* Background

According to the 2023 Forest Declaration Assessment [1], global deforestation reached 4.1 million hectares and 3.7 million hectares of primary tropical forests lost in 2022 and 2023 respectively, mostly in the Amazon, the Congo Basin, and Southeast Asia. Efforts to protect forest are insufficient, and if this trend continues, these vital ecosystems could transition from being carbon sinks to becoming carbon sources, exacerbating climate change impacts. Several forest loss detection systems, based on space borne remote sensing data, have been developed in the recent years. While most of these approaches rely on optical imagery, radar data, which are insensitive to light conditions and cloud cover, are increasingly being used. As a notable example, the TropiSCO system, initialized during the Ph.D. work of M. Balère (CNES-CESBIO) [2], uses Sentinel-1 data to detect forest loss over nearly ten countries, in an operational manner (https://www.tropisco.org/).

The upcoming ESA BIOMASS mission [3], initialized by CESBIO and to be launched in April 2025, will provide an unprecedented mean to monitor and characterize forest worldwide, including dense tropical forest, through the use of low frequency waves (P band, λ ≈ 70 cm) and of 3D imaging capabilities. The primary objectives of the mission concern the retrieval of Above Ground forest Biomass Density (AGBD), Forest Height (FH) and Forest Disturbance (FD).

The consortium proposing this subject is particularly involved in the preparation and operation of the BIOMASS mission and will manage the Level 2 FD product, as well as the Level 3 (multi-date consolidated) version of all 3 products, in the frame of post-launch activities. It currently supervises the Ph.D. work of M. Bottani (CNES-ISAE-SUPAERO), who recently proposed a Sentinel-1 based forest loss detection method with significantly improved performance [4].

* Description

This project aims at monitoring forest AGB loss in a quantitative way using BIOMASS SAR data and products, as well as other sources of information, such as SAR sensors operated at different frequency bands (Sentinel 1, NISAR, ROSE-L) or optical sensors (Sentinel 2). This objective goes well beyond the ones of official BIOMASS products, and proposes to accurately estimate a key component of forest carbon stock evolution over time.

The performance of the official ESA BIOMASS FD product will be assessed in key areas of interest, distributed over several continents. A new advanced monitoring approach, able to efficiently detect losses at high-resolution will be developed, based on geometric estimation on the manifold of covariance matrices [5] and on the use of geometric priors for SAR 3D imaging [6]. The corresponding improved FD results will be fused with those derived from operational sensors, such as Sentinel-1 and Sentinel-2 in order to fully exploit the potential of different wavelengths for mapping various kinds of forest loss events and phenomena.

Finally, these results will be combined with BIOMASS Level 3 products in a Bayesian regularization approach to enforce geophysical consistency between the estimated quantities, using covariance constraints over time, space, and forest type. Data and products from external information sources, such as other satellite missions, forest descriptor maps, will be incorporated into a high-level assimilation procedure to provide regularized, stabilized and cross-validated time series of quantitative forest AGB loss/gain estimates that could be used as inputs to forest carbon cycle models.

* Information

Supervision: L. Ferro-Famil (ISAE-SUPAERO & CESBIO), S. Labsir (IPSA & TéSA)

Consortium: Supervision team + T. Koleck (CNES), S. Mermoz (GloBEO)

* References

[1] Climate Focus Eds., Forest Declaration Assessment Partners. Off track and falling behind: Tracking progress on 2030 forest goals., 2023.

[2] M. Ballère, A. Bouvet, S. Mermoz, T. Le Toan, T. Koleck, C. Bedeau, M. André, E. Forestier, P.-L. Frison, and C. Lardeux, “SAR data for tropical forest disturbance alerts in French Guiana: Benefit over optical imagery,” RSE, vol. 252, p. 112-159, 2021.

[3] T. Le Toan et al., “The biomass mission: Mapping global forest biomass to better understand the terrestrial carbon cycle,” RSE, vol. 115, no. 11, pp. 2850–2860, 2011.

[4] M. Bottani, L. Ferro-Famil, S. Mermoz, J. Doblas, A. Bouvet, and T. Koleck, “A statistical method for near real-time deforestation monitoring using time series of sentinel-1 images,” in IGARSS 2024, pp. 11 457–11 460, (Best Student Paper Award).

[5] F. Bouchard, A. Breloy, G. Ginolhac, and F. Pascal, “Riemannian framework for robust covariance matrix estimation in spiked models,” in ICASSP 2020, pp. 5979–5983.

[6] P.-A. Bou, L. Ferro-Famil, F. Brigui, and Y. Huang, “Tropical forest characterisation using parametric SAR tomography at P band and low-dimensional models,” IEEE GRS Letters, accepted for publication.

=================

For more Information about the topics and the co-financial partner (found by the lab !); 

contact Directeur de thèse - Laurent.Ferro-Famil@isae-supaero.fr

Then, prepare a resume, a recent transcript and a reference letter from your M2 supervisor/ engineering school director and you will be ready to apply online  before March 14th, 2025 Midnight Paris time !