Spatial functional data modeling of plant reflectances

Plant reflectance spectra, the profile of light reflected by leaves across different wavelengths, supply spectral signature for a species at spatial location to enable estimation functional and taxonomic diversity plants. We consider leaf spectra as “responses” be explained spatially. These are also functions over wavelength that respond environment. Our motivating data gathered several plant families from Greater Cape Floristic Region (GCFR) in South Africa lead us develop rich novel models can explain genera within families. Wavelength responses an individual viewed function wavelength, leading modeling. Local environmental features become covariates. introduce covariate interaction, since response regressors may vary with variance. Formal modeling enables prediction reflectances unobserved locations known features. incorporate dependence, space–wavelength interaction (in spirit space–time interaction). implement out-of-sample validation model selection, finding above informative analysis. ecological interpretation results under selected model.