Remote sensing data provide a quantitative, spatially extensive and repeatable method for mapping and measuring Earth's vegetation. In particular, imaging spectrometers, which measure reflectance in many narrow, contiguous wavelengths, are sensitive to subtle differences in plant structure, biochemistry, physiology and phenology. The need for improved global assessments of biodiversity, species and plant functional type (PFT) distributions, and ecosystem function has paved the way for the proposal of a spaceborne imaging spectrometer with monthly, global coverage. In this research, I evaluated such a sensor's capabilities for accurately mapping species and PFTs in contrasting ecosystems, at varying spatial resolutions and through the year. Using airborne imaging spectrometer data acquired over five different ecosystems, I compared strategies for mapping dominant plant species.

Different combinations of training data selection, dimension reduction and classification methods yielded a range of accuracies, but overall, species could be accurately mapped both within single ecosystems and when ecosystem data were pooled. By spatially aggregating these images to simulate coarser resolution data, I assessed the accuracy of both species and PFT classification in these same five ecosystems. My results demonstrated that the best resolution for mapping varied by site and with PFT composition and diversity. Finally, I analyzed monthly field spectroscopy data in a Mediterranean ecosystem to quantify seasonal changes in spectral separability and to relate these with phenological observations. For each of three vegetation communities, separability was related to plant functional trait differences and changed in response to phenological differences.

Differences in separability were also observed between coastal and inland sites within the same community types related to local meteorological and site conditions. Overall, the results of this work support the launch of a spaceborne imaging spectrometer with global, monthly coverage. They demonstrate the utility of the entire spectrum for differentiating among species and PFTs, as well as the added value of monthly data for improving our understanding of the relationship between seasonal spectral change and phenological changes observed on the ground.