How to combine depths from a series of images
- to produce an image of the deepest depths
- to produce an image of the average depth
|"I think we should easily be able to do the data combination to obtain deepest values. |
Clearly the quality of the result will depend on what 4SM is able to produce.
Currently I am unclear how sediment, even if very slight, affects the results.
I was hoping multi-date images might illustrate the capabilities or limitations."
|Example for a 12 TM/ETM images series |
Original dataa time series of 12 co-registered images must be available,
with the same number of lines and columns, superposable
- All 12 4-bands BGRN Landsat TM or ETM original images in format image.pix must be placed in the same directory
- and each image shall have a directory with its name
First, Model all images, each in its directory
- All images shall be calibrated and modeled in their respective directories, using their respective 4SM commandline
- -MakePIX..., -Import..., -Extract..., -Calibrate..., -Model...
then Combine depth results: run script tanzaniaCZ.sh in main directory
- Once all images are modeled to the best of your satisfaction in their respetive directories, prepare a tanzaniaCZ.sh commandline script in the main directory, and execute it: it shall
- prepare a list all images.pix available
- create a tanzaniaCZ.pix image with all specified U8 channels
- import Computed depths from channel_Z250 of each modeled image
|nice -20 ./4SM.2.06 -PROCESS/24_FEB_2009/MASK_0 -Origin/NPA-FUGRO |
- The practitioner may modify
- N, the minimum number of depth results required to compute an average depth
- cSTD, the coefficient that is applied to the standard deviation for restricting depth range using the filter:
- In practice, 4SM first computes an average for all depths available: average+-standard_deviation (N=N), then
- if the condition on N is satisfied, 4SM rejects those depths that fall outside the specified range and computes a new depth and standard deviation which is written out
- if N is found to be less than 2, no average depth is written out
- else the first result is written out
- As of 09 march 2009, this is only very rough, and needs to be improved
- still, the result is quite interesting and promising.