Computer Science – Performance
Scientific paper
Dec 2008
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2008jgre..11312013l&link_type=abstract
Journal of Geophysical Research, Volume 113, Issue E12, CiteID E12013
Computer Science
Performance
2
Planetary Sciences: Solid Surface Planets: Remote Sensing, Planetary Sciences: Solid Surface Planets: Surface Materials And Properties, Planetary Sciences: Solid Surface Planets: Composition (1060, 3672), Planetary Sciences: Solid Surface Planets: Instruments And Techniques
Scientific paper
Partial least squares (PLS) regressions were applied to lunar highland and mare soil data characterized by the Lunar Soil Characterization Consortium (LSCC) for spectral estimation of the abundance of lunar soil chemical constituents FeO and Al2O3. The LSCC data set was split into a number of subsets including the total highland, Apollo 16, Apollo 14, and total mare soils, and then PLS was applied to each to investigate the effect of nonlinearity on the performance of the PLS method. The weight-loading vectors resulting from PLS were analyzed to identify mineral species responsible for spectral estimation of the soil chemicals. The results from PLS modeling indicate that the PLS performance depends on the correlation of constituents of interest to their major mineral carriers, and the Apollo 16 soils are responsible for the large errors of FeO and Al2O3 estimates when the soils were modeled along with other types of soils. These large errors are primarily attributed to the degraded correlation FeO to pyroxene for the relatively mature Apollo 16 soils as a result of space weathering and secondary to the interference of olivine. PLS consistently yields very accurate fits to the two soil chemicals when applied to mare soils. Although Al2O3 has no spectrally diagnostic characteristics, this chemical can be predicted for all subset data by PLS modeling at high accuracies because of its correlation to FeO. This correlation is reflected in the symmetry of the PLS weight-loading vectors for FeO and Al2O3, which prove to be very useful for qualitative interpretation of the PLS results. However, this qualitative interpretation of PLS modeling cannot be achieved using principal component regression loading vectors.
No associations
LandOfFree
Partial least squares methods for spectrally estimating lunar soil FeO abundance: A stratified approach to revealing nonlinear effect and qualitative interpretation does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Partial least squares methods for spectrally estimating lunar soil FeO abundance: A stratified approach to revealing nonlinear effect and qualitative interpretation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Partial least squares methods for spectrally estimating lunar soil FeO abundance: A stratified approach to revealing nonlinear effect and qualitative interpretation will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-969279