[gdal-dev] Re: GDAL and raster calibration

[Andrey Kiselev]

> > Is it possible to read and write calibration coeffisients when raster
> > data processed with GDAL? I can't find note about calibration in `Data
> > Model' section of docs. What a best method for this purpose?
> > Calibration is important thing and can be linear and stored in two
> > values (gain/intercept), poliline with several nodes, polinomial curve
> > and, moreover, it can be a lookup table. It will be very good to have a
> > possibility to use calibration.

> In theory simple gain and intercept values can be returned from a band
> via the GetOffset() and GetScale() methods. These are primarily intended
> to map between raw pixel values and some linearly related output value
> with physical meaning. A common example would be when elevation data is
> rendered to 16 bit after applying some offset or scaling operation.
> 
> More general models of calibration information should be handled as
> metadata. Metadata is a set of string name/value pairs. You would
> basically pick a set of names for your particular calibration information
> and encode the values into string values.

I'm understand that, but it will be better to have special infrastructure
like those for GCP.

> If we were to formalize a model for calibration information we could pick
> specific metadata names to be used for calibration information and define
> their semantics in the data model documentation; however, in my experience
> calibration information for different formats, sensors, and applications
> tends to vary quite a bit in organization so I doubt we are ready to do
> that yet.

Yes, in the NOAA-15 AVHRR data thermal bands calibarted using polynomial
curve but visible band has polyline calibration ;-) Anyway, we should not
forget about this issue.

-- 
Andrey V. Kiselev
Scientific Research Center for Ecological Safety Russian Academy of Sciences
Office phone: +7 812 2307834 ICQ UIN 26871517
Registered Linux user number 169907