Conference Agenda

One of the objectives of Dragon 4 Project 31470_2 is the investigation of upscaling and adaptation models and algorithms for 3D multi-functional and scales forestry inventory by using airborne data and products as basis.

The main achievements acquired during the last years could be summarized as follows:

(1) Improvement of forest carbon flux simulation by incorporating remotely sensed model with process-based model.

The improved simulation of forest carbon fluxes was conducted by incorporating a remote-sensing-based MODIS MOD_17 GPP (MOD_17) model with a process-based model (Biome-BGC) using incorporation and data assimilation. Firstly, the original remote sensing-based MODIS MOD_17 GPP (MOD_17) model was optimized using refined input data and biome-specific parameters. The key ecophysiological parameters of the Biome-BGC model were determined through the Extended Fourier Amplitude Sensitivity Test (EFAST) sensitivity analysis.

Then the optimized MOD_17 model was used to calibrate the Biome-BGC model by adjusting the sensitive ecophysiological parameters. Once the best match was found for the pre-selected forest plots for the 8-day GPP estimates from the optimized MOD_17 and from the Biome-BGC, the values of sensitive ecophysiological parameters were determined. The calibrated Biome-BGC model agreed better with the eddy covariance (EC) and tree ring measurements than the original model did. To provide a best estimate of the true state of the model, the Ensemble Kalman Filter (EnKF) was used to assimilate Global LAnd Surface Satellite (GLASS) LAI products into the calibrated Biome-BGC model. Finally, the calibrated and data-assimilated model was applied to simulate the large scale and long-term forest carbon fluxes.

(2) Estimation of forest structure parameters by using multi-source remotely sensed data

As important forest parameters, the leaf area index (LAI), canopy closure (CC), forest height (h) and forest above-ground biomass (AGB) are indispensable for ecological process models and carbon cycle models. Therefore, the accurate estimations of regional or global scale forest parameters are of great significance for a deep understanding of inherent laws of environmental change. With the diversification of remote sensing technology, the single-source remote sensing data has been unable to meet the application demand of the region and high precision. Recently, a large amount of effort has been devoted to the joint utilization of multi-source remote sensing data for the estimation of regional forest parameters. However, the regional application, topographic influence, and mixed pixel decomposition have become the three major scientific problems in the joint retrieval of the multi-source remote sensing data. In response to these three problems, this study has proposed methods for the prediction of the mountain forest height, the canopy closure, and the effective leaf area index (LAIe). Furthermore, the forest AGB model was constructed based on vegetation indices, topographic indices and these structure parameters with physical significance.

The research includes the following three main aspects:

1) Predicting forest height using the GOST model and multisource remote sensing data for sloping terrains.

2) Predicting canopy closure and effective leaf area index using the Li-Strahler geometric-optical model and multisource remote sensing data.

3) Multi-parameter synergic retrieval of forest AGB.

(3) Monitoring forest change by integrating active and passive remote sensing data

Accurate and rapid acquisition of forest land change information is critical for the study of ecological environment changes and forest management planning. At this aspect, multi-temporal Sentinel-1 Synthetic Aperture Radar (SAR) and Chinese Gaofen-1/2 (GF-1/2) optical images were applied to detect the forest changes. For SAR images, the difference image was extracted by using the improved log-ratio method. The Bayesian theory based minimum threshold error adaptive threshold selection method (Kilter and Illingworth, K&I) was used to segment the threshold and extract the change areas. For GF images, the difference image was extracted by using the multivariate change detection (MAD) algorithm, and the maximum inter-class variance method (OTSU) was used to segment the threshold and extract the change areas. Finally, incorporating the change detection results of above two tests was conducted to determine the local forest land changes. The validation based the field survey showed that the incorporation of active and passive remote sensing techniques can efficiently and timely detect the forest land changes with high spatio-temporal resolution, due to high temporal resolution (12 days) of Sentinel-1 and high spatial resolutions (GF-1:2m, GF-2:1m) of GF-1/2 data.

(4) Modeling forest above-ground biomass dynamics using multi-source data and incorporated models

Forest dominates the terrestrial carbon cycle and forest above-ground biomass (AGB) has been the critical index for carbon sequestration capacity. However, any individual method, such as ground-measurement-based method, remote-sensing-based method, and ecological model-based model, cannot efficiently describe the changing processes and driven mechanisms of forest AGB dynamics. Based on multi-mode remote sensing, time-space dynamic knowledge of forest ecological process, and continuous multi-disciplinary ground observation data, this project is planning to model spatial-temporal continuous, physical quantity-synergy forest AGB dynamics. Firstly, a highly accurate regional forest AGB product obtained by applying multi-mode remote sensing and scaling connection is used as the AGB basis. Then, the uncertainties of simulation of forest growth processes are alleviated by use of model-model and model-data fusion strategies. Finally, modeling of forest AGB dynamics is accomplished by combining forest AGB basis with succeeding dynamic forest growth processes, which taking the effects of tree mortality, forest disturbance into account. The methodology of spatio-temporal synergetic modeling of Forest AGB dynamic information proposed by this project, can explore the eco-physiological mechanisms of spatio-temporal pattern of forest AGB dynamics and the driven forces of natural and anthropogenic disturbances. Moreover, this methodology can extend the spatial and temporal dimensions of forest AGB dynamics and in order to precisely improve forest quality and promote the national ecological civilization.

Keywords: Forest above-ground biomass, carbon cycle, model coupling, data assimilation, spatio-temporal synergy

Synthetic Aperture Radar Tomography (TomoSAR) is a microwave imaging technology to focus the illuminated scatterers in the 3D space, by jointly processing multiple acquisitions from parallel trajectories. TomoSAR has been applied with success to the 3D analysis of forested environments.

In principle, TomSAR can be easily understood by considering that the availability of multiple flight lines allows the formation of a 2D synthetic aperture, which permits to focus the signal not only in the range-azimuth plane, as in conventional 2D SAR imaging, but also in elevation.

Although the concept is straightforward, the application of TomSAR using spaceborne sensors is hindered by the fact that different baselines are usually acquired at time lags on the order of days, limiting the analysis to temporally stable targets (like urban scenarios). A possible way out of this blocking circumstance is the employment of single pass interferometers, as in the case of Tandem-X (currently operating) and possible future systems.

Such systems achieve the 3D imaging capabilities by collecting a number of simultaneous interferometric pairs acquired by two satellites. The observed complex coherence corresponds to a particular vertical wavenumber of the imaged scene, depending on the interferometric baseline, i.e., the across-track distance between the two satellites. By collecting multiple pairs with varying interferometric baseline it is then possible to get multiple vertical wavenumbers, which allows the reconstruction of the vertical distribution of the backscattered power of the imaged scene through spectral estimation techniques.

Such tomographic data acquired using spaceborne sensors are characterized by some specific features that may limit the performance of classical 3D focusing techniques. Such data are generally gathered into stacks of a limited number of images, having a coarse spatial resolution, and specific correlation properties. As a result, the available 2^nd order statistical information is largely incomplete and lacks of the redundancy used by classical spectral analysis techniques to enforce a sufficient output signal quality. The achievable vertical resolution is, in general, extremely coarse, due to a limited spatial resolution of the individual SAR images and to a low-pass effect of the spectral interpolation techniques used to reconstruct the missing information.

This contribution summarizes some solutions to these intrinsic limitations. The coarseness of the naturally available vertical resolution, obtained using classical Fourier focusing, is partially compensated with super-resolution techniques based on the processing of a reconstructed covariance matrix. An improved reconstruction of a positive semi-definite covariance matrix is achieved using an original multi-resolution technique which ensures a good conditioning of the estimated information all-along the evaluation process. The validity and usefulness of this approach in the polarimetric mode is assessed using simulated spaceborne data sets obtained from airborne ESA campaigns

Accurate and large-scale access to forest height information is of great significance for the fine management of forests, carbon cycle modeling and scientific research on climate change. Interferometric synthetic aperture radar (InSAR) data without or with very low temporal de-correlation is sensitive to the vertical structure of vegetation and is one of the most potential remote sensing technologies to map forest height in large area. It has been demonstrated by few studies that TanDEM-X InSAR data can be used to map forest height by applying RVoG model or InSAR water-cloud model if the terrain was not so steep, otherwise, descending and ascending InSAR data should be used together for generating one wall-to wall map of an interested region. However, we still need much more detailed investigations on this topic for area of steep terrain in order to apply them to practical mapping activities. So we established two test sites in the Northeast forest region of China: Chaozha forest farm in Genhe district and Wangyedian forest farm in Chifeng district. Chaozha test site is relatively flat, while Wangyedian test site is of steep terrain. Firstly, the performance of forest height inversion using the difference method (DIFF method, in short, taking the forest height as the difference between the DSM from Tandem-X and the DEM from LiDAR) and the SINC model based method (SINC method, in short, where SINC model is one simplified Random Volume over Ground model) was analyzed. Secondly, the effects of signal-to-noise ratio (SNR) de-correlation, spatial baseline, ground-to-volume scattering ratio and extinction coefficient on the estimation of SINC model were studied; Finally, the influence of terrain on the SINC model was investigated, and a threshold determination method was proposed based on Monte Carlo simulation and RVoG model, so as to provide a basis for masking the region severely affected by terrain and for doing multi-track data fusion further. We established a technical process for estimating forest height from space-borne InSAR data without temporal de-correlation under steep terrain conditions, which will provide very useful technique supporting for forest resources monitoring and forestry management activities.

Forest stand height is one of the most important parameters in forest inventory and has a close relationship with other parameters (such as DBH, biomass). Nowadays airborne laser scanning is considered the most accurate remote sensing method for forest height extraction. However, these airborne surveys are relatively expensive and there is a desire to identify more affordable options for collecting or updating this information. In this paper, ZY-3 satellite stereo images are used to derive a digital surface model, which together with a high-resolution digital terrain model (DEM) from airborne laser scanning (ALS) to estimate forest stand height. Forest stand height derived from LIDAR is used as the reference for validation. The results show that ZY-3 stereo satellite images are suitable to extract forest stand height with reliable accuracy when a high-resolution DEM is available.

I. INTRODUCTION

This paper proposes a decomposition technique that accounts the influence of coherent and incoherent double-bounce scattering mechanisms. In order to assess our physical understanding of the interaction between an emitted radar wave and a forested area, a man-made miniaturized RVoG-like scene is imaged. Consisting of a volume lying above a ground, this scene highlights the presence of ground/volume double-bounce and ground/trunk double-bounce.

II. Validation on in-situ data

An electromagnetic wave encounters four potential scattering mechanisms in a forest. The single-bounce on ground, double-bounce ground/trunk, double-bounce ground/volume and volume scattering.

The equivalent distance of a wave encountering a double-bounce is the distance of a single-bounce originating from the ground. Meaning that the double-bounce mechanism is considered as a ground response due to the fact that classical imagery algorithms will represent it on the ground. Taking all potential double-bounces along the volume, it follows that a projection of volume contributions will be located on the ground beneath the volume.

Bare soil contributions k_s are therefore estimated by subtracting double-bounce contributions k_st+k_sv from total ground response, i.e. k_s = k_g − (k_st+k_sv).

PolTomSAR (Polarimetric SAR Tomography) acquisitions over a man-made miniaturized RVoG-like scene show that ground/trunk double bounce is coherent and that ground/volume double-bounce is incoherent. Existing decomposition techniques such as SKP (Sum of Kronecker Products) or HySKP (Hybrid SKP) introduced respectively in [1] and [2] are therefore able to separate the coherent double-bounce from the ground but the incoherent double-bounce remains.

The proposed approach to separate this incoherent double-bounce is to estimate volume contributions and subtract a portion of these contributions from the ground by using the Freeman decomposition for volume estimation [3].

REFERENCES

[1] S. Tebaldini, "Algebraic Synthesis of Forest Scenarios From Multibaseline PolInSAR Data," in IEEE Transactions on Geoscience and Remote Sensing, vol. 47, no. 12, pp. 4132-4142, Dec. 2009. doi: 10.1109/TGRS.2009.2023785

[2] M. Pardini and K. Papathanassiou, "On the Estimation of Ground and Volume Polarimetric Covariances in Forest Scenarios With SAR Tomography," in IEEE Geoscience and Remote Sensing Letters, vol. 14, no. 10, pp. 1860-1864, Oct. 2017. doi: 10.1109/LGRS.2017.2738672

[3] A. Freeman and S. L. Durden, "A three-component scattering model for polarimetric SAR data," in IEEE Transactions on Geoscience and Remote Sensing, vol. 36, no. 3, pp. 963-973, May 1998.

doi: 10.1109/36.673687