Hi,

I think there is a problem with the AATSR tie-point to pixel interpolation in the epr_api. It appears that when reading the pixel longitude and latitude, the epr_api is returning a point on the edge of the pixel below the one expected.

i.e. if you use epr_api to get the lat/lon of pixels A,B,C,D in the raster below, you actually get points a,b,c,d

Looking at the epr_read_band_annotation_data function, the scan_offset_x/y for AATSR are:
scan_offset_y = 0.5F;
scan_offset_x = -19.0F; // geolocation tie-points
scan_offset_x = 6.0F; // view-angle tie-points

To return the pixel centres (A,B,C,D) I think these should be:
scan_offset_y = -0.5F;
scan_offset_x = -19.5F; // geolocation tie-points
scan_offset_x = 5.5F; // view-angle tie-points

Or, to return the 'lower-left' pixel corners these should be:
scan_offset_y = 0.0F;
scan_offset_x = -19.0F; // geolocation tie-points
scan_offset_x = 6.0F; // view-angle tie-points

Hi,
Was this problem also affecting MERIS ?
If so, can you confirm that the result was an offset in lat/lon by about a pixel ?
Thanks,
François

The issue was not affecting MERIS and ASAR, only AATSR was affected.

Thanks Owen!