Sensitivity of Bistatic TanDEM-X Data to Stand Structural Parameters in Temperate Forests

Abstract

Synthetic aperture radar (SAR) satellite data provide a valuable means for the large-scale and long-term monitoring of structural components of forest stands. The potential of TanDEM-X interferometric SAR (InSAR) for the assessment of forest structural properties has been widely verified. However, present studies are mostly restricted to homogeneous forests and do not account for stratification in assessing model performance. A systematic sensitivity analysis of the TanDEM-X SAR signal to forest structural parameters was carried out with emphasis on different strata of forest stands (location of the study site, forest type, and development stage). Forest structure was parameterized by forest height metrics and stem volume. Results show that X-band volume coherence is highly sensitive to the forest canopy. Volume scattering within the canopy is dependent on the vertical heterogeneity of the forest stand. In general, TanDEM-X coherence is more sensitive to forest vertical structure compared to backscatter. The relations between TanDEM-X volume coherence and forest structural properties were significant at the level of a single test site as well as across sites in temperate forests in Germany. Forest type does not affect the overall relationship between the SAR signal and the forests' vertical structure. The prediction of forest structural parameters based on the outcome of the sensitivity analysis yielded model accuracies between 15% (relative root mean square error) for Lorey's height and 32% for stem volume. The global database of single-polarized bistatic TanDEM-X data provides an important source for mapping structural parameters in temperate forests at large scale, irrespective of forest type. and ranging (LiDAR), and synthetic aperture radar (SAR) data. A comprehensive summary of studies that aim at quantifying forest biomass from satellite sensors was compiled by [3]. AGB estimates from optical remote sensing data mostly rely on vegetation indices that parameterize the photosynthetic activity of vegetation. They imply a relationship between the foliage and the total AGB of a vegetation stand, which is further used to estimate AGB. Since the major part of AGB in forests is composed of non-photosynthetic (woody) components, such approaches imply high uncertainties. The major challenge in AGB mapping from optical data is the saturation of the signal after canopy closure which especially hampers the estimation of high-level AGB [4]. However, recent studies confirmed the general ability of optical systems like Landsat to map AGB at biomass levels around 70 Mg/ha [5].The saturation problem can be overcome by sensors, which are able to penetrate through the canopy and interact with the forest constituents. For instance, LiDAR and interferometric SAR (InSAR) systems generally provide the necessary technique because they are able to penetrate the canopy of a forest and deliver information about the vertical structure of a forest stand. LiDAR data produce high-accuracy canopy height models (CHM) but the main drawback ...

Earth Observation Based Monitoring of Forests in Germany: A Review

“…The second most frequent topic dealt with is "Forest Structure" with 39 publications. Forest structure comprises stand structure [126], canopy gaps [137], stand density [248], and vertical forest structure (e.g., tree height and tree crown diameter) [240]. Some of the publications are also assigned to the topic "Biomass/Productivity" [132], which discusses the relevance of forest structure for biomass and productivity.…”

“…Different regression analysis methods were applied by References [126,142,168,196]. Fischer et al [132] utilized the forest model FORMIND, (an individual-based vegetation model that simulates the growth of forests on the hectare scale [261]) to simulate the field forest structure and, subsequently, simulate the lidar measurements for correlation analysis to find the best remote sensing predictors for the forest structure.…”

Derivation of forest inventory parameters from high-resolution satellite imagery for the Thunkel area, Northern Mongolia. A comparative study on various satellite sensors and data analysis techniques.

“…SAR (Synthetic Aperture Radar) sensors on satellite platforms carry big names like ALOS, TerraSAR-X/TanDEM-X, KOMPSAT, and Sentinel-1. Recent publications show very reasonable accuracies for the estimation of structural parameters (RMSE of 15% for stand height and 32% for stem volumesee also : Erasmi et al, 2019) and the determination of forest above ground biomass / growing stock for carbon sequestration appraisal Zhang, 2016, Wilhelm et al, 2014). Thiel et al (2006) used JERS-1 SAR data to achieve high accuracies (90%) for delineating forest cover and detecting deforestation activities.…”

“…However, such imagery has to meet the requirement of sufficient spatial and radiometric resolution for the recognition of individual trees. Recently, UAV based imagery, as well as LiDAR and SAR have become increasingly popular to furnish three-dimensional information on tree traits and forest structure (Erasmi et al, 2019, Goldbergs et al, 2018, Gougeon and Leckie, 2003, Ke and Quackenbush, 2011b, Kwak et al, 2007, Magnussen et al, 2013, Naesset, 1997, Nevalainen et al, 2017, Park et al, 2014, Popescu et al, 2002, Vauhkonen et al, 2012.…”