Negative values are unlikely to be related to the off-nadir view angle.
- Did you perform radiometric calibration with 'flaash settings' on the geoeye image? If so, what are the basic statistics of the file?
- If you do the above, then you used an input scale factor of 1.0 with your data?
- are the negative values primarily in the blue band or also in other bands? If it is primarily in the blue band, try setting the Initial Visibility to a value such as 100KM.

Yes I use first Radiometric Calibration anbd Apply FLAASH settings, always when I select the radiance image in FLAASG I use 1 for all factors.

The basic statistics are all positiuve values (Radiance QuickStats):
Band Min Max Mean StdDev
Band 1 4.206795 30.428656 13.997359 4.296764
Band 2 2.594633 35.173603 14.268072 4.588568
Band 3 1.700370 32.744274 14.314542 4.494840
Band 4 0.594766 19.636871 10.146269 2.993326

Negative values are for all bands not just for blue (FLAASH QuickStats):
Band Min Max Mean StdDev
Band 1 -453 7957 2790.506393 1126.701203
Band 2 -252 9719 3357.410755 1261.762689
Band 3 -212 10310 4169.073778 1396.255785
Band 4 -222 8831 4331.160048 1322.316790

regards
JM Lattus

I would try the increased initial visibility. Are the negative values all over the image or primarily in shadows/water or other dark areas?

I use 30 km. Is desert a salt flat with sobe dark areas also.