[ncl-talk] Unrealistic (strange) wind values using obj_anal_ic interpolation function

[Gerardo Montoya]

 Thanks Dennis for your comments: I checked that, the variable lv_HTGL2 in
the GFS analysis, effectively has two values corresponding to 80m and 100
respectively. Here was my error. I also found that this level variable, in
the GFS forecast it has 3 values corresponding to the levels 2, 80 and 100m
respectively, as it can be seen for the following ncl_filedump: ncl_filedump
gfs.t18z.pgrb2.0p25.f001.grb2 -v lv_HTGL2 | less Variable No. 0 lv_HTGL2: 2.000
80.00 100.0 However, even I don't know the name of the level variable
(instead of lv_HTGL2) I should use in order to download the 10m wind.
Thanks for your help.

Gerardo

El mar., 5 mar. 2019 a las 22:22, Dennis Shea (<shea at ucar.edu>) escribió:

> Hello,
>
> A few questions/comments/suggestions:
>
> [1]
> I am not sure where you get 10m wind 1st guess.  Adding "*.grb2*" on the
> command line and looking at the *lv_HGT2 *variable:
>
> *%>*  *ncl_filedump*   gfs.t18z.pgrb2.0p25.anl*.grb2*    *-v lv_HTGL2*
> | less
>
> *GRD_P0_L103_GLL0 *(* lv_HTGL2*, lat_0, lon_0 )       ; <= 3D
>
> *lv_HTGL2*:    80.00       100.0      <== meters
>
> [2]  Apparently, you have created a 10 m wind: EG
>           g = addfile (DirAnData+"*wind10m.nc <http://wind10m.nc>"*,"r")
>       I have no idea how this was created.
>
> [3]  Inverting and the latitude and longitude coordinates does *NOT*
> result in the actual grid being reordered
>
> ;invert the latitude order
> lati=lat_g(::-1)
> ;convert to west latitude
> loni=lon_g-360.0
>
> [4] Please *carefully *examine the *coordinates* after each operation
>
>   dirg = "./"
>   filg = "gfs.t18z.pgrb2.0p25.anl"
>   pthg = dirg+filg+".grb2"            ; add .grb extension
>   fg   = *addfile*(pthg, "r")
>  ;print(fg)                           ; ncl_filedump
> gfs.t18z.pgrb2.0p25.anl.grb2| less
>
>   uh   = fg->UGRD_P0_L103_GLL0        ; ( lv_HTGL2, lat_0, lon_0)
>   *printVarSummary*(uh)
>   print("---")
>
>   uh   = uh(:,*::-1*,:)                 ; Invert the grid (S->N)  via NCL
> syntax
>   printVarSummary(uh)
>   print("---")
>
>   uh   =* lonFlip*(uh)                  ; -180 to 180
>   printVarSummary(uh)
>   print("---")
>
>   lat  = uh*&*lat_0                     ; extract the latitudes *associated
> with the variable*
>   lon  = uh*&*lon_0                    ;             longitudes
>   print(lat)                              ; look at values
>   print(lon)
>
> If, after you makes changes, you still have problens
> I will take a look. However, I would need the files:
>
> VarMetar1H.txt
> VieMetar1H.txt
>
> On Tue, Mar 5, 2019 at 7:55 AM Gerardo Montoya <gemonga at gmail.com> wrote:
>
>>  Dear NCL talk colleagues
>>
>> I’m interpolating wind, extracted from METAR, using  obj_anal_ic function.
>>
>> I take as first guess the wind at 10m (UGRD_P0_L103_GLL0 and VGRD_P0_L103_GLL0 variables) from GFS analysis at 2019-02-28:18Z (gfs.t18z.pgrb2.0p25.anl <https://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/gfs.2019022818/gfs.t18z.pgrb2.0p25.anl>)and use a rscan=(/0.1,0.01/).
>>
>> The METAR information is also taken for this data and time.
>>
>>
>>
>> Fig. Vgfs_A20190228_1800png  shows the GFS analysis wind and Fig. CressmanInterpolation19-02-28_18Z.png, the interpolated wind.
>>
>> These Figs, also contain wind barbs (plotted in black) at airports.
>>
>>
>>
>> Comparing these two Figs., it can be seen that, in some regions and in
>> the vicinity of airports, The interpolation function works as expected: the
>> interpolated wind deviates from the large scale flow (trade winds) and
>> tends to curve in the direction indicated by wind barbs (see the areas
>> highlighted with green arrow). Although, in some areas, specially over
>> complex terrain (central Colombia)  and Cuba, the interpolated wind is
>> extremely strong (see the areas highlighted with red arrow). It there is
>> the possibility I made a mistake in my code. However, I can’t find it. I
>> also noted that Cressman interpolation is very sensible to the rscan
>> values. I’m attaching the code. The NCL 6.5 version is used.
>>
>> Any suggestion will be greatly appreciated.
>> Gerardo Montoya, full professor (retired) at Universidad Nacional de
>> Colombia
>> _______________________________________________
>> ncl-talk mailing list
>> ncl-talk at ucar.edu
>> List instructions, subscriber options, unsubscribe:
>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20190308/9b0258d6/attachment.html>