|Landsat8 True Color image (4,3,1)||Results from the Abrams Ratio (1,2,3)|
This model applies the Abrams Ratio formula, to highlight mineral information via three derived bands displayed as an RGB composite. The algorithm was originally designed for use with Landsat 5 TM data, but can be applied to other sensors if the correct bands corresponding to the required wavelengths are selected.
When Run, this model performs a check, at the initial stage, whether the input has 7 or more bands. If true, the Abrams Ratio is applied on the input raster. If false, the model stops and informs the user through the Require Condition operator that, 'Input imagery doesn't seem to have 7 bands' (displayed in the Messages panel). If the Input raster file has more than 7 bands, the model will still run assuming that the selected bands correspond to the correct wavelengths per the algorithm used here. Refer to Abrams Ratio link for further information.
As per the formulas described above, the necessary input wavelengths/bands are (in the order presented in the Run dialog):
The band selection defaults are set for a 7-band Landsat 5 (or 4) TM image (i.e. one where band 6 is the low-resolution Thermal band): 7, 5, 4, 3, 2
For a Landsat 8 scene, the inputs would normally need to be 7, 6, 5, 4, 3.
For a Sentinel-2 scene (with all 13 bands), the inputs would normally need to be 13, 12, 8, 4, 3 (note that these are the sequential band numbers in the image file, not the band "names", which include an "8" and an "8A")
If you are lucky enough to have 16-band WorldView-3 "Supercube" data you could try 14, 11, 7, 5, 3
Hyperion imagery might use 206, 150, 48, 31, 21.
NOTE: The Run dialog presents (and expects) each band as a Range List (e.g. 7:7). However you can enter the desired band number by simply typing the number (e.g. 7) - the software will take care of turning it into a Range List for you when you click OK. If desired the model could be altered to prompt for and take an Integer or Scalar and explicitly convert to a Range List as part of the model (using the Range and Range List operators)
Each Ratio sub-model follows this form:
|"Ratio of Band5/Band7 i.e. 1.65µm/2.215µm" Sub-Model|
The formula in this instance (to produce the output band 1) would therefore be
( Band5 - (Min(Band5) - 1) ) / ( Band7 - (Min(Band7) - 1) )
The Input Image should have 7 or more bands, five of which must correspond to the required wavelengths. Really this limitation is arbitrary and that branch of the model could be removed. The user would simply need to ensure that the input image contains the five required wavelengths and the appropriate bands representing these wavelengths are selected.
Interpretation of output file:
Input Image: Name of the input multispectral image which contains 7 or more bands, covering the requisite wavelengths
Output Image: Name of the output 3 band Arams Ratio image, which should be displayed using a 1,2,3 band combination
uSWIR Band Number: The band number (indexed from 1) of the Input Image which contains DN values captured in the upper Shortwave IR wavelengths (centered around 2.215 µm). Band Numbers can be entered as a simple integer number (e.g. 7), or as a range list denoting a single value (e.g. 7:7)
SWIR Band Number: The band number (indexed from 1) of the Input Image which contains DN values captured in the lower Shortwave IR wavelengths (centered around 1.65 µm)
NIR Band Number: The band number (indexed from 1) of the Input Image which contains DN values captured in the Near IR wavelengths (centered around 0.83 µm)
Red Band Number: The band number (indexed from 1) of the Input Image which contains DN values captured in the Red wavelengths (centered around 0.66 µm)
Green Band Number: The band number (indexed from 1) of the Input Image which contains DN values captured in the Green wavelengths (centered around 0.56 µm)