Spectral Unmixing Applied to Desert Soils for the - Naval ...
Spectral Unmixing Applied to Desert Soils for the - Naval ... Spectral Unmixing Applied to Desert Soils for the - Naval ...
after atmospherically correcting the data using the FLAASH method described previously. As Figure 21 illustrates, once the endmembers have been identified, the next step is to perform MTMF to determine where endmembers occur within an image and what their abundances are within a given pixel. Once this has been conducted, maps of this information can then be generated (Boardman and Kruse, 2011). Figure 21. This figure shows the processing methods for spectral mixing analysis using the N-dimensional approach adapted from Kruse et al., (2003) and Boardman and Kruse (2011). a. The Hourglass Approach The hourglass method is suitable for the purposes of this study because it allows for accurate endmember extraction without any prior knowledge of location details (Boardman and Kruse, 2011). In general, the hourglass method converts radiance data to apparent reflectance (required for spectral analysis) using an atmospheric correction model (in this case FLAASH). The minimum noise fraction (MNF) (transformation resulting in new components being ordered by image quality (Boardman 42
and Kruse, 2011; Green et al., 1988)) portion performs noise suppression within the data as well as some reduction in both spatial and spectral space. Pixel Purity Index (PPI) determination is where the purest pixels (endmembers) are identified using convex geometry. Endmembers are then visualized through utilization of the n-dimensional visualizer in the ENVI software package. The endmembers were identified through the use of both the ASD spectral library described previously and libraries built by USGS (Clark et al., 2007). The endmembers were then run through the MTMF spectral (partial) unmixing process to determine location and abundance in the HSI data. b. The MTMF Method MTMF involves three analysis steps. Step one involves an MNF transform, step two calculates the matched filter, and step three is where mixture tuning (MT) occurs. MT utilizes convex geometry to measure pixel mixture probabilities that are composites of both the target spectrum and background spectra. MTMF is an automated process that only requires the data and predetermined endmembers (Boardman and Kruse, 2011). In order for step one to occur, pre-processing of the data must be completed. Pre-processing has two main objectives performed as part of the execution of the MNF transform (Boardman and Kruse, 2011). During the pre-processing phase, noise whitening is done via decorrelation of noise present in the data and unit variance across all dimensions of the spectra. The noise whitening and data characterization step is a crucial part of the MTMF process because it is a main element of what allows estimation and target detection within the imagery (Boardman and Kruse, 2011). Three possible options exist in estimating an MNF transform, estimation using a shift difference, using a dark current image, or through use of known noise parameters. All three methods are statistical approaches that are useful under varying conditions. The end result is to determine the eigenvectors and project the noise whitened data onto them in order to decorrelate the data using equation (7) from Boardman and Kruse (2011). 43
- Page 11 and 12: LIST OF FIGURES Figure 1. The above
- Page 13 and 14: spectrum by atmospheric effects. Re
- Page 15 and 16: emoved function showing an absorpti
- Page 17 and 18: LIST OF TABLES Table 1. This table
- Page 19 and 20: LIST OF ACRONYMS AND ABBREVIATIONS
- Page 21 and 22: I. INTRODUCTION A study published b
- Page 23 and 24: II. THE PHYSICS BEHIND REMOTE SENSI
- Page 25 and 26: sensitive a given sensor is to diff
- Page 27 and 28: Figure 3. From Green et al. (1998),
- Page 29 and 30: analyzing imagery spectra, it is mo
- Page 31 and 32: After data have been converted to r
- Page 33 and 34: Collins et al. (1997) was able to s
- Page 35 and 36: These purposes include, but are not
- Page 37 and 38: III. DESERT ECOSYSTEM CHARACTERISTI
- Page 39 and 40: sagebrush of Utah, Montana, and the
- Page 41 and 42: in desert regions include argids, o
- Page 43 and 44: 2. Biological Soil Crusts (BSCs) Bi
- Page 45 and 46: 2004), especially in cases where ma
- Page 47 and 48: IV. STUDY SITES The focus area of t
- Page 50 and 51: Figure 13. This figure illustrates
- Page 52 and 53: Following the uplift that occurred
- Page 54 and 55: the Mazourka Canyon OHV park betwee
- Page 56 and 57: wavelengths being analyzed to obtai
- Page 58 and 59: 2. Field Spectroscopy An Analytical
- Page 60 and 61: A spectral library was then built a
- Page 64 and 65: where: is the mean corrected and no
- Page 66 and 67: also be seen in Figure 23. The leve
- Page 68 and 69: Figure 24. This figure is a compari
- Page 70 and 71: Figure 25. This figure shows ASD co
- Page 72 and 73: Looking at Figure 25 it is apparent
- Page 74 and 75: A. IMAGERY DERIVED ENDMEMBERS The i
- Page 76 and 77: Figure 28. The above shows some of
- Page 78 and 79: spectrometer, reflectance values we
- Page 80 and 81: such an inference can be made (Ben-
- Page 82 and 83: While this is lower than the hoped
- Page 84 and 85: While the lower value would initial
- Page 86 and 87: Figure 33. This figure shows the ad
- Page 88 and 89: Inset C of Figure 35 is the same da
- Page 90 and 91: However, the presences of BSCs are
- Page 92 and 93: A B C Figure 37. Inset A shows the
- Page 94 and 95: A B C Figure 38. Inset A shows a co
- Page 96 and 97: differences in the studies by other
- Page 98 and 99: small concentrations making them un
- Page 100 and 101: area making it possible to tell wha
- Page 102 and 103: Clark, R. N., Swayze, G. A., Livo,
- Page 104 and 105: Kruse, F. A., Boardman, J. W., and
- Page 106 and 107: Sharp, R. P., and Glazner, A. F., (
- Page 108: INITIAL DISTRIBUTION LIST 1. Defens
after atmospherically correcting <strong>the</strong> data using <strong>the</strong> FLAASH method described<br />
previously. As Figure 21 illustrates, once <strong>the</strong> endmembers have been identified, <strong>the</strong> next<br />
step is <strong>to</strong> per<strong>for</strong>m MTMF <strong>to</strong> determine where endmembers occur within an image and<br />
what <strong>the</strong>ir abundances are within a given pixel. Once this has been conducted, maps of<br />
this in<strong>for</strong>mation can <strong>the</strong>n be generated (Boardman and Kruse, 2011).<br />
Figure 21. This figure shows <strong>the</strong> processing methods <strong>for</strong> spectral mixing analysis using<br />
<strong>the</strong> N-dimensional approach adapted from Kruse et al., (2003) and<br />
Boardman and Kruse (2011).<br />
a. The Hourglass Approach<br />
The hourglass method is suitable <strong>for</strong> <strong>the</strong> purposes of this study because it<br />
allows <strong>for</strong> accurate endmember extraction without any prior knowledge of location<br />
details (Boardman and Kruse, 2011). In general, <strong>the</strong> hourglass method converts radiance<br />
data <strong>to</strong> apparent reflectance (required <strong>for</strong> spectral analysis) using an atmospheric<br />
correction model (in this case FLAASH). The minimum noise fraction (MNF)<br />
(trans<strong>for</strong>mation resulting in new components being ordered by image quality (Boardman<br />
42