Canopy Height Estimation in Mediterranean Forests of Spain with TanDEM-X Data

Abstract

Canopy height is an essential feature in forest inventory, and for assessment of biomass and carbon budgets. Spatially explicit maps of forest height over large areas can be derived from satellite Synthetic Aperture Radar (SAR) data. We aimed to evaluate the capacity of TanDEM-X (TDX) data to assess canopy height in Mediterranean forests of Spain, which are of relatively short height (typically < 20 m), diverse in species and structure, and adapted to summer drought. Interferogram coherence was retrieved from single-pol image pairs. Forest height estimation was carried out by previously fitting a sinc-type function, with two empirical parameters, to the data measured. Six types of forest were defined to assess the convenience of stratification for model implementation. The influence of terrain slope, forest type, and interferometric baseline on model performance was evaluated, and a strategy for large area mapping was proposed and tested. TDX-derived heights were compared to a contemporaneous LiDAR-derived canopy height model for assessment of quality. Results limited to slopes below 10 degrees provided the best results, reaching R2 = 0.91 and RMSE = 1.24 m in one of the study sites. However, in some areas the results were much worse, especially in regions characterized by rugged terrain with broadleaved species. This work demonstrates the feasibility of deriving a forest height map over the entire area of Spain from TDX data. Stratification per slope interval, as well as selection of long interferometric baselines are recommended.