| 11 Jan 2010 01:40 PM |
|
|
Hi everyone,
I am trying to run FLAASH on a CASI image (72 bands, 2-byte unsigned int. BIL) but I keep receiving an error message and the routine quits without completing the processing. I am using the following parameters:
Sensor type Unknown-HSI
No water or aerosol retrieval
No spectral polishing
No wavelength recalibration
No tiled processing
Single scale factor for all bands (1.0000)
Set FWHM to band spacing
and each time I attempt to run FLAASH I get the following:
Problem: The FLAASH program exited with the following errors:
ACC error: avrd:
IDL error:
REBIN: Result dimensions must be integer factor of original dimensions
ACC_AVRD
ACC_HYPER2
ACC_ACC
ENVI_ACC_EVENT
ENVI_FLAASH_PROCESS_MANAGER
IDLRTMAIN
$MAIN$
The IDL error code recorded in the journal file is -178, but I haven't been able to find a listing of error codes anywhere. I've tried increasing the cache size and image tile size to no avail. When I turn on water retrieval, the error message does not show up but instead I get an unusable image where almost all the pixels have extremely high reflectance and it is impossible to make out any visual details. Does anyone have any idea what might be going wrong?
Thanks
S.
|
|
|
|
MariM
Veteran Member
Posts:2396  
| 11 Jan 2010 01:47 PM |
|
|
The REBIN error can be caused by using the 0.94um water feature. If your CASI bands only go to about 1.0um then FLAASH may not be able to find the water absorption feature in this region. Try the 0.82um region instead.
The results you get with water retrieval off are likely caused by an incorrect scale factor (you said you used 1). You need to use scale factors that convert the data to floating point radiance in units of uW/(cm^2*sr*nm). If you do not use the correct scale factors, the output data can be off by a factor or more and cause saturation in the high and low ends of the data values.
|
|
|
|
Deleted User
New Member
Posts: 
| 11 Jan 2010 03:06 PM |
|
|
You are correct, the REBIN error occurs when the 940 nm feature is selected. The 820 nm feature is better. I still got an extremely noisy image, but modifying the scale factors helps. Thanks!
|
|
|
|
Deleted User
New Member
Posts:
| 04 Mar 2010 01:18 AM |
|
|
Hi everyone
I'm also trying to run FLAASH on a Hyperion image (196 selected bands BIL format) . It runs okay but the following error message appears in the journal:
Warning: water column calculation used invalid parts of the water column look-up table. Please examine the ACC archive file (.fla) used. You may need to carry out another MODTRAN calculation to extend the humidity range covered. ACC will display a map of the invalid column locations, and store it in "LUTinval.dat" 3876 pixels out of 102400 pixels are invalid.
Enter minimum reflectance for water column display (0.1 recommended): script command 0.000000
Pixels below minimum reflectance are displayed in blue
Minimum H2O = 1.34146 g/cm2 = 1669.01 atm-cm displayed in white
Maximum H2O = 2.13961 g/cm2 = 2662.06 atm-cm displayed in black
Mean H2O = 1.78440 gm/cm2 = 2220.12 atm-cm = 64.4293% of maximum MODTRAN water column.
Standard deviation of H2O distribution= 0.286906 gm/cm2 = 356.963 atm-cm
(For comparison, MODTRAN standard atmosphere values [atm-cm] are:
5119 -- tropical; 3636 -- mid-latitude summer; 1060 -- mid-latitude winter;
2589 -- sub-arctic summer; 518 -- sub-arctic winter; 1762 -- US Standard)
Failed water table look-ups in red.
What does this mean? Tried opening the file with the .fla extension but haven't found any software that will actually open it.
Would really appreciate any help.
Thanks
B
|
|
|
|