Retrieval of forest biomass from spaceborne synthetic aperture radar data: on the coupling of an architectural plants model with an electromagnetic model

Abstract : Mapping forest parameters such as aerial biomass is important for sustainable forest monitoring, and highiight forest ecosystems functioning. To this aim, SAR data offer good potentialities due to their use day and night independently of cloud cover. Cross polarized HV low frequency (P- and L-band) backscatter showed good sensitivity to forest biomass up to 200-300 m3[ha. Most of past studies were realized over “simple” forest medium (homogeneous coniferous plantations over flat terrain) and demonstrated that additionai works are needed to take into account architectural tree properties and topographic effects on forest SAR backscatter. Towards the use of various SAR in various forest conditions, a better understanding of the bio-physical link between SAR backscatter and forest parameters is needed by using a theoretical electromagnetic modeiling approach. However, a reajistic and complete canopy architectural description for ail scatterers (trunk, branches, needles or leaves) as a function of growth stage is lacking. An alternative is to deveiop a new approach using the architectural plant model cailed AMAP developed by CIRAD to derive input parameters to theoretical modeis. The first goal of this thesis cailed “direct problem” is to improve the interpretation of wave niatter interaction, using a coupling approach of the vector radative transfer model developed by MIT!CESBIO with the AMAP model. The second goal cailed “inverse problem” is to obtain biomass map using SAR data, AMAP or driving AMAP by parameters obtained using SAR data or other sources through a GIS. In ail cases, effects of terrain siope must be addressed and removed using a SAR post-processing chain developed at LCT. The forested area under study is mainly composed of homogeneous forest stands of Austrian pine plantations located in the central part of Lozère departement (south of France) which offer a large range of topographic situations. Various spaceborne SAR data (ERS, JERS and SIR-C) have been acquired over the test site. A GIS-based methodology was developed towards the analysis of radar backscatter for different growth stages and topographics situations. On the other hand, the forest canopy description based on the AMAP model was implemented into a new interface called AMAP2SAR. Highest sensitivity of corrected backscatter with forest parameters related to aboveground biomass was achieved at L-HV (55°) for bole volume up to 150 m3/ha. Sensitivity was found similar to the flat terrain case, after applying specific radiometric corrections using a semi empirical angular correction model. Moreover, modeling resuits indicates the need to take into account the vertical variability within the canopy by using the AMAP model wich was lacking before. Else, the AMAP2SAR flexibility providing information on any part of the canopy appear to be a powerful tool to interpret forest SAR backscatter. Thus, the link between L-HV backscatter and primary branches density and diameter, wich are correlated through allometric equations to forest classical parameters such as trunk density and basal area and thus with bole volume (m3/ha) was explicited. Also, the modelling approach helped us to refine and validate the semi empirical angular correction model. Finaily, forest parameters inversion using SAR data, AMAP, or their coupling through a GIS are presented. Results show the potentialities and limits in using such coupled tools and leaded us to discuss the potentials of these biomass retrieval methods.
Complete list of metadatas

Cited literature [294 references]  Display  Hide  Download

https://hal-agrosup-dijon.archives-ouvertes.fr/tel-02152966
Contributor : Administrateur Agrosupdijon <>
Submitted on : Tuesday, June 11, 2019 - 6:23:37 PM
Last modification on : Tuesday, October 15, 2019 - 5:30:02 PM

File

Thèse - Castel Thierry.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02152966, version 1

Citation

Thierry Castel. Retrieval of forest biomass from spaceborne synthetic aperture radar data: on the coupling of an architectural plants model with an electromagnetic model. Silviculture, forestry. Ecole Nationale du Génie Rural des Eaux et des Forêts, 1998. English. ⟨tel-02152966⟩

Share

Metrics

Record views

29

Files downloads

14