Estimation of Forest Biomass From Two-Level Model Inversion of Single-Pass InSAR Data
Artikel i vetenskaplig tidskrift, 2015

A model for aboveground biomass estimation from single-pass interferometric synthetic aperture radar (InSAR) data is presented. Forest height and canopy density estimates Delta h and eta(0), respectively, obtained from two-level model (TLM) inversion, are used as biomass predictors. Eighteen bistatic VV-polarized TanDEM-X (TDM) acquisitions are used, made over two Swedish test sites in the summers of 2011, 2012, and 2013 (nominal incidence angle: 41 degrees; height-of-ambiguity: 32-63 m). Remningstorp features a hemiboreal forest in southern Sweden, with flat topography and where 32 circular plots have been sampled between 2010 and 2011 (area: 0.5 ha; biomass: 42-242 t/ha; height: 14-32 m). Krycklan features a boreal forest in northern Sweden, 720-km north-northeast from Remningstorp, with significant topography and where 31 stands have been sampled in 2008 (area: 2.4-26.3 ha; biomass: 23-183 t/ha; height: 7-21 m). A high-resolution digital terrain model has been used as ground reference during InSAR processing. For the aforementioned plots and stands and if the same acquisition is used for model training and validation, the new model explains 65%-89% of the observed variance, with root-mean-square error (RMSE) of 12%-19% (median: 15%). By fixing two of the three model parameters, accurate biomass estimation can also be done when different acquisitions or different test sites are used for model training and validation, with RMSE of 12%-56% (median: 17%). Compared with a simple scaling model computing biomass from the phase center elevation above ground, the proposed model shows significantly better performance in Remningstorp, as it accounts for the large canopy density variations caused by active management. In Krycklan, the two models show similar performance.

interferometric model

canopy density

interferometric synthetic aperture radar (InSAR)

Aboveground biomass (AGB)

two-level model (TLM)


forest height


Maciej Soja

Chalmers, Rymd- och geovetenskap, Radarfjärranalys

H. Persson

Lars Ulander

Chalmers, Rymd- och geovetenskap, Radarfjärranalys

IEEE Transactions on Geoscience and Remote Sensing

0196-2892 (ISSN)

Vol. 53 5083-5099