EMPLINE

Description

Computes parameters of band-specific radiometric transformations to transform multiband image values to estimates of scene reflectance A reference reflectance spectrum (typically acquired on the ground) is required for each of one or more targets that appear in the image data.

Parameters

Name	Type	Caption	Length	Value range
FILE *	String	Input file name	1 - 192
SPECFILE *	String	Spectrum library file name	1 - 192
RTLEVEL	Integer	Radiometric transformation level	0 - 1
WLENINT	Float	Wavelength interval	0 - 2

* Required parameter

Parameter descriptions

FILE

Specifies the name of the image file that contains data for which the radiometric transformation is to be computed.

SPECFILE

Specifies the spectral library file that contains one or more reflectance spectra that correspond to targets appearing in the image data.

RTLEVEL

Specifies the radiometric sequence of radiometric transformations already represented in the image metadata that are to be applied to the stored pixel values in order to generate the image values that will be involved in the computation of the new radiometric transformation.

WLENINT

Specifies that the selected bands are to be restricted to those whose center wavelength is either inside or outside a closed interval, specified in nanometers. The default is no restriction.

This parameter has no effect if the input file contains no band center wavelength metadata.

Details

EMPLINE computes parameters of band-specific radiometric transformations to transform multiband image values to estimates of scene reflectance A reference reflectance spectrum (typically acquired on the ground) is required for each of one or more targets that appear in the image data.

Each reflectance spectrum in the spectrum library (SPECFILE) must contain an ROI (region of interest) field that indicates the corresponding target region in the image. If the ROI field specifies a bitmap segment (rather than a point-and-neighborhood or a rectangular region), that bitmap segment must be held in the file specified by FILE (Input Data Set).

The reflectance spectra are not required to have the same wavelength sampling as the image data. If they do not, the spectra are convolved with the relative response profiles for the image bands stored in the metadata, and the convolution results are used to compute the new radiometric transformation.

If an RTLEVEL (Radiometrica Transformation) value of i is specified, the computed radiometric transformation is appended to the end of the pre-existing sequence consisting of the first i transformations in the metadata. Pre-existing transformations beyond the first i is discarded. The radiometric quantity of reflectance ("refl" in the metadata) is associated with the new radiometric transformation.

A placeholder radiometric transformation defined by an offset of 0 and gain of 1 is stored for all bands outside the specified wavelength interval (WLENINT). Furthermore, a radiometric transformation will only be computed for bands whose center wavelengths are within the intersection of the input spectra wavelength ranges.

If the reflectance spectra or image data for a particular wavelength are such that transformation parameters cannot be computed for that wavelength, a placeholder radiometric transformation defined by an offset of 0 and a gain of 1 will be stored for that wavelength.

Example

In this, spectra derived from cupref.pix (apparent reflectance data resulting from a model inversion) will be used to transform cuprad.pix (at-sensor radiance data).

First, extract spectra from cupref.pix using the Focus Spectra Plotting panel and store them in a text file named ref.txt. The spectra must represent radiometric transformation level 1, which is the transformation of D.N. values to reflectance.

Each spectrum should be the mean spectrum for a 3x3 neighborhood, to reduce random noise. The spectra should represent a wide range of scene cover types. A recommended set of neighborhood center (X,Y) locations is {(203,535), (465,1307), (79,1058), (339,1116), (167,1803), (147,1652)}.

Compute and save the transformation parameters:

EASI>file	=	"cuprad.pix"
EASI>specfile	=	"ref.txt"
EASI>rtlevel	=		! new r.t. will be added to the end of existing sequence
EASI>wlenint	=		! no wavelength restrictions

EASI>RUN empline

Extract reflectance spectra from cupref.pix (use radiometric transformation level 1) and estimated reflectance spectra from cuprad.pix (use radiometric transformation level 2) at matching locations and compare them in the Focus Spectra Plotting panel. The spectra in each pair should closely match, except at wavelengths for which the empirical line transformation could not be computed. These wavelengths are indicated with "mask" quality values in cupref.pix.

Algorithm

For each input reflectance spectrum, the image spectra at all pixel locations within the corresponding ROI are evaluated at the specified radiometric transformation level, and averaged to create a single image-derived spectrum.

In the case of a a single input reflectance spectrum, the resulting transformation for each band j is represented by an offset of 0 and a gain of Rj/Ij, where Rj is the reflectance value for band j (taken from the reflectance spectrum) and Ij is the image value for band j (taken from the image-derived spectrum).

In the case of two or more input reflectance spectra, the resulting transformation for each band is computed by regressing the band reflectance values for all of the ROIs on the image values for all of the ROIs. A straight-line linear regression is performed. For a particular band, let the slope of the regression line be 's' and the intersection of the line with the image value axis be 'a'. Then, the regression-estimated relation between image value (I) and reflectance value (R) for that band is R = (I-a)*s . The computed radiometric transformation for that band therefore has gain = s and offset -a*s . If the image values are estimates of radiance, 'a' may be interpreted as an estimate of atmospheric radiance.

References

Reviews and a case studies of the empirical line method, including a description of some of the potential pitfalls, are provided in the following papers:

Smith, Geoffrey M., and Edward J. Milton, "The Use of the Empirical Line Method to Calibrate Remotely Sensed Data to Reflectance", International Journal of Remote Sensing, Vol. 20, No. 1, (1999), pp. 2653-2662.M

Berk, A. et al., 1999, "Modtran4 User's Manual", Air Force Research Laboratory, Space Vehicles Directorate, Air Force Materiel Command, Hanscom AFB, MA

Matthew, M.W. et al. "Status of Atmospheric Correction Using a Modtran4-Based Algorithm", SPIE Proceeding, Algorithms for Multispectral, Hyperspectral, and Ultraspectral Imagery VI, Volume 4049, April 2000