Biomass Estimation from Forest Vertical Structure: Potentials and Challenges for Multi-Baseline Pol-InSAR Techniques

Kurzfassung

A central parameter of the terrestrial carbon budget is forest biomass which represents a proxy for the stored carbon. Despite its crucial role in the terrestrial carbon budget, forest biomass is poorly quantified across most parts of the planet due to the great difficulties in measuring biomass on the ground and consistently aggregating measurements across scales. Accurate and reliable estimation of forest biomass is today one of the “hot topics” within the radar (and lidar) remote sensing community. In this work, we discuss the potential and limitations for estimating forest biomass through vertical forest structure parameters obtained from interferometric syntheric aperture radar (SAR) remote sensing techniques. A first approach for biomass estimation was proposed in 2004, basing on the allometric relationship between forest biomass and forest (top canopy) height. This method provides reliable results for forest systems without strong density variations. However, changes in forest density and structure (e.g. due to tree species diversity, management system or disturbances) introduce deviations from the “standard” forest height to biomass relationship and limit the performance of biomass estimation from height data. An important extension has been recently proposed, according to which the height to biomass allometry has been extended by means of the vertical biomass distribution function. This compensates widely the structural heterogeneity within and across forest systems and opens the door for accurate and reliable biomass estimation by means of remote sensing techniques. The estimation of forest vertical structure by means of SAR is a challenge, especially when addressed in terms of a space-borne mission implementation. Conventional SAR tomography has demonstrated the potential to “image” vertical structure by means of multiple acquisitions. However, the lack of multiple space-borne SAR configurations able to perform these acquisitions, combined with (temporal) scene decorrelation that limits the temporal baseline of the acquisitions, reduce the number of suitable (i.e. coherent) acquisitions in a realistic space-borne scenario drastically. Given the availability of only a limited number of acquisitions, alternative approaches have to be used in order to assess vertical structure information by means of SAR. Interferometric approaches are based on the fact that the (volume) interferometric coherence is directly related to the vertical distribution of scatterers seen by the radar and thus to the vertical structure of forests. This information can be either extracted by model based inversion or by approximating the structure function through a weighted sum of a series of (orthogonal) basis functions. The individual parameterization has then to be inverted using a (limited) number of interferometric measurements at the same or different polarizations. In this work, the biomass estimation performance from forest height and vertical structure measurements from interferometric SAR at L-band is addressed. In particular, the pros and cons of the different vertical structure estimation approaches are discussed with respect to biomass estimation. Moreover, the performance of individual techniques is demonstrated using multi-baseline SAR data acquired by DLR’s E-SAR airborne sensor in the frame of different relevant and actual experiments.