysplit?

For the moment I have

Code: Select all

arlDI.LevelVarList.add(['HGTS','TEMP','UWND','VWND','RH']) 

- but is it RH or RELH, I have found all of them somewhere in the usermanual, but don't get what is what!

And the other two lines read

Code: Select all

gvars = ['Temperature_isobaric','Temperature_height_above_ground',\
   'Relative_humidity_isobaric','Relative_humidity_height_above_ground',\
   'u-component_of_wind_isobaric','v-component_of_wind_isobaric',\
   'Pressure_reduced_to_MSL_msl','Geopotential_height_isobaric']
avars = ['TEMP',T02M',RH',RH2M','UWND','VWND','PRSS','HGTS']

where again I'm unsure about the RH names, but also if I have the correct pressure.

And then there is the levels, I have 2m values and then isobaric values between 1 and 1000mbar in my grd-file. From what I see level=0 corresponds to the 2m (but also some 10m) values in the example script. I have 2m values from relative humidity and temperature, and a surface pressure. Is that correct, or do I also need 10m velocities as done in the example:

Code: Select all

arlDI.LevelVarList.add(['PRSS','T02M','U10M','V10M'])

I'm a bit confused about these variables.

Thanks a lot for you time and help,
Martina

[Martina]

Hei Yaqiang,

thanks again for your quick reply!

Thanks for the hint to the variable list, however RELH is not listed there, only RH2M, that's why I was wondering. But ok, then it's RELH und not RH since RH is too short .

With respect to the needed input variables, I read the user guide (section convert to ARL, ECMWF ERA), the FAQ and some other post in the forum dealing with converting ERA interim data, unfortunately they do not agree totally.
I'm quite sure, that I need geopot. height (HGTS), temperature (TEMP), wind velocities (UWND, VWND, WWND) and rel. humidity (RELH).
Then, there might be T02M (2m temp), U10M/V10M (10m wind velocities), terrain height/surface pressure - however not all of them are listed everywhere.
Since I have them, the save thing is probably simply to use them all, but what would happen for example if I drop the 10m winds (since sometimes they are listed as optional)? Would my calculations be less precise or could there be convergence problems or what would be the consequence?
And with respect to "terrain height or surface pressure" for the surface fields, does it make a difference if I take terrain height (HGTS?) or PRSS (pressure)? And do I need pressure PRSS or pressure at mean-sea level MSLP or is both possible as long as I read the correct value from my CSFR datafiles?

I saw also (now that finally all websites are back online), that there is also a tool ncr2arl which should work with my data, however when I try it, it just says "start processing" and nothing else happens.

Thanks for your patience!

Martina

[Martina]

Hi,

mostly my CFSR data conversion has now been working perfectly fine with MeteoInfo (conversion grib2 to arl).
However, now I have two datasets, where MeteoInfo seems not to recognize the correct time step (or there is a problem in the data, I don't know!). This happens already before the conversion, just when opening the original grib2 file in the MeteoInfoGUI.
Normally the time should be always 00 06 12 18 and that's also what wgrib2 and grib_ls show me. However MeteoInfo finds for some days data at 01 07 13 19 or 00 05 11 17.

When I try to use these files in hysplit I get

Code: Select all

*ERROR* metset: meteorological data time interval varies
  Changed from          360  min to          300  min
  At day/hr           31           5

what is, as far as I understand linked to this irregular timeinterval.

I wonder now, how can I find out if there is a problem with my data or if wgrib2/grib_ls or MeteoInfo give the correct information?
Is there a way to use the data anyway in hysplit or can I calculate trajectories only when all my input data has the same time intervals?

One of the files is under the link below, for this one I find the hours 01 07 13 19 from the 28.3. on.
https://dl.dropboxusercontent.com/u/22600087/March.grb2

Thanks for your help,
Martina

[Martina]

Thanks for the quick reply, for me it's still the same with V0.9.8. For the moment I'm resetting times which are not 0 6 12 18 to what they should be....
Martina