Original Code Research Articles

Scientific modelling can be accessible, interoperable and user friendly: A case study for pasture and livestock modelling in Spain.

This article describes the adaptation of a non-spatial model of pastureland dynamics, including vegetation life cycle, livestock management and nitrogen cycle, for use in a spatially explicit and modular modelling platform (k.LAB) dedicated to make data and models more interoperable. The aim is to showcase to the social-ecological modelling community the delivery of an existing, monolithic model, into a more modular, transparent and accessible approach to potential end users, regional managers, farmers and other stakeholders. This also allows better usability and adaptability of the model beyond its originally intended geographical scope (the Cantabrian Region in the North of Spain). The original code base (written in R in 1,491 lines of code divided into 13 files) combines several algorithms drawn from the literature in an opaque fashion due to lack of modularity, non-semantic variable naming and implicit assumptions. The spatiotemporal rewrite is structured around a set of 10 namespaces called PaL (Pasture and Livestock), which includes 198 interoperable and independent models. The end user chooses the spatial and temporal context of the analysis through an intuitive web-based user interface called k.Explorer. Each model can be called individually or in conjunction with the others, by querying any PaL-related concepts in a search bar. A scientific dataflow and a provenance diagram are produced in conjunction with the model results for full transparency. We argue that this work demonstrates key steps needed to create more Findable, Accessible, Interoperable and Reusable (FAIR) models beyond the selected example. This is particularly essential in environments as complex as agricultural systems, where multidisciplinary knowledge needs to be integrated across diverse spatial and temporal scales in order to understand complex and changing problems.

Numerical solution of non-isothermal flow in oil reservoirs using OpenACC

In this work, we performed a numerical simulation of the non-isothermal flow in an oil reservoir. Also, we used the OpenACC API to parallelize specific parts of the original code, which allowed the simultaneous execution of different tasks on a NVIDIA GTX 970 G1 video card in a shared memory architecture. The problem was studied using a vertical producer well and static heating wells in a two-dimensional domain. We used the Control Volume Finite Difference (CVFD) method to discretize the governing equations and the Conjugate Gradients method to obtain the solutions (pressure and temperature) of the systems of algebraic equations. As a result, concerning computational performance, a significant reduction in execution time was obtained with the parallelized version.

Incidence of the Sample Size Distribution on One-Shot Federated Learning

Federated Learning (FL) is a learning paradigm where multiple nodes collaboratively train a model by only exchanging updates or parameters. This enables to keep data locally, therefore enhancing privacy - statement requiring nuance, e.g. memorization of training data in language models. Depending on the application, the number of samples that each node contains can be very different, which can impact the training and the final performance. This work studies the impact of the per-node sample size distribution on the mean squared error (MSE) of the one-shot federated estimator. We focus on one-shot aggregation of statistical estimations made across disjoint, independent and identically distributed (i.i.d.) data sources, in the context of empirical risk minimization. In distributed learning, it is well-known that for a total number of m nodes, each node should contain at least m samples to equal the performance of centralized training. In a federated scenario, this result remains true, but now applies to the mean of the per-node sample size distribution. The demo enables to visualize this effect as well as to compare the behavior of the FESC (Federated Estimation with Statistical Correction) algorithm - a weighting scheme which depends on the local sample size - with respect to the classical federated estimator and the centralized one, for a large collection of distributions, number of nodes, and features space dimension. **This is an MLBriefs article, the source code has not been reviewed!**<br> **The original source code is [[available here|https://github.com/gonzaq94/one-shot-fed-learning]] (last checked 2023/02/12).**

AirSeaFluxCode: Open-source software for calculating turbulent air-sea fluxes from meteorological parameters

The turbulent exchanges, or fluxes, of heat, moisture and momentum between the atmosphere and the ocean play a crucial role in the Earth’s climate system. Direct measurements of turbulent fluxes are very challenging and sparse, and do not span the full range of environmental conditions that exist over the ocean. This means that empirical “bulk formulae” parameterizations that relate direct flux observations to concurrent measurements of the mean meteorological and sea surface variables contain considerable uncertainty. In this paper, we present a Python 3.6 (or higher) open-source software package “AirSeaFluxCode” for the computation of the heat (latent and sensible) and momentum fluxes. Ten different parameterizations are included, each based on published descriptions or code and each derived from a different set of observations, or different assumptions about the turbulent exchange processes. They represent a range of current expert opinion on how the fluxes depend on mean properties and can be used to explore uncertainty in calculated fluxes. AirSeaFluxCode also allows the adjustment of the mean meteorological input parameters (air temperature, humidity and wind speed) from the height at which they are obtained to a user-defined output height. This height adjustment enables the comparison of measurements, or model-derived values, made at different heights above sea-level. The parameterizations calculate the fluxes using input parameters that are relatively easily to measure, or are available as model output: wind speed, air temperature, sea surface temperature, atmospheric pressure and humidity. Where original code is available we have compared its output with that of AirSeaFluxCode. Any changes made to increase consistency across algorithms by standardizing computational methods or calculation of meteorological variables, for example, are discussed and the impacts quantified: these are shown to be insignificant except for a few cases where conditions were extreme, and AirSeaFluxCode is shown to be robust. We also investigate the impact on the fluxes caused by different assumptions about the exchange processes, or the choices inherent in the implementation of the parameterizations. For example, sea surface temperature usually refers to data typically obtained at depths of between 1 and 10 m. However, since some parameterizations require a “skin” sea surface temperature, code that adjusts temperature at depth to skin temperature is included: this has a very significant impact on the fluxes. Selecting a parameterization that is appropriate for the available sea surface temperature will avoid the need to adjust the sea temperature data and the uncertainties associated with that adjustment, and will also avoid the biases due to use of the “wrong” measure of temperature. Significant differences also resulted from assumptions about the size of reduction in sea surface humidity to account for salinity effects: the uncertainty in the reduction factor needs to be quantified in future analyses. Fluxes in extreme conditions are particularly uncertain since the transfer coefficients in the different parameterizations vary most at very high and very low wind speeds. Low wind speeds are also challenging for numerical implementation since choices have to be made regarding: convergence criteria for the iterative calculation, inclusion of a parameterization for convective gustiness, or application of ad hoc limits to various parameters. All of these choices can significantly affect the flux estimates for light winds.

The Grossberg Code: Universal Neural Network Signatures of Perceptual Experience

Two universal functional principles of Grossberg’s Adaptive Resonance Theory decipher the brain code of all biological learning and adaptive intelligence. Low-level representations of multisensory stimuli in their immediate environmental context are formed on the basis of bottom-up activation and under the control of top-down matching rules that integrate high-level, long-term traces of contextual configuration. These universal coding principles lead to the establishment of lasting brain signatures of perceptual experience in all living species, from aplysiae to primates. They are re-visited in this concept paper on the basis of examples drawn from the original code and from some of the most recent related empirical findings on contextual modulation in the brain, highlighting the potential of Grossberg’s pioneering insights and groundbreaking theoretical work for intelligent solutions in the domain of developmental and cognitive robotics.

Statistical Analysis of the Probability of Interaction of Globular Clusters with Each Other and with the Galactic Center on the Cosmological Time Scale According to Gaia DR2 Data

This study is aimed at investigating the dynamic evolution of the orbits of stellar globular clusters (GCs). To integrate the orbits backward in time, the authors use models of the time-varying potentials derived from cosmological simulations, which are closest to the potential of the Galaxy. This allows for estimating the probability of close passages (collisions) of GCs with respect to each other and the Galactic center (GalC) in the Galaxy undergoing dynamic changes in the past. To reproduce the dynamics of the Galaxy in time, five potentials selected from the IllustrisTNG-100 large-scale cosmological database, which are similar in their characteristics to the current physical parameters of the Milky Way, are used. With these time-varying potentials, we have reproduced the orbital trajectories of 143 GCs 10 Gyr back in time using our original phi-GPU N-body code. Each GC was treated as a single physical particle with the assigned position and velocity of the cluster center from the Gaia DR2 observations. For each of the potentials, 1000 initial conditions were generated with randomized initial velocities of GCs within the errors of the observational data. In this study, we consider close passages to be passages with a relative distance of less than 100 pc and a relative speed of less than 250 km/s. To select clusters that pass at close distances from the GalC, the following criterion is applied based only on the relative distance: it must be less than 100 pc. Applying the above criteria, the authors obtained statistically significant rates of close passages of GCs with respect to each other and to the GalC. It has been determined that GCs during their evolution have approximately 10 intersecting trajectories with each other on the average and approximately 3 to 4 close passages near the GalC in 1 Gyr at a distance of 50 pc for each of the chosen potentials.

The possibility of translating ancient Latin poetry: some French, English and Bulgarian versions of Ovidian poetry and of a Latin epitaph

The present text tries to compare different translations and interpretations of Latin poetic works. An everlasting discussion about the possibility to preserve the initial meaning and to transfer original codes into a new language medium, is illustrated within translations in three modern languages, from different linguistic groups – French, Bulgarian and English. Though the examples given in Latin are of different authorships, they all belong to the poetic genre and aim to illustrate how modern languages could translate and interpret classic verses and to what extent contemporary languages are able to stay loyal to the original. Classical Latin poems of Ovid and epigraphic epitaphs are chosen as being well known and challenging for many translators devoted to ancient texts. The poems of the roman poet Ovid have been studied by a lot of interpreters of classical antiquity and those chosen herein like Shilling, Sharankov, Bonniec and Frazer are some of the most representative. In Bulgarian medium a significant scientist in translation of old texts is Nicolay Sharankov whose interest in roman epigraphic inscriptions led him to some outstanding achievements, which are ana-lyzed in this research. The study also compares different traditions and outlooks of how Latin poetry should be transmitted into the so-called source language or langue cible. Many diach-ronic studies show strong and feeble parts of different approaches. An overview of transla-tions made in long periods of time makes possible to outline different paradigms for transmit-ting meaning and content. One less known material referred to is the work of Desjardins and his studies of Latin inscriptions found in Bulgaria.

Monocular Depth Estimation: a Review of the 2022 State of the Art

We compare five monocular depth estimation methods based on deep learning. This comparison focuses on how well methods generalize rather than a quantitative comparison on a specific dataset. This study shows that while monocular depth estimation methods work well on images similar to training images, they often show artifacts when applied on images out of the training distribution. We evaluate the different methods with images similar to training data and images with unusual point of views (e.g. top-down) or paintings. The readers are invited to judge by themselves about the advantages and drawbacks of all methods by submitting their own images to the online demo associated with the present paper. **This is an MLBriefs article, the source code has not been reviewed!**<br> <span style="color:red">**BEST PAPER MLBRIEFS 2022**</span>. <br> <br> The codes used in the demos and publicly available are the following: * [[MiDaS method|MiDaS-main.zip]], * [[DPT method|DPT-master.zip]], * [[Adabins method|Adabins-main.zip]], * [[GLPDepth method|GLPDepth-main.zip]], and the original source codes are available here (last checked 2023/01/23): * [[MiDaS and DPT methods|https://github.com/isl-org/MiDaS]], * [[Adabins method|https://github.com/shariqfarooq123/AdaBins]], * [[GLPDepth method|https://github.com/vinvino02/GLPDepth]], * [[3DShape method|https://github.com/aim-uofa/AdelaiDepth]]

Performance Tests and Improvements on the rmcdhf and rci Programs of GRASP

The latest published version of GRASP (General-purpose Relativistic Atomic Structure Package), i.e., GRASP2018, retains a few suboptimal subroutines/algorithms, which reflect the limited memory and file storage of computers available in the 1980s. Here we show how the efficiency of the relativistic self-consistent-field (SCF) procedure of the multiconfiguration-Dirac–Hartree–Fock (MCDHF) method and the relativistic configuration-interaction (RCI) calculations can be improved significantly. Compared with the original GRASP codes, the present modified version reduces the CPU times by factors of a few tens or more. The MPI performances for all the original and modified codes are carefully analyzed. Except for diagonalization, all computational processes show good MPI scaling.

Structure and performance analysis of fusion positioning system with a single 5G station and a single GNSS satellite

ABSTRACT NaGlobal vigation Satellite System (GNSS) positioning technology is widely used for its high precision, global, and all-weather service. However, in complex environments such as urban canyons, GNSS performance is often degraded due to signal occlusion and even fails to achieve positioning due to the insufficient visible satellites. Because of the characteristics of large bandwidth, low latency, and high Base Station (BS) density, the fifth-Generation mobile communication (5G) technology has gradually become a trend for positioning in cities while offering traditional communication service. To supply the communication demands of the User Equipment (UE), only one BS is usually considered to establish a connection with the UE during the BS construction. However, the positioning accuracy with a single BS in urban canyons will be significantly reduced. To further improve the positioning accuracy in such extreme scenarios, this paper proposes a simplified 5G/GNSS fusion positioning system architecture using observations from only a 5G BS and a GNSS satellite. In this system, the GNSS receiver is mounted on the 5G BS, and the measurements provided by the receiver are used to form the differential code and complete the position estimation. The positioning mathematical models of the system based on the original code and differential code are derived. Then, the impacts of the measurements noise and the time synchronization error on the positioning accuracy are analyzed theoretically. Finally, the positioning performance is investigated by a set of simulation experiments. Numerical results show that under the existing 5G measurement noise and 2 m’s code measurement noise, the improvement of the differential code based fusion positioning compared with the 5G-only positioning is more than 32%, which is also about 6% higher than the original code based fusion positioning. Besides, this improvement is not affected by the time synchronization error between the BS and the GNSS satellite.

Development of an interactive web dashboard to facilitate the reexamination of pathology reports for instances of underbilling of CPT codes

Current Procedural Terminology Codes is a numerical coding system used to bill for medical procedures and services and crucially, represents a major reimbursement pathway. Given that pathology services represent a consequential source of hospital revenue, understanding instances where codes may have been misassigned or underbilled is critical. Several algorithms have been proposed that can identify improperly billed CPT codes in existing datasets of pathology reports. Estimation of the fiscal impacts of these reports requires a coder (i.e., billing staff) to review the original reports and manually code them again. As the re-assignment of codes using machine learning algorithms can be done quickly, the bottleneck in validating these reassignments is in this manual re-coding process, which can prove cumbersome. This work documents the development of a rapidly deployable dashboard for examination of reports that the original coder may have misbilled. Our dashboard features the following main components: (1) a bar plot to show the predicted probabilities for each CPT code, (2) an interpretation plot showing how each word in the report combines to form the overall prediction, and (3) a place for the user to input the CPT code they have chosen to assign. This dashboard utilizes the algorithms developed to accurately identify CPT codes to highlight the codes missed by the original coders. In order to demonstrate the function of this web application, we recruited pathologists to utilize it to highlight reports that had codes incorrectly assigned. We expect this application to accelerate the validation of re-assigned codes through facilitating rapid review of false-positive pathology reports. In the future, we will use this technology to review thousands of past cases in order to estimate the impact of underbilling has on departmental revenue.

Corrections to: “Accurate computation of gravitational field of a tesseroid” by Fukushima (2018) in J. Geod. 92(12):1371–1386

An accurate method with conditional split, double exponential quadrature rule, and numerical differentiation has been proposed in the paper “Accurate computation of gravitational field of a tesseroid” (Fukushima in J Geod 92(12):1371–1386, https://doi.org/10.1007/s00190-018-1126-2, 2018) to compute the gravitational field (i.e. gravitational potential, gravitational acceleration vector, and gravity gradient tensor) of a tesseroid. This study presents the corrections for some formulas in the main paper and electronic supplementary material of Fukushima (J Geod 92(12):1371–1386, https://doi.org/10.1007/s00190-018-1126-2, 2018). Moreover, the FORTRAN subroutines gtess (or qgtess) and ggtess (or qggtess) in the original codes xtess.txt (or xqtess.txt) in double (or quadrature) precision provided by Fukushima (J Geod 92(12):1371–1386, https://doi.org/10.1007/s00190-018-1126-2, 2018) are revised. The revised parts have impacts on the calculation of these components of the gravitational acceleration vector (g_{varPhi } and g_{varLambda }) and gravity gradient tensor (varGamma _{varPhi varPhi }, varGamma _{varPhi varLambda }, varGamma _{varPhi H}, varGamma _{varLambda varLambda }, varGamma _{varLambda H}, and varGamma _{H H}). The revised FORTRAN codes xtess.f90 and xqtess.f90 in double and quadrature precision are presented at the GitHub website https://github.com/xiaoledeng/xtess-xqtess. These revised FORTRAN codes can accurately compute the gravitational field of a tesseroid in double and quadrature precision no matter the computation point is located outside, near the surface of, on the surface of, or inside the tesseroid. They can be applied to calculate the gravitational field of the different layers (e.g. atmosphere, topography, crust, and mantle) of the Earth or other celestial bodies, which helps investigate the various geoscience applications, e.g. geoid determination in geodesy and gravity interpretation in geophysics.

Easy-to-use MPC tool for controlling chemical processes in a rigorous simulation environment

Rigorous process simulation has become a tool that academic and industrial environments are exploiting, mainly to extract information useful for maximizing profit. As a matter of fact, the detailed thermodynamic models contained in commercial or open-source software are able to represent the behavior of a chemical process far better than a linearized model. On the other hand, designing customized model predictive controllers (MPC) has proven to enhance process performance over traditional control architectures. Therefore, in this paper, we present the interaction of an easy-to-use MPC algorithm developed in Python with the rigorous simulator UniSim Design®. The communication exploits the UniSim Design® spreadsheets as the variables database to be read/written by Python, by stopping or not the simulation before every control action. The software communication has been properly developed so to maintain the flexibility of the original MPC code and to exploit different controller designs. Two different test cases are presented to show the effectiveness of the proposed methodology: a simple two-phase separator and a more complex debutanizer column. System identification is used to build the controller's linear models, various MPC designs differing in considering disturbances as measurable have been analyzed and satisfactory results are obtained.

A surface porosity approach for eliminating artificial ponding in coastal salt marsh simulations

Hydrodynamic processes over marsh topography are significantly affected by surface defects such as cuts and rills on channel berms and platforms. These meter-scale features are often missing in the model representation due to the spatial resolution available from data sources, as well as incomplete resolution in the model grid itself. To minimize the artificial hydraulic isolation in the numerical models, we propose implementing an effective porosity algorithm on the marsh surface by considering the fine-scale topography over marsh depressions that control the drainage process. The modification is carried out to eliminate artificial ponding effects observed in model simulations in Bombay Hook National Wildlife Refuge, DE, USA using the original FVCOM code. Results from the revised and original FVCOM models are compared with pressure gauge data collected from an isolated depression in the marsh platform. The new implementations for proper wetting and drying are efficient and accurate for hydrodynamic modeling inside a complex salt-marsh system, which constitutes a major breakthrough in the context of increasing need for better understanding of physical and morphological changes in valuable coastal ecosystems. • A permeable-seabed technique is proposed to improve wetting and drying in artificial hydraulic depressions over salt marsh surfaces. • A new set of the formula is implemented in a widely used coastal ocean model FVCOM and tested the model performance in a salt marsh system. • The detailed comparison against the traditional approach showed a substantial improvement in the marsh flooding and draining accuracy.

Introducing atmospheric photochemical isotopic processes to the PATMO atmospheric code

The study of production and depletion of chemical species is vital for the understanding of composition and evolution of planetary atmospheres. We present an implementation of photoinduced isotopic effects into the PATMO code (Ávila et al., 2021) code designed for the study of stable isotopes and photo-induced isotopic effects. With respect to the original code PATMO, where the photochemistry was not included, this report extends capability of the model to set photochemical processes for stable isotopes and thus enhancing its applicability. The PATMO code is flexible and allows the edition of new chemical reactions without need for hard code them. We also test how changes in spectral resolution affects the calculation of isotopic effects during the photodissociation of oxygen. We found that for a highly structured spectrum such as the Schumann-Runge band a spectral resolution larger than 0.005 nm is necessary for accurately modeling these isotopic effects. We also show that SO2 and SO photodissociation couples in a complex shielding fashion and significantly affects the photo-induced isotopic effects. The model was also benchmarked against today’s Earth atmosphere, where the solar UV flux and the ozone profile of the US Standard atmosphere of 1976 was reproduced with the simple Chapman mechanism and improved with the implementation of NOx and HOx radicals.

Image Unprocessing: A Pipeline to Recover Raw Data from sRGB Images

Access to high quality datasets is an essential condition for data-driven methods as it is known that mismatches between the distributions of training and test data may cause learning-based methods to fail. This issue has led to one of the most active research subjects in learning-based image restoration. For instance neural networks trained on unrealistic synthetic data may not generalize to real data even if they perform well on those synthetic data. This is specially problematic for image and video processing tasks, such as denoising, which are performed on raw data, since acquiring real raw datasets is not straightforward and is even impossible in some cases (acquiring a video dataset of real noise with clean ground-truth, for instance). Consequently, CNNs are often trained on synthetic data. Synthesizing realistic raw data is a difficult task and requires to invert properly the image processing pipeline. This paper focuses on the backward pipeline proposed by Brooks et al. [Unprocessing images for learned raw denoising, CVPR 2019] which aims at producing raw data from sRGB images. **This is an MLBriefs article, the source code has not been reviewed!**<br> **The original source code is [[available here|https://github.com/timothybrooks/unprocessing]] (last checked 2022/12/30).**

Metagenomic classification with KrakenUniq on low-memory computers.

The KrakenUniq software classifies reads from metagenomic samples to establish which organisms are present in the samples and estimate their abundance. The software is widely used used by researchers and clinicians in medical diagnostics, microbiome and environmental studies.Typical databases used by KrakenUniq are tens to hundreds of gigabytes in size. The original KrakenUniq code required loading the entire database in RAM, which demanded expensive high-memory servers to run it efficiently. If a user did not have enough physical RAM to load the entire database, KrakenUniq resorted to memory-mapping the database, which significantly increased run times, frequently by a factor of more than 100. The new functionality described in this paper enables users who do not have access to high-memory servers to run KrakenUniq efficiently, with a CPU time performance increase of 3 to 4-fold, down from 100+.

Rescue of a stone bridge with respect to current condition and standards

A three span stone masonry bridge dating back to the middle of the 19th century, still in the roadway network, is assessed to prevent its demolition. Its industrial heritage value and ecological concerns were the principal reasons for the assessment. The carbon footprint of the stone arch replacement would be approximately 200 t CO2 emission owing to 430 m3 reinforced concrete in the new structure. Besides the cultural monument, considerable energy, CO2 emissions and natural resources can be saved in accordance with the sustainable development goals. Standards, guides and commercial software do not provide an adequate support for the assessment of masonry arch bridges, therefore an innovative two phase application of a commercial linear sttructural analysis code and an original in-house code was developed for the purpose. Thousands of stone ach bridges still in service worlwide give the case study more general importance.

A Brief Analysis of the SwinIR Image Super-Resolution

SwinIR is a recent image restoration method based on the Swin Transformer architecture. In contrast to other traditional convolutional neural networks, SwinIR is capable of capturing sophisticated attention between image patches, leading to remarkable results. In this paper, we focus on the aspect of single-image super-resolution by SwinIR. We discuss the characteristics of the architecture of this algorithm and compare it to other deep learning methods. **This is an MLBriefs article, the source code has not been reviewed!**<br> **The original source code is [[available here|https://github.com/JingyunLiang/SwinIR]] (last checked 2022/12/13).**

GLODAPv2.2022: the latest version of the global interior ocean biogeochemical data product

Abstract. The Global Ocean Data Analysis Project (GLODAP) is a synthesis effort providing regular compilations of surface-to-bottom ocean biogeochemical bottle data, with an emphasis on seawater inorganic carbon chemistry and related variables determined through chemical analysis of seawater samples. GLODAPv2.2022 is an update of the previous version, GLODAPv2.2021 (Lauvset et al., 2021). The major changes are as follows: data from 96 new cruises were added, data coverage was extended until 2021, and for the first time we performed secondary quality control on all sulfur hexafluoride (SF6) data. In addition, a number of changes were made to data included in GLODAPv2.2021. These changes affect specifically the SF6 data, which are now subjected to secondary quality control, and carbon data measured on board the RV Knorr in the Indian Ocean in 1994–1995 which are now adjusted using certified reference material (CRM) measurements made at the time. GLODAPv2.2022 includes measurements from almost 1.4 million water samples from the global oceans collected on 1085 cruises. The data for the now 13 GLODAP core variables (salinity, oxygen, nitrate, silicate, phosphate, dissolved inorganic carbon, total alkalinity, pH, chlorofluorocarbon-11 (CFC-11), CFC-12, CFC-113, CCl4, and SF6) have undergone extensive quality control with a focus on systematic evaluation of bias. The data are available in two formats: (i) as submitted by the data originator but converted to World Ocean Circulation Experiment (WOCE) exchange format and (ii) as a merged data product with adjustments applied to minimize bias. For the present annual update, adjustments for the 96 new cruises were derived by comparing those data with the data from the 989 quality-controlled cruises in the GLODAPv2.2021 data product using crossover analysis. SF6 data from all cruises were evaluated by comparison with CFC-12 data measured on the same cruises. For nutrients and ocean carbon dioxide (CO2) chemistry comparisons to estimates based on empirical algorithms provided additional context for adjustment decisions. The adjustments that we applied are intended to remove potential biases from errors related to measurement, calibration, and data handling practices without removing known or likely time trends or variations in the variables evaluated. The compiled and adjusted data product is believed to be consistent to better than 0.005 in salinity, 1 % in oxygen, 2 % in nitrate, 2 % in silicate, 2 % in phosphate, 4 µmol kg−1 in dissolved inorganic carbon, 4 µmol kg−1 in total alkalinity, 0.01–0.02 in pH (depending on region), and 5 % in the halogenated transient tracers. The other variables included in the compilation, such as isotopic tracers and discrete CO2 fugacity (fCO2), were not subjected to bias comparison or adjustments. The original data, their documentation, and DOI codes are available at the Ocean Carbon and Acidification Data System of NOAA NCEI (https://www.ncei.noaa.gov/access/ocean-carbon-acidification-data-system/oceans/GLODAPv2_2022/, last access: 15 August 2022). This site also provides access to the merged data product, which is provided as a single global file and as four regional ones – the Arctic, Atlantic, Indian, and Pacific oceans – under https://doi.org/10.25921/1f4w-0t92 (Lauvset et al., 2022). These bias-adjusted product files also include significant ancillary and approximated data, which were obtained by interpolation of, or calculation from, measured data. This living data update documents the GLODAPv2.2022 methods and provides a broad overview of the secondary quality control procedures and results.