PRISMA L2D loading: incorrect geocoding of panchromatic and hyperspectral band

[antoine-lavrard]

I believe there is an issue with the geocoding of the panchromatic/ hyperspectral data, causing a shift of 15m for hyperspectral data and 2.5m for panchromatic data. Due to this incorecct geocoding, hyperspectral and panchromatic raster do not have the same extent.

I think the origin of the problem are here. Indeed, according to this link, pixel coordinate in a raster correspond to the upper left coordinate when using gdal.

deleting the offset in the code should fix the issue (see below)
`

def spatialInfo(self, metadata, res):
extent = QgsRectangle(
float(metadata['Product_ULcorner_easting']),
float(metadata['Product_LRcorner_northing']) -res,
float(metadata['Product_LRcorner_easting']) +res,
float(metadata['Product_ULcorner_northing']),
)
crs = QgsCoordinateReferenceSystem.fromEpsgId(int(metadata['Epsg_Code']))
geoTransform = (extent.xMinimum(), res, -0., extent.yMaximum(), -0., -res)
return crs, extent, geoTransform

`

[janzandr]

Thanks for pointing that out. I can confirm that the SPECTRAL and the PAN products don't match up.
Currently, I'm not fully convinced, that the proposed fix is correct. I will talk to the PRISMA team at ASI for clarification.

[antoine-lavrard]

You're welcome! Is it possible to keep me informed when you will have thoses clarrification?