[mapserver-users] GRIB files - gdal_translate

Thu May 21 08:49:47 EDT 2009

Roberto Garcia,MSc wrote:
> Size is *161, 98*
> Coordinate System is:
> .....
> Origin = (-150.000000000000000,-61.246000000000002)
> Pixel Size = (0.938000000000000,-0.001836734693878)
...
> Band 1 Block=*161x1* Type=Float64, ColorInterp=Undefined

> As u can see the size according to the its header is 161x98, but all the 
> bands are related as 161x1 block. Why does it happen? 

Roberto,

This means that the file (and bands) are 161x98, but the bands
are subdivided into blocks which are 161x1 implying that the natural
access chunk is a scanline.  This is normal.

> Another thing I 
> noticed is that the dif between upper and lower latitude in the corner 
> coordinates is very small (-61.246 to -61.426).

This appears to be screwed up.  Perhaps you could provide the file,
and file a ticket about this?

> The issue is that, when mapserver renders this layer it only draws a 
> straight line below South America, exactly where the whole image should 
> start, as my CTL below shows.

What is "CTL"?

PS. this would likely be better addressed on gdal-dev.  Any ticket
should be filed in the GDAL Trac.

   http://trac.osgeo.org/gdal/

Best regards,

> dset ^T126L28.grb
> title pac/sa/atlan sector: troposphere file
> undef 1e+20
> dtype grib
> index ^T126L28.gmp
> xdef 161 linear -150.000015 0.937500
> ydef 98 levels
> -61.246 -60.311
> -59.376 -58.441 -57.506 -56.571 -55.636 -54.701 -53.766 -52.831 -51.896 
> -50.961
> -50.026 -49.091 -48.156 -47.221 -46.286 -45.350 -44.415 -43.480 -42.545 
> -41.610
> -40.675 -39.740 -38.805 -37.870 -36.935 -36.000 -35.065 -34.130 -33.195 
> -32.260
> -31.325 -30.389 -29.454 -28.519 -27.584 -26.649 -25.714 -24.779 -23.844 
> -22.909
> -21.974 -21.039 -20.104 -19.169 -18.234 -17.299 -16.364 -15.429 -14.493 
> -13.558
> -12.623 -11.688 -10.753  -9.818  -8.883  -7.948  -7.013  -6.078  -5.143  
> -4.208
>  -3.273  -2.338  -1.403  -0.468   0.468   1.403   2.338   3.273   
> 4.208   5.143
>   6.078   7.013   7.948   8.883   9.818  10.753  11.688  12.623  13.558  
> 14.493
>  15.429  16.364  17.299  18.234  19.169  20.104  21.039  21.974  22.909  
> 23.844
>  24.779  25.714  26.649  27.584  28.519  29.454
> zdef 7 levels
> 1000 925 850 700 500 300 200   ===> set Z 1 ou 2 ou 3 ... ou 7 (GRADS)
> tdef 1 linear 6Z16may2009  6hr
> vars 15
> psnm      0    2,102,  0,  0 SEA LEVEL PRESSURE [hPa]
> tsfc      0  187,  1,  0,  0 SURFACE TEMPERATURE [K]
> prec      0   61,  1,  0,  0 TOTAL PRECIPITATION [Kg/m2/day]
> cbnv      0   71,  3,  0,  0 CLOUD COVER [0-1]
> role      0  114,  8,  0,  0 OUTGOING LONG WAVE AT TOP [W/m2]
> mask      7  137,100 MASK [-/+]
> uvel      7   33,100 ZONAL WIND (U) [m/s]
> vvel      7   34,100 MERIDIONAL WIND (V) [m/s]
> omeg      7   39,100 OMEGA [Pa/s]
> fcor      7   35,100 STREAM FUNCTION [m2/s]
> potv      7   36,100 VELOCITY POTENTIAL [m2/s]
> zgeo      7    7,100 GEOPOTENTIAL HEIGHT [gpm]
> temp      7   11,100 ABSOLUTE TEMPERATURE [K]
> umrl      7   52,100 RELATIVE HUMIDITY [no Dim]
> umes      7   51,100 SPECIFIC HUMIDITY [kg/kg]
> endvars
> 
> Can someone help me to find out what is happening?
> 
> Tks a lot
> 
> Roberto Garcia
> 
> 
> Frank Warmerdam wrote:
>> Roberto Garcia,MSc wrote:
>>> Hi, tks for the answers.
>>>
>>> According to my gdainfo output file attached, can anyone tell me what 
>>> element could be a CLASSITEM?
>>
>> Roberto,
>>
>> In MapServer you always classify rasters based on the item [pixel] but
>> you do not need to explicitly state a classitem.
>>
>> This is a relatively simple example of raster classification.
>>
>> LAYER
>>   NAME grid1
>>   TYPE raster
>>   STATUS default
>>   DATA data/float.tif
>>   CLASS
>>     NAME "red"
>>     EXPRESSION ([pixel] < -3)
>>     COLOR 255 0 0
>>   END
>>   CLASS
>>     NAME "green"
>>     EXPRESSION ([pixel] >= -3 and [pixel] < 3)
>>     COLOR 0 255 0
>>   END
>>   CLASS
>>     NAME "blue"
>>     EXPRESSION ([pixel] >= 3)
>>     COLOR 0 0 255
>>   END
>> END
>>
>>> I can see my grib using Grads but what tool do I use to translate 
>>> individual bands into GeoTIFF?
>>
>> You can use gdal_translate to extract particular bands from a grib file
>> as a geotiff (or all the bands).  For instance, the following would
>> extract band 3 (total precipitation):
>>
>> gdal_translate -b 3 T126L28.grb total_precip.tif
>>
>>> Converting the data on-the-fly to an image is the right way to work? 
>>
>> Having MapServer access the grib file on the fly rather than depending
>> on pre-translated extracts will reduce data duplication and may help
>> avoid confusion about what is up to date.  However, it is possible that
>> there will be a performance penalty for extracting directly from GRIB.
>> If so, it may not be significant enough to matter to you.
>>
>>> What are the problems with Apache if I do not do so?
>>
>> I don't see how either approach relates to Apache.  It is a
>> data management and possibly performance issue with MapServer.
>>
>> Best regards,
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> mapserver-users mailing list
> mapserver-users at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-users

-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent