Topic: Problems with optimization of SPOT imagery.
The situation is as follows:
The target area is in the surrounding of a dam,where one of his slopes is sliding by some meters down. This is a fact measurable over ground and quite conspicuous. To check this using COSIS-CORR The following data is available:
SPOT V original scenes covering 2007 up to 2011 years.
An elevation model 5m pixel size, projected into UTM zone 30 with ETRS89 datum, this is an WGS datum with small corrections.In the area this accounts for some centimeters deviations in X and Y and it was ignored and treated as WGS84 to be consistente with all the images and data. From this elevation model I made a shaded dem of the same pixel size, using the same sun altitude and azimuth corresponding to the existents for the raw SPOT data, as reported by the creation of the .ANC file
The same SPOT data orthorectified with the elevation model and with the same datum by a another source with 2.5 m of pixel size.
A mosaic of aerial foto 0.25cm pixel size, orthorectified with the same mosaic by the same source.
I checked the goodness of the orthorectification of both aerial and "orthorectified SPOT" by other program by viewing SPOT and Aerial over the shaded elevation model and both fitted perfectly over conspicuos buildings as over the dam wall.
To get a First image, I cropped the first SPOT image over a small area surrounding the dam , and the the aeril imagery and the same SPOt data over an area bigger than the this small area. 4290 Km2 for the elevation model and aerial and/or SPOT, and 219.49 km2 for the first raw image to be used. Mapinfo in the ENVI files was set to 1.00 for both aerial and orthorectified Spot and to 0.0 for the raw original SPOT.
Tie points were taken in different places, scattered over the original, but not abundant over the dam slopes nor the highest points of the surrounding hills.Number was beetween 50-40 for the aerial and SPOT and some 20 for the shade of the dem. RMS was around 14-15m and no point lies out of the dem area. There are no problems to convert these tie-point to GCPs and the elevation is OK.
However there is no way to achieve any correlation when the optimization of the satellite image is intended, nor by frequency nor by statistic mode, nor using aerial, nor orto -spot nor the shaded dem.The windows Master and Slaves only shows the Master, no traces of any identified point at the slave image.
No problems of different time nor illumination may arise beetween the raw SPOT and the orto-referenced one, they are of course of the same date! That may be the case with the aerial, but some tie point should be identified and I got rid of points surrounded by changing crops. Not even this for the shaded elevation...
What can be done and what I am doing wrong?