Date of Award

5-2017

Document Type

Thesis

Degree Name

Master of Science (MS)

Legacy Department

Forest Resources

Committee Member

Dr. Skip J. Van Bloem, Committee Chair

Committee Member

Dr. Saara J. DeWalt

Committee Member

Dr. Christina E. Wells

Abstract

The use of hyperspectral imagers provides the capability to obtain quantitative and qualitative information on vegetation by remotely collecting reflectance over broad ranges of the ultraviolet and infrared spectrum. Reflectances are determined by the chemical composition and three-dimensional structure of leaves. However, the success of this approach depends on our ability to understand how factors affecting trees influences our ability to interpret reflectance data. The primary objectives of this study were to: 1) Understand the relationship between specific leaf constituents and spectral reflectance patterns of trees in a tropical dry forest, and 2) understand the effects that seasonality has on leaf reflectance and other leaf characteristics of tropical tree species in the reserve. To answer these questions, leaves and hyperspectral data were collected from 83 individuals of 30 species in the Guánica Dry Forest in two missions representing two different seasonal periods (dry and wet). Leaf constituents (Chl-A, Chl-B, carotenoids, leaf water content, and nitrogen) were measured. Reflectance data were obtained from a hyperspectral sensor. The exotic legume species Leucaena leucocephala and Prosopis juliflora and the grasses Uniola virgata and Urochloa maxima tended to be the most distinct from the other species selected for this study. We found seasonal differences for Chl-A, total chlorophyll, and nitrogen concentrations, but not for reflectance at various wavelengths. Among wavelengths recommended in the literature for having strong correlations between water concentration and reflectance, only the wavelengths 1180 and 1190 nm differed among species in this study. Moreover, among the recommended wavelengths for nitrogen, only the 1191 and 1225 nm varied among species. Our results showed no correlation between leaf pigments and spectral reflectance data suggesting that these remotely sensed data will be insufficient for classifying trees species based on their pigment concentrations. Based on our results, the interval (1180-1225 nm) from the electromagnetic spectrum was the most sensitive for the use of reflectance to differentiate among tropical tree species from the Guánica Dry Forest.

Share

COinS