Large-Scale Forest Height Mapping by Combining TanDEM-X and GEDI Data

Choi, Changhyun; Cazcarra-Bes, Victor; Guliaev, Roman; Pardini, Matteo; Papathanassiou, Konstantinos; Qi, Wangdong; Armston, John; Dubayah, Ralph

doi:10.1109/jstars.2023.3244866

Cited by 11 publications

(15 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The corrected mean pro les are nally used in (1). A more detailed description and discussion can be found in (Choi et al, 2023). Next, we applied Eq.…”

Section: Methodsmentioning

confidence: 99%

Mapping Large-Scale Pantropical Forest Canopy Height by Integrating GEDI Lidar and TanDEM-X InSAR Data

Qi,

Armston,

Choi

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Our ability to infer the impact of land use changes such as deforestation and reforestation on concentrations of atmospheric CO2 rests upon accurate and spatially resolved estimates of forest structure, namely canopy height, aboveground biomass (AGB) and biomass density (AGBD). Since April 2019, NASA’s Global Ecosystem Dynamic Investigation (GEDI) mission has been collecting billions of lidar waveforms over the Earth’s temperate and pantropical forests. However, GEDI is a sampling mission and there are large gaps between tracks, as well as those caused by clouds. As a result, the standard gridded height products created from this mission are at 1 km resolution which provides nearly continuous coverage, but which may be too coarse for some applications. One way to provide wall-to-wall maps at finer spatial resolution is through fusion with other remotely sensed data that are responsive to ecosystem structure. The TanDEM-X twin satellites (abbreviated as TDX for convenience all through this study) have provided an unprecedented dataset of global SAR interferometry at X-band since 2010 and have been shown to be highly sensitive to height and other ecosystem structure, but with limited accuracy as compared to lidar. Building on our previous research for fusion of TDX and GEDI, we present a new method of mapping high spatial resolution forest heights across large areas using data from these two missions. Our method uses GEDI waveforms to provide the vertical profile of scatterers needed to invert a physically-based model to solve for canopy height. We assess the impact of using profiles generalized over large areas and develop a calibration method based on GEDI canopy heights to improve model performance. Our method reduces regional errors in forest height caused by the limited penetration capability of the X-band signal in dense tropical forests and the impact of terrain slope using adaptive wavenumber (kZ)-based calibration models and over 2 years of GEDI height observations. In comparison to applying a general country-scale calibration model, the adaptive method selects more representative calibration coefficients for different forest types and landscapes. We apply the method over the entirety of Gabon, Mexico, French Guiana and most of the Amazon basin to produce continuous forest height products at 25m and 100 m. We find that the regional calibration approach produces the best results with a bias of 0.31 m, RMSE = 8.48 m (30.02%) at 25 m and a bias of 0.46 m, RMSE = 6.91 m (24.08%) at 100 m when cross-validated against airborne lidar data. In comparison to existing height data products that have used Machine Learning based approaches to fuse GEDI with passive optical data, such as Landsat and Sentinel-2, our methods produce maps with greatly reduced bias, lower RMSE, and they do not saturate for tall canopy heights up to 56 m. An important feature of this study is that our canopy height product is complemented with an uncertainty of prediction map which is a measure of the predictor’s uncertainty around the actual value rather than the standard error (a square root of estimated variance which quantifies the predictor’s expectation) used by earlier studies. The approach outlined here shows how the integration of GEDI data with TDX InSAR images enables high-resolution mapping of wall-to-wall forest canopy heights, providing an essential foundation for the global mapping of aboveground biomass.

show abstract

“…The corrected mean pro les are nally used in (1). A more detailed description and discussion can be found in (Choi et al, 2023). Next, we applied Eq.…”

Section: Methodsmentioning

confidence: 99%

Mapping Large-Scale Pantropical Forest Canopy Height by Integrating GEDI Lidar and TanDEM-X InSAR Data

Qi,

Armston,

Choi

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…The SeEm-SINC model incorporates two semi-empirical parameters with the help of a small amount of LiDAR data, which makes the model adaptive to the forest scenario and the baseline configuration [43]. The correlation between InSAR coherence and forest height is preferred in this model than in SINC function model, so it reduces certain systematic bias present in the SINC model in some extent [34].…”

Section: B Multi-sinc Modelmentioning

confidence: 99%

Estimation of Canopy Height From a Multi-SINC Model in Mediterranean Forest With Single-Baseline TanDEM-X InSAR Data

Zhang,

Fu,

Zhu

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

TanDEM-X interferometric synthetic aperture radar (InSAR) data have demonstrated promising advantages and potential in recent years for the inversion of forest height. InSAR coherence becomes the primary input feature when a precise digital terrain model (DTM) is unavailable, but the relationship between InSAR coherence and forest height remains uncertain because of the complexity of forest scenes. In this paper, a method for retrieving canopy height in Mediterranean forests, characterised by short and sparse trees, using a single-pass bistatic TanDEM-X InSAR dataset is proposed. To improve the accuracy of forest height inversion from the uncertain correlation between InSAR coherence and canopy height, we begin by using the established SINC model with two semi-empirical parameters and then expand the single curve into a collection of three curves, forming the Multi-SINC model. To determine the optimal relationship (curve) between TanDEM-X InSAR coherence and canopy height, the problem is shifted from parameter inversion to classification. To solve the problem, we used optical remote sensing data, a small amount of LiDAR data, and TanDEM-X InSAR data in combination with machine learning for classification. As a proof-of-concept, we conducted forest height retrieval at two study sites in Spain with complex terrain and diverse forest types. The results were verified by comparing them with LiDAR product forest height, which demonstrated improved performance (RMSE = 2.49 m and 1.7 m) compared to the SeEm-SINC model (RMSE = 3.28 m and 2.36 m).

show abstract

“…In order to help better contextualize the performance of the proposed DL model, we implemented a high level comparison of the baseline scenario with the RVoG model, which is often used in the literature [23], [24], [73], [74]. Given the relationship expressed in (3), the RVoG model parametrizes the vertical reflectivity function F (z) as a two-layer model consisting of a Dirac-like ground component and a vegetation volume component, which is modelled as a continuously extended volume layer of randomly oriented scatterers [21].…”

Section: Comparison With the Rvog Modelmentioning

confidence: 99%

“…A canopy height estimation accuracy was evaluated over the tropical forest area of the Lopé National Park (Gabon), resulting in a peak performance of 8.62m RMSE (r 2 = 0.40). In [24], the authors presented a similar scheme for mapping forest height from TanDEM-X single-polarization data by combining interferometric coherence maps and GEDI waveform measurements over Tasmania. They validated the technique against airborne LiDAR reference data, achieving a RMSE of 7.3m (r = 0.66).…”

Section: Introductionmentioning

confidence: 99%

A Deep Learning Framework for the Estimation of Forest Height From Bistatic TanDEM-X Data

Carcereri,

Rizzoli,

Ienco

et al. 2023

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Up-to-date canopy height model (CHM) estimates are of key importance for forest resources monitoring and disturbance analysis. In this work we present a study on the potential of Deep Learning (DL) for the regression of forest height from TanDEM-X bistatic interferometric (InSAR) data. We propose a novel fully convolutional neural network (CNN) framework, trained in a supervised manner using reference CHM measurements derived from the LiDAR LVIS airborne sensor from NASA. The reference measurements were acquired during the joint NASA-ESA 2016 AfriSAR campaign over five sites in Gabon, Africa, characterized by the presence of different kinds of vegetation, spanning from tropical primary forests to mangroves. Together with the DL architecture and training strategy, we present a series of experiments to assess the impact of different input features on the network estimation accuracy (in particular of bistatic InSAR-related ones). When tested on all considered sites, the proposed DL model achieves an overall performance of 1.46m mean error, 4.2m mean absolute error and 15.06% mean absolute percentage error. Furthermore, we perform a spatial transfer analysis aimed at deriving preliminary insights on the generalization capability of the network when trained and tested on data sets acquired over different locations, combining different kinds of tropical vegetation. The obtained results are promising and already in line with state-of-the-art methods based on both physical-based modelling and data-driven approaches, with the remarkable advantage of requiring only one single TanDEM-X acquisition at inference time.

show abstract

Large-Scale Forest Height Mapping by Combining TanDEM-X and GEDI Data

Cited by 11 publications

References 45 publications

Mapping Large-Scale Pantropical Forest Canopy Height by Integrating GEDI Lidar and TanDEM-X InSAR Data

Mapping Large-Scale Pantropical Forest Canopy Height by Integrating GEDI Lidar and TanDEM-X InSAR Data

Estimation of Canopy Height From a Multi-SINC Model in Mediterranean Forest With Single-Baseline TanDEM-X InSAR Data

A Deep Learning Framework for the Estimation of Forest Height From Bistatic TanDEM-X Data

Contact Info

Product

Resources

About