The effect of seasonal spectral variation on species classification in the Panamanian tropical forest

2012 
Abstract Variation in the leaf optical properties imposed by variation in genetics and location has been addressed in recent literature, but those stemming from forest seasonality and phenology have been less well explored. Here, we explore the effect of inter-seasonal spectral variation on the potential for automated classification methods to accurately discern species of trees and lianas from high-resolution spectral data collected at the leaf level at two tropical forest sites. Through the application of a set of data reduction techniques and classification methods to leaf-level spectral data collected at both rainforest and seasonally dry sites in Panama, we found that in all cases the structure and organization of spectrally-derived taxonomies varied substantially between seasons. Using principle component analysis and a non-parametric classifier, we found at both sites that species-level classification was possible with a high level of accuracy within a given season. Classification across season was not, however, with accuracy dropping on average by a factor of 10. This study represents one of the first systematic investigations of leaf-level spectro-temporal variability, an appreciation for which is crucial to the advancement of species classification methods, with broad applications within the environmental sciences.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    47
    References
    59
    Citations
    NaN
    KQI
    []