[GRASS-user] TRMM (3B43) handling

[nikos]

hey list,

Sorry that I took so much to respond,

I'm orientating in a different direction right now, and I'd like an
opinion (on TRMM 3B43 ofc).

Here are my thoughts:

a) the hdf that delivers the product only wraps a table 1400 long x 440
tall. Each tile is represented by known cords. eg the upper left tile
( [0][0] ) have the coords of -180deg, 50deg. This is a fact. And also
it is a fact the dArc from each tile to another tile (horizontally and
vertically) its 0.25arcs.
b) So we can controllably somehow "dump" the data into a csv of x,y z
where z are mm/hr (3b43 specification). x,y in a φ,λ coordinate system
(reprojections and projections are not my strong point). 

   results are 1440 x 400 points with known known coords and values.

c)  interpolate for the area you're interested for.

d) You got yourself a raster to have fun with.

Taken into account the reprojection from φ,λ is possible and a viable
solution, the hard part seems to be getting the data. Here's my solution
on that: Giovanni Interface so kindly provided by nasa.gov . You get the
data in ascii format for the area you're researching for and free to
work with htem. 

Here's an example on data dump: 

http://disc2.nascom.nasa.gov/daac-bin/Giovanni/tovas/Giovanni_cgi.pl?west=16.0&north=43.0&east=30.0&south=31.0&params=1|3B43_V6&plot_type=Area+Plot&byr=2006&bmo=05&eyr=2006&emo=05&begin_date=1998/01&end_date=2010/10&cbar=cdyn&cmin=&cmax=&yaxis=ydyn&ymin=&ymax=&yint=&ascres=0.25x0.25&global_cfg=tovas.global.cfg.pl&instance_id=TRMM_V6&prod_id=3B43&action=ASCII+Output


My thoughts are that a python script that utilities the above method is
possible.  

Any thoughts on my comments would be really appreciated.


Best Regards,
Nick Ves



On Mon, 2010-11-15 at 11:03 +0100, Markus Metz wrote:
> On Sun, Nov 14, 2010 at 8:41 PM, Markus Neteler <neteler at osgeo.org> wrote:
> > On Sun, Nov 14, 2010 at 7:42 PM, nikos <vesnikos at gmail.com> wrote:
> >> On Sun, 2010-11-14 at 16:35 +0100, Markus Neteler wrote:
> >>> On Sun, Nov 14, 2010 at 4:01 PM, nikos <vesnikos at gmail.com> wrote:
> >>> >
> >>> >
> >>> > I'm battling to import the 43b3 Product (TRMM perspiration + other
> >>>
> >>> (http://disc.sci.gsfc.nasa.gov/precipitation/documentation/TRMM_README/TRMM_3B42_readme.shtml)
> 
> You could also use the TRMM tools [1] to dump the data to a raw binary
> file. Last time I did that I noted that the binary file is rotated by
> 90°, i.e. north is right and west is up (or north is left and west is
> down, forgot the rotation direction).
> 
> >>
> >> I tried
> >>
> >> ./gdal_translate -a_ullr  -180 50 -50 180
> >> HDF4_SDS:UNKNOWN:"/DIR/OF/TRMM_IMAGE/3B43.060801.6.HDF":1 test.tif
> >>
> >> which makes an interesting tif, that is not empty, but when I load it
> >> with qgis (use psedocolor) its square(?). The image should have a
> >> 1440/400 ratio.
> 
> What does
> gdalinfo HDF4_SDS:UNKNOWN:"/DIR/OF/TRMM_IMAGE/3B43.060801.6.HDF":1
> say? There should be 400 rows and 1440 columns, and not vice versa,
> 1440 rows and 400 columns (suspected 90° rotation mentioned above).
> 
> >
> > Wait - AFAIK you need to use "gdalwarp" on it, not gdal_translate.
> >
> 
> Hm, there is no coordinate system specified in the hdf, so gdalwarp
> would only work with -s_srs wgs84 -t_srs <something else than wgs84>.
> Or translate first, add gcps during translation, then warp?
> 
> Markus M
> 
> [1] ftp://disc2.nascom.nasa.gov/software/trmm_software/Read_HDF/READTRMM_V2.tar