Conference Agenda

Over the past 100 years, CO2 concentrations have increased from 280ppm to 400ppm with an
average temperature increase of 0.85 ℃ (IPCC 2013),Greenhouse gases from the burning of large amounts of fossil fuels are the main drivers of climate change (IPCC 2014).The optical system of the three detection channels of the STIIS prototype is designed in detail.The assembly and development of the STIIS prototype are completed.The preliminary test results show that the inversion accuracy of CO2 meets the requirement of better than 4ppm.According to the above work of the STIIS prototype, it provides the technical basis for the next generation of high spatiotemporal resolution detection of greenhouse gases.

Current satellite-based methods for measuring aerosols necessitate a homogeneous surface and pre-assumed surface albedo or reflectance, rendering them unsuitable for urban areas characterized by highly inhomogeneous surfaces. However, with the development of high-resolution satellites, building shadows can be clearly identified in satellite images. A new algorithm has been proposed that retrieves Aerosol Optical Depth (AOD) and surface albedo by using the differences between building shadows and adjacent sun-shined bright pixels. The algorithm was validated using GF-2 satellite images with a 4-meter spatial resolution, successfully retrieving AOD and surface albedo for locations near the Beijing Olympic Center. Comparing the AOD derived from the shadow method with those from MODIS aerosol products and ground-based sun photometer measurements, it has been found that the AOD retrieved by the shadow method align well with ground photometer data, exhibiting differences of less than±0.03. The results indicate that the shadow method can accurately retrieve aerosol data over megacities at a finer spatial resolution.

As the spatial resolution of space-borne imaging spectrometers is rapidly improving and moving towards sub-kilometre scale, three-dimensional (3D) cloud effects become more prominent in the retrieval of atmospheric trace gases and aerosols. In the Sentinel-5P (S5P) nitrogen dioxide (NO₂) algorithm an one dimensional (1D) horizontal homogeneous Lambertian cloud layer is used for cloud correction. Furthermore, 1D radiative transfers models are used in the NO₂ algorithm and also in the S5P UV Absorbing Aerosol Index (AAI) algorithm. However, in reality clouds are 3D objects, they are not spatially homogeneous in brightness, and they can have effects on neighbouring clear-sky pixels by casting shadows on lower clouds or on the ground surface or by scattering light into the pixels.

In the S5P NO₂ retrieval algorithm the retrieved slant column density is translated to a vertical column density (VCD) by correcting for the light path using pre-calculated air-mass factors (AMF) When a cloud shadow is cast on the surface the downward light intensity is reduced, altering the average observed light path. This lowers the sensitivity of the measurement for the lower atmospheric layers and thus influences the AMF and the resulting NO₂ VCD.

Aided by SUOMI-NPP VIIRS data to identify the pixels affected by cloud shadows, we observed such cloud shadow effects in S5P NO₂ data. These cases revealed that the cloud shadow impact on the NO₂ product is almost entirely mitigated by an albedo correction present in the algorithm.

In addition, we use the newly developed 3D radiative transfer KNMI model (MONKI), to simulate different cloud and scene scenarios. These simulations give more insight into the effect of cloud shadows on the NO₂ and AAI retrievals and the different sensitivities. For the AAI these simulations explain that the traditional AAI formula intrinsically already corrects for the cloud shadow effects.

References

Trees, V. J. H. et al., Cancellation of cloud shadow effects in the absorbing aerosol index retrieval algorithm of TROPOMI, submitted to AMTD, 2024b.

The Greenhouse gases Monitoring Instrument (GMI) on Chinese hyperspectral satellite GF5-02 can provide more abundant observations of global atmospheric CO₂ which plays an important role in climate research. CO₂ retrieval precision is the key to determine the application value of the GMI. In order to reduce the influence of atmospheric scattering on retrieval, we combined the Directional Polarimetric Camera (DPC) data on the same satellite to improve the anti-interference ability of GMI's CO₂ retrieval and ensure its retrieval precision. To realize the reliability and feasibility of the collaborative use of GMI and DPC, this paper designs the pointing registration method of the GMI based on the coastline observations, the spatial resolution matching method and the collaborative cloud screening method of the GMI and DPC observations. With the combination of DPC which supplied the spectral data and aerosol product, the retrieval ability of the Coupled Bidirectional reflectance distribution function CO₂ Retrieval (CBCR) method developed for GMI CO₂ retrieval was improved as the retrieval efficiency of CO₂ products increased by 27% and the CO₂ retrieval precision increased to 1.5 ppm. Meanwhile, the collaborative use not only guaranteed the GMI's ability to detect global and area CO₂ concentration distribution characteristics like the significant concentration differences between the northern and southern hemispheres in winter and high CO₂ concentration in urban agglomeration areas caused by human activities, but also extended GMI’s potential of monitoring anomalous events like Tonga volcanic eruption.

The DaQi-1 (DQ-1) equipped with the Aerosol and Carbon Detection Lidar (ACDL) payload was successfully launched into a sun-synchronous polar orbit on 16 April 2022 with an equator-crossing local time of around 13:30 p.m., and a 51-day repeat cycle. The ACDL is the world’s first space-borne integrated path differential absorption (IPDA) lidar instrument. It was primarily designced to measure XCO2 and aerosol .

The LIDAR onboard DQ-1 for XCO2 is based on IPDA mechanism, this technique is different from the Near Infrared (NIR) hyperspectral, thus, the CO2 profile are weighted by different weight function from NIR and IPDA, which can lead to differences in XCO2. In order to quantify the differences, a Carbon Model are used for simulating the XCO2 from NIR and IPDA. Moreover, a method is established for compensating or reducing the difference, and a feasible trick for calibration of XCO2 of IPDA by TCCON are proposed.

As the DQ-1 and OCO-2 in A-train orbit with 2 hours away, therefore, we focus on studying the difference between XCO2 from OCO-2 and DQ-1. On the differential analysis, we found uncovered intriguing findings that not only allow for the identification whether the CO2 concentration near the surface is higher than that in other layers, but also provide fresh insights into the vertical transport mechanisms of CO2, enhancing our understanding of its distribution and dynamics within the Earth's atmosphere

Atmospheric carbon dioxide (CO2) is the primary anthropogenic driver of climate change, accounting for more than half of the total effective radiative forcing (ERF). The accurate monitoring of carbon dioxide is essential to study the global carbon cycle and radiation budget on Earth. The Aerosol and Carbon Detection Lidar (ACDL) instrument, as the first space-borne integrated path differential absorption (IPDA) light detection and ranging (Lidar), was successfully launched in April 2022 onboard the DaQi-1 (DQ-1) satellite. ACDL enables observations to be taken at all latitudes and all times of year owing to their illumination, which allows a new perspective to quantify the global spatial distribution of atmospheric CO2. Although the active techniques have some significant benefits, evaluating and attributing the measurement error is necessary for a rigorous error budget to ensure a high-quality CO2 measurement. In this paper, the performance of the IPDA lidar was evaluated to meet the global weighted column-averaged dry air mixing ratio of carbon dioxide (XCO2) measurement requirements of less than 1 ppm. The random errors resulting from the noise associated with the detection of the lidar signals were assessed. The simulations of ACDL lidar were conducted. Results showed that the random error was distributed in the range of 0-1.5 parts per million (ppm) with 50 km averaging over land surfaces and 50 km averaging over oceans. In addition, the systematic errors arising from the laser pulse energy, spectral purity, doppler shift, and atmospheric factors, transmitter, and receiver were also analyzed. The uncertainties of surface pressure were found to be the major source of error, followed by the Doppler shift. This study can help to improve the understanding of the measurement uncertainties and provide a reference for CO2 retrievals and validation.

This study conducts an in-depth analysis of the deployment and integration of renewable energy technologies within the United Kingdom and evaluates their impact on atmospheric carbon dioxide (CO₂) emissions. Data from the Orbiting Carbon Observatory-2 (OCO-2), which employs advanced spectrometers for precise measurements of carbon columns, was utilized to quantify atmospheric CO₂ levels. The current data show that despite an increase in the deployment of renewable technologies in the UK, a continuous rise in atmospheric CO2 has been observed. The findings suggest that a more rapid and comprehensive uptake of renewable energy technologies, along with other measures are pivotal for mitigating the increase in CO2 levels. The research underscores the global nature of atmospheric challenges, highlighting that actions taken in one region can have a global effect. A closer look at the OCO-2 data also sheds light on the complexity of atmospheric CO2 dynamics, offering insights into how different regions contribute to global CO2 levels. By combining an in – depth analysis of OCO-2 data with a critical evaluation
of renewable energy strategies, this research contributes valuable insights on climate action and sustainable energy policy development.

This paper evaluates the recently introduced phase histogram (PH) technique for estimating forest height and vertical structure using theoretical modeling and synthetic aperture radar (SAR) data experiment, and makes comparison with the well-known SAR tomography (TomoSAR) technique. Using multiple SAR images, TomoSAR allows for a direct imaging of the three-dimensional (3D) electromagnetic structure of the vegetation layer, from which biophysical parameters such as forest height and terrain topography can be extracted[1], [2]. The PH technique assigns each pixel in a SAR interferogram to a specific height bin based on the value of the interferometric phase, allowing for a local estimation of the vertical profile of forest scattering by accumulation of pixels fall within a given spatial window[3]–[5].

The main goal of this paper is to understand the limits of the PH technique and the main contributions of this paper are:

1) Using multi-polarized monostatic L-Band TomoSense data to analyze the performance of PH technique in reconstructing the vertical structure of the forest;

2) Introducing a new physical modeling of PH technique, and quantitatively characterizing the height dispersion of the phase histogram as a function of the number of scatterers within any SAR pixel.

To do that, we first compare the PH technique against TomoSAR on an experimental ground by analyzing L-Band tomographic data from the ESA airborne campaign TomoSense, flown in 2020 at the Kermeter area in the Eifel Park, North-West Germany. The data analyzed in this paper feature 30+30 overpasses acquired along two opposite flight headings, and provide a vertical resolution consistently better than 5 m on the whole area of interest[6].

The analysis we present considers the evaluation of the vertical profile of forest scattering and forest height retrieval. In the evaluation of the forest vertical profile, we take the result by TomoSAR as the auxiliary verification against which to evaluate the PH technique, by virtue of the fine vertical resolution ensured by TomoSense data. Certain areas of interest in leaf area density (LAD) products from terrestrial laser scanning (TLS), Unoccupied Aerial Vehicle Lidar (UAV-LS) and airborne laser scanning (ALS) are taken as references[7]–[9]. Concerning the estimation of forest height, both TomoSAR and the PH technique are evaluated against the canopy height model (CHM) generated by ALS. The observed results are interpreted in light of a simple physical model to characterize phase histograms depending on the number of scatterers within the SAR resolution cell, on which basis we derive analytical expressions to predict height dispersion in phase histograms.

The conclusion from both experimental and theoretical results is that phase histograms cannot correctly reproduce forest structure, unless the distribution of scatterers within the SAR resolution cell is characterized by a single dominant scatterer. However, the PH technique can be used to produce a fairly good estimate of forest height. In particular, TomoSAR and the PH technique are observed to produce an average root mean square error (RMSE) of 2.8 m and 4.45 m in NW flight data, and 1.84 m and 5.46m in SE flight data, respectively. Overall, the analysis in this paper demonstrates, both theoretically and experimentally, that the PH technique cannot achieve the same performance as multi-baseline tomography when applied to lower frequency data at a resolution of few meters. Yet, even in these conditions we remark that the PH technique is inherently best suited for the analysis of high- or very-high resolution data, which suggests its use in the context of higher frequency SAR Missions (e.g.: Tandem-X) and when there are few acquisitions available.

[1] S. Tebaldini, D. Ho Tong Minh, M. Mariotti d’Alessandro, L. Villard, T. Le Toan, and J. Chave, “The status of technologies to measure forest biomass and structural properties: State of the art in SAR tomography of tropical forests,” Surv. Geophys., vol. 40, no. 4, pp. 779–801, 2019.

[2] M. M. D’Alessandro and S. Tebaldini, “Digital Terrain Model Retrieval in Tropical Forests Through P-Band SAR Tomography,” IEEE Trans. Geosci. Remote Sens., vol. 57, no. 9, pp. 6774–6781, 2019.

[3] R. N. Treuhaft et al., “Vegetation profiles in tropical forests from multibaseline interferometric synthetic aperture radar, field, and lidar measurements,” J. Geophys. Res. Atmos., vol. 114, no. D23, 2009.

[4] G. H. X. Shiroma and M. Lavalle, “Digital terrain, surface, and canopy height models from InSAR backscatter-height histograms,” IEEE Trans. Geosci. Remote Sens., vol. 58, no. 6, pp. 3754–3777, 2020.

[5] Y. Lei, R. Treuhaft, and F. Gonçalves, “Automated estimation of forest height and underlying topography over a Brazilian tropical forest with single-baseline single-polarization TanDEM-X SAR interferometry,” Remote Sens. Environ., vol. 252, p. 112132, 2021.

[6] S. Tebaldini et al., “TomoSense: A unique 3D dataset over temperate forest combining multi-frequency mono-and bi-static tomographic SAR with terrestrial, UAV and airborne lidar, and in-situ forest census,” Remote Sens. Environ., vol. 290, p. 113532, 2023.

[7] F. Pimont, D. Allard, M. Soma, and J.-L. Dupuy, “Estimators and confidence intervals for plant area density at voxel scale with T-LiDAR,” Remote Sens. Environ., vol. 215, pp. 343–370, 2018.

[8] B. Brede, H. M. Bartholomeus, N. Barbier, F. Pimont, G. Vincent, and M. Herold, “Peering through the thicket: Effects of UAV LiDAR scanner settings and flight planning on canopy volume discovery,” Int. J. Appl. Earth Obs. Geoinf., vol. 114, p. 103056, 2022.

[9] D. L. B. Jupp, D. S. Culvenor, J. L. Lovell, G. J. Newnham, A. H. Strahler, and C. E. Woodcock, “Estimating forest LAI profiles and structural parameters using a ground-based laser called ‘Echidna®,” Tree Physiol., vol. 29, no. 2, pp. 171–181, 2009.

Synthetic Aperture Radar Interferometry (InSAR) is widely used to identify potentially unstable slopes and monitor typical landslides. However, the atmospheric delay distortion has a great effect on InSAR landslide detection and monitoring. Generally, landslide detection and boundary delineation are realized by screening InSAR deformation rates on coherent points, but inaccurate deformation distorted by the atmospheric delay results in poor performances. In addition, long-term InSAR monitoring is important in landslide mechanism analysis and early warning, which is sensitive to atmospheric delay distortion. Although the conventional empirical phase-elevation models or external databased methods can correct atmospheric delay to some extent it is still limited by atmospheric delays in alpine valley areas due to the highly heterogeneous refractivity distributions.

To correct the InSAR tropospheric delay over wide areas with steep terrains, we propose a multi-temporal moving-window linear model (MMLM). It is a linear phase-elevation model that estimates tropospheric delay phases from multi-temporal unwrapped phases over overlapping sliding windows. The proposed atmospheric correction model has the following three innovations:

(1) Because the relationship between tropospheric delay and elevation can be regarded as linear within a localized area. We solved the linear model of the SDFP (Slow Decorrelating Filtered Phase) points within the local spatial window and performed global smooth interpolation on the estimated parameters to ensure better inversion of the atmospheric phase of spatial heterogeneity.

(2) Since the turbulence mixing varies randomly in space and time and the elevation-dependent stratified delay presents a seasonal trend in time, the abundant observations ensure the robustness of linear estimation in local windows. Therefore, we exploit multi-temporal unwrapped phases to mitigate the influence of local turbulence.

(3) In the local window, the traditional LM method is sensitive to deformation signals and phase unwrapping errors, especially the one related to local terrains. To address them, we use the estimated deformation rate and phase closure residuals of unwrapping phases as the weight to restrain the influence of local-scale landslide deformation and unwrapping jump errors on the parameter estimation.

A simulation experiment was first conducted to demonstrate the effectiveness of the proposed model. We then evaluated the MMLM model through comparison with the ERA5, GACOS, spatial-temporal filtering, and traditional linear model using descending and ascending Sentinel-1 data over the reservoir area of the Lianghekou hydropower station. Our major findings can be summarized as follows:

(1) The MMLM models multi-temporal unwrapping phases to estimate tropospheric delay, mitigating the influence of local turbulent phase, local landslide deformation, and local phase unwrapping jump error on parameter estimation.

(2) The MMLM establishes an empirical linear model for the phase within each sliding local window and interpolates the parameters to each SDFP target, to effectively characterize the spatial heterogeneous atmospheric path delay.

(3) The comparisons between MMLM and various tropospheric delay correction methods demonstrated its strengths of tropospheric corrections for individual interferograms, deformation rates, and displacement time series. Among the above-mentioned methods, the STD value of the original unwrapped phases shows the largest decrease of more than 35% and 50% after correction by the MMLM model for the descending and ascending Sentinel-1 tracks, respectively.

(4) Refined InSAR landslides investigation in the Lianghekou reservoir area was carried out with the correction by the MMLM model. The corrected deformation rate map improved the landslide hazard identification and boundary delineation, which has been validated by UAV image interpretation and field survey. Moreover, the refined movement evolution time series was retrieved, which can be helpful for disaster early warning and the analysis of landslide mechanisms.

Synthetic Aperture Radar (SAR) has emerged as a powerful remote sensing technique for mapping and characterizing the Earth's surface regardless of weather and illumination conditions. By exploiting the interferometric properties of SAR data acquired from multiple imaging geometries, SAR tomography (TomoSAR) enables the reconstruction of three-dimensional images, offering insights into the vertical structure of scattering targets.

The principle underlying SAR tomography is rooted in the exploitation of the phase differences between coherent SAR observations acquired from slightly different viewpoints. These phase variations encode information about the spatial distribution of scattering sources within the imaged volume. Through inversion algorithms, SAR tomography can reconstruct the vertical backscatter distribution of forest areas and provide crucial information for the evaluation of global carbon stocks, tracking changes in natural habitats, and territorial planning and management.

To obtain the vertical profile of scattering mechanisms, one should observe the scene with a long wavelength instrument (P-bands or L-band) to ensure sufficient penetration. Moreover, the vertical resolution of the tomographic product depends on the baselines between the acquisitions. The viewpoints should be different enough otherwise, land and canopy cannot be distinguished. The aforementioned conditions are realized by the SAOCOM L-band mission, a two-satellite constellation launched by CONAE in 2018 and 2020. The low orbital control makes it possible to obtain interferometric stacks with distributed vertical wavenumbers.

In this work, we have demonstrated the feasibility of measuring the forest height with L-band spaceborne data. The utilized SAOCOM stack consists of nine images acquired over the Amazonas, the largest Brazilian state, which is mostly covered by tropical jungle. The data spans over two months and is sampled regularly every eight days. The research explored different TomoSAR techniques, including Fourier beamforming, Capon estimation, and MUSIC estimation. We have found that by using the first two methods, the vertical resolution was insufficient to discriminate between canopy and ground. However, the MUSIC approach showed impressive results, clearly separating the two layers.

The outcome of the analysis is a 2D forest height map, considering the upper TomoSAR envelope. It was validated with a reference tree canopy height map provided by UND GLAD, generated by integrating GEDI and Landsat. The accordance between the two measures is very good, especially considering the non-favorable vertical resolution of the SAOCOM stack. Residual was investigated, showing that the estimation is almost unbiased and has a standard deviation of about 3 m. Results are promising, especially in view of future long wavenumber repeat pass missions, such as BIOMASS.