[gdal-dev] Gdalwarp producing odd results around 0 degree longitude

[Simon Proud]

Hello,

I am warping ECMWF grib data (full globe, 0.25 degree resolution) to 
match the SEVIRI geostationary satellite projection. I use this command:

gdalwarp -overwrite -ot Float32 -t_srs "+proj=geos +a=6378169 
+b=6356583.8 +lon_0=0.0 +h=35785831" -te -5570248.832537 5570244.832537 
5567248.429179 -5567252.429179 -ts 3712 3712 -wo NUM_THREADS=6 -wo 
SOURCE_EXTRA=1000 -q -r cubicspline --config CENTER_LONG 180 
/data/ECMWF_T2_20181223T00_000.grib /data/cubicspline.tiff

The output projection looks good, and if I try this with the ECMWF 
land-sea mask then I can see that it is correct for most of the view.
However, along a vertical centred on zero degrees longitude I see this 
artefact: https://i.stack.imgur.com/tQn5F.png


Any ideas what is causing that, and how to resolve it? It artefact 
appears for all resampling methods except nearest neighbour.

Thanks,
Simon

-- 
Dr Simon R Proud
NERC Fellow in Aviation Meteorology,
Department of Atmospheric, Oceanic and Planetary Physics,
University of Oxford,
Clarendon Lab,
Parks Road,
OX1 3PU,
Oxford,
UK

Tel (office): +44 (0) 1865 282431
Tel (mobile): +44 (0) 7563 639685
Email: simon.proud at physics.ox.ac.uk
Twitter: @simon_sat