From: Bryce L N. <bno...@fs...> - 2007-03-01 20:20:26
|
geo...@li... wrote on 03/01/2007 05:25:09 AM: > This data does not have associated > a CRS because it has not been georectified yet. It has some GCP that one > could use to geolocate and then > georectify this data set. > Should I force this dataset to have some spatial CRS? The answer is no since > we got to > preserve the geolocation information. Un-georectified sensor data has an EngineeringCRS. Actually it has several EngineeringCRSes along the way from "spatial location on the focal plane" to "spatial location on the ground". But usually you have "the raw image" and "the resampled image" with all the intermediate steps hidden. The resampled image is typically georectified and can be assigned a "regular" CRS. How does this impact us? If you want to serve MODIS data, the server could advertise the "nativeCRS" as EngineeringCRS and offer the layer in "regular" CRSes as well. The appropriate place to implement a resampler would be the CV_ReferenceableGrid class from 19123. But Geoserver should not care about this petty detail. It treats the gridded data from MODIS the same as it treats gridded data from an orthophoto, using the evaluate() method. The coverage is responsible to convert between real-world and grid coordinates. The coverage reader is responsible for knowing that MODIS data is being read and for installing the appropriate MODIS sensor model in the CV_ReferenceableGrid object. An engineering CRS _is_ a spatial CRS. It's just not a _geo_spatial CRS. :) Bryce |