Alignment Procedure Research Articles

Key Frame Selection for Temporal Graph Optimization of Skeleton-Based Action Recognition

Graph neural networks (GNNs) are extensively utilized to capture the spatial–temporal relationships among human body parts for skeleton-based action recognition. However, due to the inefficient information propagation caused by redundant sampling of video frames in the temporal domain, we focus on refining temporal graphs through key frame selection. To this end, we propose a multi-stage key frame selection (MSKFS) method, aiming to find the most representative frames as the graph nodes to learn compact temporal graph representations of human skeletons for action recognition. The MSKFS progressively selects key frames in two stages: (1) salient posture frame selection based on the global dynamics of body parts and (2) key frame refinement and alignment according to intra-frame correlations. The first stage captures the most salient information and aligns the corresponding information of skeleton sequences within the same category. The second stage enriches the subtle information for the integrity of the information derived by the salient frames. Moreover, variational inference is applied to differentiate the key frame refinement and alignment procedure, allowing the end-to-end optimization of arbitrary graph-based models to represent the obtained compact graph for skeleton-based action recognition. Our MSKFS method achieves state-of-the-art performances on two challenging action recognition datasets.

Five-analyzer Johann spectrometer for hard X-ray photon-in/photon-out spectroscopy at the Inner Shell Spectroscopy beamline at NSLS-II: design, alignment and data acquisition.

Here, a recently commissioned five-analyzer Johann spectrometer at the Inner Shell Spectroscopy beamline (8-ID) at the National Synchrotron Light Source II (NSLS-II) is presented. Designed for hard X-ray photon-in/photon-out spectroscopy, the spectrometer achieves a resolution in the 0.5-2 eV range, depending on the element and/or emission line, providing detailed insights into the local electronic and geometric structure of materials. It serves a diverse user community, including fields such as physical, chemical, biological, environmental and materials sciences. This article details the mechanical design, alignment procedures and data-acquisition scheme of the spectrometer, with a particular focus on the continuous asynchronous data-acquisition approach that significantly enhances experimental efficiency.

Quantifying marine growth on pipelines with archival remotely operated vehicle footage – a case study

Abstract. Marine growth can substantially affect the hydrodynamic properties of subsea infrastructure. The ability to accurately quantify this growth is essential for designing and maintaining subsea structures. Traditionally, quantifying marine growth has been done by applying simple ratio equations putting in relationship the known distance of an object and its length observed in the image to the length of an unknown object only observed in an image. This method is very sensitive if measurements are not taken in the same plane. Hence, this study aims to establish a method for creating a 3D model of a subsea pipeline segment based on Structure from Motion (SfM). The proof of concept is done using a case study. For this study, Remotely Operated Vehicle (ROV) footage from a 2019 inspection survey is utilised. The scenario in which the procedure is tested represents a difficult, but realistic test case. Despite the lack of metadata, video quality and obstructions (e.g. watermarks), meaningful results were produced. The SfM workflow included masking obstructions (e.g. watermarks), and an iterative camera alignment procedure to generate 3D models, enabling the measurement of marine growth using either the models or ortho images created. While the resulting models were sufficiently accurate to provide meaningful measurements for harder growth types, the method faced challenges in accurately modelling softer growth types, suggesting the need for further refinement. Nevertheless, SfM was proven to be a more reliable method of measuring biofouling than the ratio approach that been traditionally used.

Met4DX: A Unified and Versatile Data Processing Tool for Multidimensional Untargeted Metabolomics Data.

Liquid chromatography-mass spectrometry (LC-MS) is a powerful tool in untargeted metabolomics, enabling the high-sensitivity and high-specificity characterization of metabolites. The integration of ion mobility (IM) with LC-MS, known as LC-IM-MS, enhances the analytical depth, facilitating more comprehensive metabolite profiling. However, the complexity of data generated by these technologies presents significant challenges in data processing. Addressing these challenges, we developed Met4DX, a unified and versatile software tool for processing both 3D and 4D untargeted metabolomics data. Met4DX incorporates a new MS1-oriented peak detection approach coupled with our bottom-up assembly algorithm, enabling highly sensitive and comprehensive peak detection in untargeted metabolomics data. Additionally, Met4DX employs a uniform quantification strategy to enhance the precision of peak integration across different samples. The software provides a user-friendly interface that simplifies data processing with default parameter sets, consolidating peak detection, alignment, quantification, and other procedures into a single streamlined workflow. Together, Met4DX offers a comprehensive solution for multidimensional metabolomics data processing, transforming raw data from diverse MS instruments into a final feature table containing quantification and identification results. We postulate Met4DX facilitates metabolite discovery in biological samples by deciphering the complex untargeted metabolomics data. Met4DX is freely available on the Internet (https://met4dx.zhulab.cn/).

Preliminary beam formations after the upgrade of the TIARA two microbeam systems

This paper presents a comprehensive study on the enhancements made to the heavy-ion microbeam (H-MB) and light-ion microbeam (L-MB) systems at the Takasaki Institute for Advanced Quantum Science, part of the National Institutes for Quantum Science and Technology (QST), and their implications for future research in ion microbeam technology. Our work addresses the critical need for submicron beam sizes in advanced ion microbeam applications, a challenge previously unmet due to limitations in the original system designs and aging components. The manuscript details the extensive upgrades undertaken for both the H-MB and L-MB systems, including the implementation of new quadruple magnets, refined demagnification strategies, and advanced control systems, culminating in the successful formation of submicron ion beams. The meticulous alignment procedures for both the H-MB and L-MB have enabled precise ion beam focusing. The current beam sizes have been measured at 1.0 × 1.1 µm2 with a beam current of approximately 10 pA for the H-MB, and 500 × 600 nm2 with a beam current of approximately 50 pA for the L-MB at a 3 MeV H+ beam.

An instrumentation guide to measuring thermal conductivity using frequency domain thermoreflectance (FDTR).

Frequency Domain Thermoreflectance (FDTR) is a versatile technique used to measure the thermal properties of thin films, multilayer stacks, and interfaces that govern the performance and thermal management in semiconductor microelectronics. Reliable thermal property measurements at these length scales (≈10nm to ≈10μm), where the physics of thermal transport and phonon scattering at interfaces both grow in complexity, are increasingly relevant as electronic components continue to shrink. While FDTR is a promising technique, FDTR instruments are generally home-built; they can be difficult to construct, align, and maintain, especially for the novice. Our goal here is to provide a practical resource beyond theory that increases the accessibility, replicability, and widespread adoption of FDTR instrumentation. We provide a detailed account of unpublished insights and institutional knowledge that are critical for obtaining accurate and repeatable measurements of thermal properties using FDTR. We discuss component selection and placement, alignment procedures, data collection parameters, common challenges, and our efforts to increase measurement automation. In FDTR, the unknown thermal properties are fit by minimizing the error between the phase lag at each frequency and the multilayer diffusive thermal model solution. For data fitting and uncertainty analysis, we compare common numerical integration methods, and we compare multiple approaches for fitting and uncertainty analysis, including Monte Carlo simulation, to demonstrate their reliability and relative speed. The instrument is validated with substrates of known thermal properties over a wide range of isotropic thermal conductivities, including Borofloat silica, quartz, sapphire, and silicon.

Testing the Capability of Embedding-Based Alignments on the GST Superfamily Classification: The Role of Protein Length.

In order to shed light on the usage of protein language model-based alignment procedures, we attempted the classification of Glutathione S-transferases (GST; EC 2.5.1.18) and compared our results with the ARBA/UNI rule-based annotation in UniProt. GST is a protein superfamily involved in cellular detoxification from harmful xenobiotics and endobiotics, widely distributed in prokaryotes and eukaryotes. What is particularly interesting is that the superfamily is characterized by different classes, comprising proteins from different taxa that can act in different cell locations (cytosolic, mitochondrial and microsomal compartments) with different folds and different levels of sequence identity with remote homologs. For this reason, GST functional annotation in a specific class is problematic: unless a structure is released, the protein can be classified only on the basis of sequence similarity, which excludes the annotation of remote homologs. Here, we adopt an embedding-based alignment to classify 15,061 GST proteins automatically annotated by the UniProt-ARBA/UNI rules. Embedding is based on the Meta ESM2-15b protein language. The embedding-based alignment reaches more than a 99% rate of perfect matching with the UniProt automatic procedure. Data analysis indicates that 46% of the UniProt automatically classified proteins do not conserve the typical length of canonical GSTs, whose structure is known. Therefore, 46% of the classified proteins do not conserve the template/s structure required for their family classification. Our approach finds that 41% of 64,207 GST UniProt proteins not yet assigned to any class can be classified consistently with the structural template length.

Determination of five-parameter grain boundary characteristics in nanocrystalline Ni-W by scanning precession electron diffraction tomography

Determining the full five-parameter grain boundary characteristics from experiments is essential for understanding grain boundaries impact on material properties, improving related models, and designing advanced alloys. However, achieving this is generally challenging, in particular at nanoscale, due to their 3D nature. In our study, we successfully determined the grain boundary characteristics of an annealed nickel-tungsten alloy (NiW) nanocrystalline needle-shaped specimen (tip) containing twins using Scanning Precession Electron Diffraction (SPED) Tomography. The presence of annealing twins in this face-centered cubic (fcc) material gives rise to common reflections in the SPED diffraction patterns, which challenges the reconstruction of orientation-specific virtual dark field (VDF) images required for tomographic reconstruction of the 3D grain shapes. To address this, an automated post-processing step identifies and deselects these shared reflections prior to the reconstruction of the VDF images. Combined with appropriate intensity normalization and projection alignment procedures, this approach enables high-fidelity 3D reconstruction of the individual grains contained in the needle-shaped sample volume. To probe the accuracy of the resulting boundary characteristics, the twin boundary surface normal directions were extracted from the 3D voxelated grain boundary map using a 3D Hough transform. For the sub-set of coherent Σ3 boundaries, the expected {111} grain boundary plane normals were obtained with an angular error of <3° for boundary sizes down to 400 nm². This work advances our ability to precisely characterize and understand the complex grain boundaries that govern material properties.

Optimizing Sample Preparation for Destructive Testing in Resistance Spot Welding

Abstract The integrity and reliability of resistance spot welding, a crucial process in various manufacturing industries, are significantly influenced by the quality of sample preparation. This study focuses on advancing the methodologies and practices in the preparation of samples for resistance spot welding destructive tests, aiming to improve the precision and repeatability of the results. We discuss comprehensive approaches encompassing material selection, surface treatment, dimensional accuracy, and alignment procedures, emphasizing the importance of meticulous handling and preparation techniques, A significant focus is placed on the utilization of sophisticated alignment and measurement tools to enhance the consistency of the welding area subjected to destructive testing. our study demonstrates a significant improvement in the reliability and accuracy of welding machine calibration, ultimately contributing to the overall quality of the welding process. By integrating these refined processes, the study highlights a significant improvement in the reliability of welding tests, contributing to enhanced welding quality and performance in practical applications. Our findings underscore the critical role of detailed and standardized sample preparation in achieving accurate and reliable resistance spot welding destructive test outcomes, providing valuable insights for both industrial applications and further research in welding technology.

An alignment algorithm using coherent twin boundaries as internal reference in 3D-EBSD.

A three-dimensional (3D) microstructural volume is reconstructed from a stack of two-dimensional sections which was obtained by serial sectioning coupled with electron back scattering diffraction (EBSD) mapping of a 316L austenitic stainless steel. A new alignment algorithm named linear translation by minimising the indicator (LTMI) is proposed to reduce the translational misalignments between adjacent sections by referencing to coherent twin boundaries which are flat and lying on {111} planes. The angular difference between the measured orientation of a flat twin boundary and that of the {111} plane is used as an indicator of the accuracy of the alignment operations. This indicator is minimised through linear translations of the centroids of triangular facets, which constitute grain boundaries at a distance not restricted by the in-plane step size of the EBSD maps. And hence the systematic trend in the translational misalignments can be effectively reduced. The LTMI alignment procedure proposed herein effectively corrects the misalignments remained by other methods on a 3D-EBSD data prepared using serial sectioning methods. The accuracy in distinguishing between coherent and incoherent twin boundaries is significantly improved.

Sensor-Fused Nighttime System for Enhanced Pedestrian Detection in ADAS and Autonomous Vehicles.

Ensuring a safe nighttime environmental perception system relies on the early detection of vulnerable road users with minimal delay and high precision. This paper presents a sensor-fused nighttime environmental perception system by integrating data from thermal and RGB cameras. A new alignment algorithm is proposed to fuse the data from the two camera sensors. The proposed alignment procedure is crucial for effective sensor fusion. To develop a robust Deep Neural Network (DNN) system, nighttime thermal and RGB images were collected under various scenarios, creating a labeled dataset of 32,000 image pairs. Three fusion techniques were explored using transfer learning, alongside two single-sensor models using only RGB or thermal data. Five DNN models were developed and evaluated, with experimental results showing superior performance of fused models over non-fusion counterparts. The late-fusion system was selected for its optimal balance of accuracy and response time. For real-time inferencing, the best model was further optimized, achieving 33 fps on the embedded edge computing device, an 83.33% improvement in inference speed over the system without optimization. These findings are valuable for advancing Advanced Driver Assistance Systems (ADASs) and autonomous vehicle technologies, enhancing pedestrian detection during nighttime to improve road safety and reduce accidents.

Deep Learning‐Enhanced Fiber‐Coupled Laser Diode‐Based Photoacoustic Microscopy for High‐Resolution Microvasculature Imaging

AbstractBlood vessel imaging provides crucial information for physiological monitoring and disease diagnosis. Optical‐resolution photoacoustic microscopy (PAM) is gaining intense attention in high‐resolution in vivo imaging of blood vessels. However, its applicability in clinical settings is hindered by the requirement of bulky and costly pulsed lasers as well as complex alignment procedures. Here, as a novel approach to high‐quality and cost‐effective microvasculature imaging, a dual‐wavelength fiber‐coupled laser diode‐based PAM (FC‐LD‐PAM) system is demonstrated on the mouse ear and brain in vivo. Furthermore, assisted with a deep learning method, the proposed technique effectively combines the advantages when using both small and large core‐size multimode fibers—successfully enhances the resolution of blurry mouse microvascular images obtained using a large core‐size fiber, revealing fine details that are comparable to those achieved using a small core‐size fiber; significantly reduces the overall time for data acquisition up to fourfold, improving the efficiency significantly in the imaging process. In addition, the proposed approach can achieve accurate and high‐resolution mapping of oxygen saturation in vivo, providing functional insight on living tissue. Therefore, the FC‐LD‐PAM can serve as a translational tool for high‐resolution imaging with enhanced accessibility and versatility in the biomedical field.

Effect Of Transcatheter Heart Valve Size And Deployment Height On Coronary Flow Obstruction

Coronary artery disease (CAD) is commonly found in patients with aortic stenosis; therefore, its prevalence is high in patients that are candidate for transcatheter aortic valve replacement (TAVR). As the TAVR procedure is expanding into younger and lower-risk population, the importance of early CAD monitoring and detection, as well as determination of potential impact of implant design or deployment strategy on CAD progression is of great importance. In this study, we investigate the effect of transcatheter heart valve (THV) size and deployment height on coronary flow obstruction during normotensive hemodynamic flow conditions. We employ unsteady, pulsatile flow, generated by a left heart model, and utilize clinically relevant geometry, pressure, and flow waveforms. Of particular interest is to understand the interaction between THV and anatomy regarding coronary flow obstruction. For a wide range of parameters, we examine two distinct THV deployment scenarios: nominal deployment – high coronary blockage (ND-HCB), and high deployment – low coronary blockage (HD-LCB). Additionally, we investigated the impact of the locations of coronary ostia and THV commissure alignment onto the risk of coronary flow obstruction. With the aid of 2D-2C particle image velocimetry (PIV), the distinctive flow patterns within the aortic sinus space are identified from the response of the flow to the presence of THV. Comparison between two THV deployment scenarios revealed that ND-HCB configurations lead to formation of stagnant flow zones, while the HD-LCB configuration provided an adequate washout of the entire coronary sinus. The optimal washout of the aortic root sinus is achieved when the distance between native leaflets and coronary ostia is maximized. Furthermore, results showed that up to 4% improvement in the coronary flow rate can be achieved if commissure alignment procedure is performed. As the coronary artery obstruction phenomenon is additive in its nature, a pre-existing stenotic condition in association with suboptimal THV deployment may lead to increased risk of full coronary blockage and immediate need for coronary intervention. In conclusion, this study using an experimental coronary assessment tool may provide potential insights to inform pre-procedural TAVR strategies in challenging patients’ anatomies, thereby maximizing the associated benefit.

Mapping Flows in the Cassava, Rice, Milk, and Fish Supply Chains of the Eastern Democratic Republic of the Congo: Is There Any Quiet Revolution Taking Place?

By applying a stacked survey approach to farmers, processors, and intermediaries, this study estimates and maps aggregate flows of fresh and processed cassava, rice, milk, and fish within the Bukavu-Uvira-Kalemie economic corridor of the Eastern Democratic Republic of the Congo (DRC). To achieve this, we develop and implement a comprehensive accounting method to better capture the complex set of activities within all four supply chains as well as an alignment procedure to ensure consistency in flows across their stacks. Apart from operating beyond mere subsistence levels, the findings of this study indicate that much of the vibrant agri-food value chain (AVC) developments observed in other low- and middle-income countries (LMIC) are yet to occur in the Eastern DRC: processing levels and new product development are low; intermediaries remain very present in agri-food markets; food losses are high in the midstream segments; and advance payments to farmers have not disappeared. While these observations provide for a dissonant case study, it underscores the huge potential for future upgrading of AVC in the Eastern DRC.

Measurement of Si pixel sensor alignment for the ALICE ITS detector

Abstract A Large Ion Collider Experiment (ALICE) experiment is one of the four experiments at the Large Hadron Collider (LHC) designed to investigate the status of matter under very high energy densities produced during heavy-ion collisions. The ALICE inner tracking system (ITS) consists of seven concentric cylindrical layers of monolithic silicon pixel sensors known as ALICE pixel detector (ALPIDE). The sensors are used to reconstruct the paths of charged particles generated in the collisions. The sensor alignment of the detector must be adjusted to a high precision standard. The adjustment objective is to obtain a detector that can undertake high-resolution measurements. This paper introduces a method for measuring the reference markers utilized in sensor alignment determination. Markers engraved at the chip corners have been detected using the Hough transform, Canny edge detection, and template matching techniques. The distances between two markers are measured to determine the accuracy of the pixel sensor alignment before and after assembly. The proposed methods exhibit an accuracy exceeding 99% and demonstrate high speed analysis. The average processing times for detecting the circle and cross markers are 105.9 ms/image and 113.8 ms/image, respectively. The sensor alignment of the detector must be adjusted to a high precision standard. However, recent studies have shown deviations of up to 5 μ m above the desired value in the measured sensor position. Such deviations do not represent a major issue, nevertheless it is important to measure them in order to speed-up and make more accurate the recursive track-based alignment procedure used to reconstruct the position of each pixel sensor in the tracking detector. The proposed method offers a promising solution to deliver precise and rapid measurements for a large number of examined objects.

Correction: A pre‑whitening with block‑bootstrap cross‑correlation procedure for temporal alignment of data sampled by eddy covariance systems

Correction: A pre‑whitening with block‑bootstrap cross‑correlation procedure for temporal alignment of data sampled by eddy covariance systems

The JCMT Transient Survey: Six Year Summary of 450/850 μm Protostellar Variability and Calibration Pipeline Version 2.0

The James Clerk Maxwell Telescope (JCMT) Transient Survey has been monitoring eight Gould Belt low-mass star-forming regions since 2015 December and six somewhat more distant intermediate-mass star-forming regions since 2020 February with the Submillimeter Common User Bolometer Array 2 on board JCMT at 450 and 850 μm and with an approximately monthly cadence. We introduce our pipeline v2 relative calibration procedures for image alignment and flux calibration across epochs, improving on our previous pipeline v1 by decreasing measurement uncertainties and providing additional robustness. These new techniques work at both 850 and 450 μm, where version 1 only allowed investigation of the 850 μm data. Pipeline v2 achieves better than 0.″5 relative image alignment, less than a tenth of the submillimeter beam widths. The version 2 relative flux calibration is found to be 1% at 850 μm and <5% at 450 μm. The improvement in the calibration is demonstrated by comparing the two pipelines over the first 4 yr of the survey and recovering additional robust variables with version 2. Using the full 6 yr of the Gould Belt survey, the number of robust variables increases by 50%, and at 450 μm we identify four robust variables, all of which are also robust at 850 μm. The multiwavelength light curves for these sources are investigated and found to be consistent with the variability being due to dust heating within the envelope in response to accretion luminosity changes from the central source.

Devising a procedure for the brightness alignment of astronomical frames background by a high frequency filtration to improve accuracy of the brightness estimation of objects

The object of this study is the background substrate of astronomical frames. To detect and compare the image of an object in a frame with its real image from astronomical catalogs, it is necessary to uniformly distribute the brightness of the background image substrate. Most often, the background alignment of astronomical frames is performed using the hardware calibration method applying the construction of service frames. However, it does not make it possible to eliminate the background from temporary stray light. Therefore, to solve this problem, a procedure has been proposed for brightness alignment of the background frame using high-pass filtering. For high-pass filtering of images, three high-pass filters were considered – an ideal filter, a Butterworth filter, and a Gaussian filter. To remove coarse-grained image components from the image, a high-pass filter was used, which attenuates low-frequency harmonics of the image spectrum while simultaneously passing high-frequency harmonics. Applying the devised procedure for brightness alignment of the background substrate of the frame has made it possible to increase the signal-to-noise ratio and reduce the dynamic range of the background substrate of the image. The study showed that when assessing brightness and identifying frames, the fitting provides better accuracy of reference to the starry sky. Also, the standard deviation of frame identification errors in this case is 5–7 times less than without using the devised procedure. The devised procedure for brightness alignment of the background frame substrate was tested in practice within the framework of the CoLiTec project. It was implemented at the stage of intra-frame processing in the Lemur software for automated detection of new objects and tracking of known objects

Spatially synchronous fringe detection: a method to facilitate the alignment of an echelle grating mosaic by separating the compensative angular errors

Alignment of mosaic gratings is traditionally supported by two interferometric verifications: on the zero order to verify the grating surfaces and on the blaze to verify the groove direction. In the case of low frequency echelle grating an interferometric measurement on the zero order is hardly feasible due to extremely low contrast of the fringes. The complete alignment has then to be carried out on high order (close to the blaze) where the two misalignment errors (the tip and rotation) show the same effect on the interferogram. The acquisition of a low and a high diffraction order image simultaneously, referred to as the spatially synchronous fringe detection method (SSFD), is used to analyze the misalignment. Iterative adjustment with the autocollimation configuration at the low and high order is used to separate the compensative errors of Δθ x and Δθ z . A prototype mosaic with two 110mm×220mm segments has been aligned with the support of this method. A numerical simulation of the alignment procedure as well as the error orientation analysis of this mosaic grating are presented. The mosaic grating with an accuracy of Δθ x &lt;0.64µrad, Δθ y &lt;1.13µrad, Δθ z &lt;0.65µrad and a wavefront RMS error of 0.149λ has been completed. This method can greatly facilitate the alignment of an echelle mosaic for an astronomical spectrograph.

A pre-whitening with block-bootstrap cross-correlation procedure for temporal alignment of data sampled by eddy covariance systems

The eddy covariance (EC) method is a standard micrometeorological technique for monitoring the exchange rate of the main greenhouse gases across the interface between the atmosphere and ecosystems. One of the first EC data processing steps is the temporal alignment of the raw, high frequency measurements collected by the sonic anemometer and gas analyser. While different methods have been proposed and are currently applied, the application of the EC method to trace gases measurements highlighted the difficulty of a correct time lag detection when the fluxes are small in magnitude. Failure to correctly synchronise the time series entails a systematic error on covariance estimates and can introduce large uncertainties and biases in the calculated fluxes. This work aims at overcoming these issues by introducing a new time lag detection procedure based on the assessment of the cross-correlation function (CCF) between variables subject to (i) a pre-whitening based on autoregressive filters and (ii) a resampling technique based on block-bootstrapping. Combining pre-whitening and block-bootstrapping facilitates the assessment of the CCF, enhancing the accuracy of time lag detection between variables with correlation of low order of magnitude (i.e. lower than -1\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$-1$$\\end{document}) and allowing for a proper estimate of the associated uncertainty. We expect the proposed procedure to significantly improve the temporal alignment of the EC time-series measured by two physically separate sensors, and to be particularly beneficial in centralised data processing pipelines of research infrastructures (e.g. the Integrated Carbon Observation System, ICOS-RI) where the use of robust and fully data-driven methods, like the one we propose, constitutes an essential prerequisite.