| Due to a number of socioeconomic and environmental factors the tropical dry forest has been among the most anthropogenically disturbed, least protected and understudied systems. By taking advantage of the available satellite imagery and investigating the possibility of using these data for inferring characteristics of the tropical dry forest at large scales, the main objective of this thesis is to explore the use of remote sensing to assess tropical dry forest structure and diversity. Initially, the accuracy of four global and regional forest cover assessments are compared at the national level for Costa Rica and then examined in greater detail for two tropical dry forest sites in Costa Rica and Mexico. Significant errors were found systematically throughout each data set. When these errors are examined in terms of carbon sequestration forecasted over a ten-year period, the discrepancies between the maps are valued in the millions of dollars. Second, a comprehensive calibration methodology was established for leaf area index (LAI) by using a combination of litter traps, species specific leaf area values and optical estimates of LAI. In the calibration of the ground-based optical LAI estimates, it was found that the instrument underestimated the actual LAI by 40% or more in the wet season. Next, a new method of estimating LAI from satellite imagery using Bayesian Networks was explored followed by an examination of the effects of season and successional stage on forest structure and spectral vegetation indices for three Mesoamerican dry forests. Differences among the sites are attributed to both climate and varying land use and land management practices. Finally, the structure and floristic diversity of a dry forest in Costa Rica is estimated from hyperspectral satellite imagery (Hyperion). It was found that the dry season image produced the best results using a selection of wavelet decomposition coefficient elements. The final chapter summarizes the challenges for future monitoring of tropical dry forests in the context of ecological succession and remote sensing. |
