Publications of Alexander Barth
Bookmark and Share    
Full Text
See detailClimatological distribution of dissolved inorganic nutrients in the western Mediterranean Sea (1981–2017)
Belgacem, Malek; Schroeder, Katrin; Barth, Alexander ULiege et al

in Earth System Science Data (2021), 13

Detailed reference viewed: 27 (1 ULiège)
Full Text
See detailA New Global Ocean Climatology
Shahzadi, Kanwal; Pinardi, Nadia; Barth, Alexander ULiege et al

in Frontiers in Marine Science (2021), 9

A new global ocean temperature and salinity climatology is proposed for two time periods: a long time mean using multiple sensor data for the 1900–2017 period and a shorter time mean using only profiling ... [more ▼]

A new global ocean temperature and salinity climatology is proposed for two time periods: a long time mean using multiple sensor data for the 1900–2017 period and a shorter time mean using only profiling float data for the 2003–2017 period. We use the historical database of World Ocean Database 2018. The estimation approach is novel as an additional quality control procedure is implemented, along with a new mapping algorithm based on Data Interpolating Variational Analysis. The new procedure, in addition to the traditional quality control approach, resulted in low sensitivity in terms of the first guess field choice. The roughness index and the root mean square of residuals are new indices applied to the selection of the free mapping parameters along with sensitivity experiments. Overall, the new estimates were consistent with previous climatologies, but several differences were found. The cause of these discrepancies is difficult to identify due to several differences in the procedures. To minimise these uncertainties, a multi-model ensemble mean is proposed as the least uncertain estimate of the global ocean temperature and salinity climatology. [less ▲]

Detailed reference viewed: 32 (3 ULiège)
Full Text
See detailSeaDataCloud Data Products for the European marginal seas and the Global Ocean
Simoncelli, Simona; Coatanoan, Christine; Myroshnychenko, Volodymyr et al

Conference (2021, May)

Data products, based on in situ temperature and salinity observations from SeaDataNet infrastructure, have been released within the framework of SeaDataCloud (SDC) project. The data from different data ... [more ▼]

Data products, based on in situ temperature and salinity observations from SeaDataNet infrastructure, have been released within the framework of SeaDataCloud (SDC) project. The data from different data providers are integrated and harmonized thanks to standardized quality assurance and quality control methodologies conducted at various stages of the data value chain. The data ingested within SeaDataNet are earlier validated by data providers who assign corresponding quality flags, but a Quality Assurance Strategy has been implemented and progressively refined to guarantee the consistency of the database content and high quality derived products. Two versions of aggregated datasets for the European marginal seas have been published and used to compute regional high resolution climatologies. External datasets, the World Ocean Database from NOAA and the CORA dataset from the Copernicus Marine Service in situ Thematic Assembly Center, have been integrated with SDC data collections to maximize data coverage and minimize the mapping error. The products are available through the SDC catalogue accompanied by Product Information Documents containing the specifications about product’s generation, characteristics and usability. Digital Object Identifiers are assigned to products and relative documentation to foster transparency of the production chain, acknowledging all actors involved from data providers to information producers. [less ▲]

Detailed reference viewed: 42 (1 ULiège)
Full Text
See detailImplementation and Validating of the Regional Ocean Model System (ROMS) for the Sunda Strait connecting the Java Sea to the Indian Ocean
Subekti, Mujiasih ULiege; Beckers, Jean-Marie ULiege; Barth, Alexander ULiege

Conference (2021, April 30)

Regional Ocean Model System (ROMS) has been simulated for the Sunda Strait, the Java Sea, and the Indian Ocean. The simulation was undertaken for thirteen months of data period (August 2013 – August 2014 ... [more ▼]

Regional Ocean Model System (ROMS) has been simulated for the Sunda Strait, the Java Sea, and the Indian Ocean. The simulation was undertaken for thirteen months of data period (August 2013 – August 2014). However, we only used four months period for validation, namely September – December 2013. The input data involved the HYbrid Coordinate Ocean Model (HYCOM) ocean model output by considering atmospheric forcing from the European Centre for Medium-Range Weather Forecasts (ECMWF), without and with tides forcing from TPXO and rivers. The output included vertical profile temperature and salinity, sea surface temperature (SST), seas surface height (SSH), zonal (u), and meridional (v) velocity. We compared the model SST to satellite SST in time series, SSH to tides gauges data in time series, the model u and v component velocity to High Frequency (HF) radial velocity. The vertical profile temperature and salinity were compared to Argo float data and XBT. Besides, we validated the amplitude and phase of the ROMS seas surface height to amplitude and phase of the tides-gauges, including four constituents (M2, S2, K1, O1). [less ▲]

Detailed reference viewed: 76 (11 ULiège)
Full Text
See detailDetection of shadows in high spatial resolution ocean satellite data using DINEOF
Alvera Azcarate, Aida ULiege; Van der Zande, Dimitry; Barth, Alexander ULiege et al

in Remote Sensing of Environment (2021), 253

Cloud shadows present in high spatial resolution remote sensing datasets can affect the quality of the data if they are not properly detected and removed. When working with ocean data, cloud shadows are ... [more ▼]

Cloud shadows present in high spatial resolution remote sensing datasets can affect the quality of the data if they are not properly detected and removed. When working with ocean data, cloud shadows are often difficult to differentiate from non-shadow values, since they show similar spectral characteristics than water pixels. A methodology to detect cloud shadows over the ocean is proposed. The present approach combines a series of tests applied directly to the physical variables derived from the satellite measured radiances, and it therefore does not depend on the wavebands measured by a specific satellite sensor. The tests include a departure from an EOF basis calculated using DINEOF, a threshold test, a proximity to cloud test and a ray tracing test. The weighing of the different tests can be adapted to each case or domain of study. The results are compared to manually detected shadows and to another shadow detection method. The approach works with cloud shadows of all sizes, and also with very small objects shadows, like the shadows projected by offshore windmills. [less ▲]

Detailed reference viewed: 74 (8 ULiège)
Full Text
See detailVariational interpolation of high-frequency radar surface currents using DIVAnd
Barth, Alexander ULiege; Troupin, Charles ULiege; Emma, Reyes et al

in Ocean Dynamics (2021), 71

DIVAnd (Data-Interpolating Variational Analysis, in n-dimensions) is a tool to interpolate observations on a regular grid using the variational inverse method. We have extended DIVAnd to include ... [more ▼]

DIVAnd (Data-Interpolating Variational Analysis, in n-dimensions) is a tool to interpolate observations on a regular grid using the variational inverse method. We have extended DIVAnd to include additional dynamic constraints relevant to surface currents, including imposing a zero normal velocity at the coastline, imposing a low horizontal divergence of the surface currents, temporal coherence and simplified dynamics based on the Coriolis force and the possibility of including a surface pressure gradient. The impact of these constraints is evaluated by cross-validation using the HF (High-Frequency) radar surface current observations in the Ibiza Channel from the Balearic Islands Coastal Ocean Observing and Forecasting System (SOCIB). A small fraction of the radial current observations are set aside to validate the velocity reconstruction. The remaining radial currents from the two radar sites are combined to derive total surface currents using DIVAnd and then compared to the cross-validation data set and to drifter observations. The benefit of the dynamic constraints is shown relative to a variational interpolation without these dynamical constraints. The best results were obtained using the Coriolis force and the surface pressure gradient as a constraint which are able to improve the reconstruction from the Open-boundary Modal Analysis, a quite commonly used method to interpolate HF radar observations, once multiple time instances are considered together. [less ▲]

Detailed reference viewed: 151 (12 ULiège)
Full Text
See detailCreation of high resolution suspended particulate matter data in the North Sea from Sentinel-2 and Sentinel-3 data.
Alvera Azcarate, Aida ULiege; Barth, Alexander ULiege; Troupin, Charles ULiege et al

in 2021 IGARSS: IEEE International Geoscience & Remote Sensing Symposium. Proceedings : July 12-16, 2021, Brussels (Belgium) (2021)

A combined Sentinel-2 and Sentinel-3 Suspended Particulate Matter reconstruction is performed using an Empirical Orthogonal Function technique, called DINEOF (Data Interpolating Empirical Orthogonal ... [more ▼]

A combined Sentinel-2 and Sentinel-3 Suspended Particulate Matter reconstruction is performed using an Empirical Orthogonal Function technique, called DINEOF (Data Interpolating Empirical Orthogonal Functions). The combination of these two datasets allows us to retain both the high spatial resolution of the Sentinel-2 data while increasing the temporal resolution thanks to the addition of Sentinel-3 data on days when no Sentinel-2 data are available. Results show an increased variability on the reconstruction of Sentinel-3 data, and a low error of the overall reconstruction. [less ▲]

Detailed reference viewed: 41 (3 ULiège)
Full Text
See detailAnalysis of 23 Years of Daily Cloud-Free Chlorophyll and Suspended Particulate Matter in the Greater North Sea
Alvera Azcarate, Aida ULiege; Van der Zande, Dimitry; Barth, Alexander ULiege et al

in Frontiers in Marine Science (2021)

Satellite-derived estimates of ocean color variables are available for several decades now and allow performing studies of the long-term changes occurred in an ecosystem. A daily, gap-free analysis of ... [more ▼]

Satellite-derived estimates of ocean color variables are available for several decades now and allow performing studies of the long-term changes occurred in an ecosystem. A daily, gap-free analysis of chlorophyll (CHL) and suspended particulate matter (SPM, indicative of light availability in the subsurface) at 1 km resolution over the Greater North Sea during the period 1998–2020 is presented. Interannual changes are described, with maximum average CHL values increasing during the period 1998–2008, a slightly decreasing trend in 2009–2017 and an stagnation in recent years. The typical spring bloom is observed to happen earlier each year, with about 1 month difference between 1998 and 2020. The duration of the bloom (time between onset and offset) appears also to be increasing with time, but the average CHL value during the spring bloom does not show a clear trend. The causes for earlier spring blooms are still unclear, although a rising water temperature can partially explain them through enhanced phytoplankton cell division rates or through increased water column stratification. SPM values during winter months (prior to the development of the spring bloom) do not exhibit a clear trend over the same period, although slightly higher SPM values are observed in recent years. The influence of sea surface temperature in the spring bloom timing appears to be dominant over the influence of SPM concentration, according to our results. The number of satellites available over the years for producing CHL and SPM in this work has an influence in the total amount of available data before interpolation. The amount of missing data has an influence in the total variability that is retained in the final dataset, and our results suggest that at least three satellites would be needed for a good representation of ocean color variability. [less ▲]

Detailed reference viewed: 36 (2 ULiège)
Full Text
See detailReconstruction of missing data in satellite images of the Southern North Sea using a convolutional neural network (DINCAE)
Barth, Alexander ULiege; Alvera Azcarate, Aida ULiege; Troupin, Charles ULiege et al

in 2021 IGARSS: IEEE International Geoscience & Remote Sensing Symposium. Proceedings : July 12-16, 2021, Brussels (Belgium) (2021)

A neural network with the architecture of a convolutional autoencoder is used to reconstruct missing data in satellite images of the Southern North Sea. The technique is applied to a multi-satellite data ... [more ▼]

A neural network with the architecture of a convolutional autoencoder is used to reconstruct missing data in satellite images of the Southern North Sea. The technique is applied to a multi-satellite data product of chlorophyll-a and total suspended particulate matter (SPM) concentration (representing 20 years of data). The presence of clouds significantly reduces the extent of the ocean that can be measured by satellite sensors using the visible or infrared spectrum. The accuracy of the reconstruction is assessed using cross-validation (i.e. increasing the actual extent of the cloud coverage). The results of the neural network compare favourably the data withheld for cross-validation. [less ▲]

Detailed reference viewed: 62 (3 ULiège)
Full Text
See detailHydrodynamic variability in the Southern Bight of the North Sea in response to typical atmospheric and tidal regimes. Benefit of using a high resolution model
Ivanov, Evgeny ULiege; Capet, Arthur ULiege; Barth, Alexander ULiege et al

in Ocean Modelling (2020), 154

In this paper, the hydrodynamics of the Southern Bight of the North Sea (SBNS) and in particular, the Belgian Coastal Zone (BCZ) is investigated on daily to seasonal time scales using a high resolution ... [more ▼]

In this paper, the hydrodynamics of the Southern Bight of the North Sea (SBNS) and in particular, the Belgian Coastal Zone (BCZ) is investigated on daily to seasonal time scales using a high resolution hydrodynamical model. The Regional Ocean Modeling System (ROMS) is implemented over the SBNS with 5 km resolution and downscaled at 1 km resolution over the BCZ in a two-way nesting configuration run over a three years period (i.e. 2006–2008). The benefit of using a high resolution model over the BCZ is assessed through an extensive comparison of model results with data from satellite and in-situ fixed platforms as well as reference products available for the region. The validation exercise and the results analysis are conducted with a particular focus on hydrodynamic features that are expected to impact the sediment transport. We find that despite the validation procedure does not allow to clearly demonstrate better performance of the high resolution model compared to the coarse resolution model in terms of overtidal circulation, sea surface temperature (SST) and salinity (SSS), the high resolution model resolves additional details in the variability of residual circulation and Scheldt salinity plume dynamics. The analysis of the response of the simulated hydrodynamics to atmospheric regimes for neap and spring tide highlights the major role played by the wind direction on the averaged currents and plume extension. The strongest currents and minimum plume extension are obtained under southwestern winds and neap tide while when northeastern winds prevail, the plume extension is at its maximum and the circulation is the weakest. We show that while neap tides allow the establishment of streamlined circulation, the spring tides induce more turbulent circulation which can favor the retention of transported elements. This latter property could not be resolved with the 5 km resolution model. [less ▲]

Detailed reference viewed: 64 (10 ULiège)
See detailMultiplatform analysis of a large anticyclonic eddy in the Algero-Provencal basin in 2019
Alvera Azcarate, Aida ULiege; Barth, Alexander ULiege; Troupin, Charles ULiege et al

Conference (2020, May)

A large anticyclonic eddy formed in April 2019 in the Algero-Provencal basin between Mallorca and Sardinia, and lasted until November 2019. While mesoscale activity is usually high in this part of the ... [more ▼]

A large anticyclonic eddy formed in April 2019 in the Algero-Provencal basin between Mallorca and Sardinia, and lasted until November 2019. While mesoscale activity is usually high in this part of the Mediterranean basin, the formation of such large (about 150 km in diameter) and long-lived eddies is not common. The eddy formed from a filament originated in the Algerian coast and was visible in multiple sources of satellite data, including sea surface temperature and ocean colour from Sentinel-3, until summer. Because of the warming of the surface layer, during summer months the eddy remained as a subsurface structure, evidenced by the sea level anomaly derived from altimetry data. A surface signal developed again in November, and the eddy finally dissipated in December 2019. According to CMEMS model data, in its strongest period the eddy reached about 300 m in depth, and during its sub-surface period the center was located at about 100 m depth. While at the surface the temperature signal was very clear, model data suggest the salinity anomaly was stronger than temperature, especially at depth. Such large and long-lived eddies have an impact in the basin currents, specifically in the transport of cold water from the northern to the southern part of the western Mediterranean basin, influencing the ecosystem there. The impact of the presence of this eddy, its long duration and the additional mesoscale and submesoscale activity that originated in its surroundings are investigated using a combination of remote sensing data, in situ data and model data. [less ▲]

Detailed reference viewed: 12 (1 ULiège)
Full Text
See detailDINCAE: multivariate convolutional neural network with error estimates to reconstruct gridded and along-track satellite observations
Barth, Alexander ULiege; Alvera Azcarate, Aida ULiege; Troupin, Charles ULiege et al

Conference (2020)

DINCAE (Data INterpolating Convolutional Auto-Encoder) is a neural network to reconstruct missing data (e.g. obscured by clouds) in satellite data. The technique has been described in Barth et al. (2020 ... [more ▼]

DINCAE (Data INterpolating Convolutional Auto-Encoder) is a neural network to reconstruct missing data (e.g. obscured by clouds) in satellite data. The technique has been described in Barth et al. (2020, https://doi.org/10.5194/gmd-13-1609-2020) for a single variable. Contrary to standard image reconstruction with neural networks, this application requires a method to handle missing data (or data with variable accuracy) in the training phase. Instead of using a standard L2 (or L1) cost function, the neural network (U-net type of network) is optimized by minimizing the negative log likelihood assuming a Gaussian distribution (characterized by a mean and a variance). As a consequence, the neural network also provides an expected error variance of the reconstructed field (per pixel and per time instance).The method has been extended to handle multivariate data (an example will be shown with sea-surface temperature, chlorophyll and wind fields) and the structure of the neural network has been updated. The improvement of this network is demonstrated in the Adriatic Sea. The code has been ported from Python TensorFlow 1.15 to Julia with Knet.jl which reduces the training time from 3.5 to 1.9 hours for the same network architecture. The speed-up is primarily thanks to a more efficient data transformation which is used to expand the training dataset by data augmentation. The first convolutional layers and the cost function have been modified so that also unstructured data can be used as inputs to obtain gridded fields as output. To demonstrate this, the neural network is applied to along-track altimetry data in the Mediterranean Sea. Results from a 20-year reconstruction are presented and validated. Hyperparameters are determined using Bayesian optimization and minimizing the error relative to a development dataset. [less ▲]

Detailed reference viewed: 25 (2 ULiège)
Full Text
See detailDINCAE 1.0: a convolutional neural network with error estimates to reconstruct sea surface temperature satellite observations
Barth, Alexander ULiege; Alvera Azcarate, Aida ULiege; Licer, Matjaz et al

in Geoscientific Model Development (2020), 13(3), 1609--1622

A method to reconstruct missing data in sea surface temperature data using a neural network is presented. Satellite observations working in the optical and infrared bands are affected by clouds, which ... [more ▼]

A method to reconstruct missing data in sea surface temperature data using a neural network is presented. Satellite observations working in the optical and infrared bands are affected by clouds, which obscure part of the ocean underneath. In this paper, a neural network with the structure of a convolutional auto-encoder is developed to reconstruct the missing data based on the available cloud-free pixels in satellite images. Contrary to standard image reconstruction with neural networks, this application requires a method to handle missing data (or data with variable accuracy) in the training phase. The present work shows a consistent approach which uses the satellite data and its expected error variance as input and provides the reconstructed field along with its expected error variance as output. The neural network is trained by maximizing the likelihood of the observed value. The approach, called DINCAE (Data INterpolating Convolutional Auto-Encoder), is applied to a 25-year time series of Advanced Very High Resolution Radiometer (AVHRR) sea surface temperature data and compared to DINEOF (Data INterpolating Empirical Orthogonal Functions), a commonly used method to reconstruct missing data based on an EOF (empirical orthogonal function) decomposition. The reconstruction error of both approaches is computed using cross-validation and in situ observations from the World Ocean Database. DINCAE results have lower error while showing higher variability than the DINEOF reconstruction. [less ▲]

Detailed reference viewed: 197 (10 ULiège)
Full Text
See detailA convolutional neural network with error estimates to reconstruct sea surface temperature satellite observations (DINCAE)
Barth, Alexander ULiege; Alvera Azcarate, Aida ULiege; Licer, Matjaz et al

Conference (2020)

A method to reconstruct missing data in satellite data using a neural network is presented. Satellite observations working in the optical and infrared bands are affected by clouds, which obscure part of ... [more ▼]

A method to reconstruct missing data in satellite data using a neural network is presented. Satellite observations working in the optical and infrared bands are affected by clouds, which obscure part of the ocean underneath. In this paper, a neural network with the structure of a convolutional auto-encoder is developed to reconstruct the missing data based on the available cloud-free pixels in satellite images. However, it is unclear how to handle missing data (or data with variable accuracy) in a neural network when using incomplete satellite data in the training phase. The present work shows a consistent approach which uses essentially the satellite data and its expected error variance as input and provides the reconstructed field along with its expected error variance as output. The approach is motivated by the way models and observations are combined in the frame of data assimilation. The neural network is trained by maximizing the likelihood of the observed value. The corresponding error variances are estimated during training and do not need to be known a priori. The approach, called DINCAE (Data-Interpolating Convolutional Auto-Encoder) is applied to a relatively long time-series of Advanced Very High Resolution Radiometer (AVHRR) sea surface temperature data and compared to DINEOF (Data Interpolating Empirical Orthogonal Functions), a method to reconstruct missing data based on an EOF decomposition. The reconstruction error of both approaches is computed using cross-validation and in situ observations from the World Ocean Database. DINCAE results have lower error, while showing higher variability than the DINEOF reconstruction. The resulting error estimates are also validated using the cross-validation data and they follow closely the expected Gaussian distribution. [less ▲]

Detailed reference viewed: 27 (5 ULiège)
Full Text
See detailD10.8: DIVA online operational in VRE
Troupin, Charles ULiege; Barth, Alexander ULiege; Beckers, Jean-Marie ULiege

Conference (2019, October 15)

Detailed reference viewed: 34 (4 ULiège)
Full Text
See detailPlankton Data Products
Barth, Alexander ULiege; Beauchard, Olivier; Herman, Peter et al

Conference (2019, May 15)

Detailed reference viewed: 29 (6 ULiège)
Full Text
See detailOnline DIVAnd Recent progress & hurdles
Barth, Alexander ULiege; Beckers, Jean-Marie ULiege; Troupin, Charles ULiege

Conference (2019, May 10)

Detailed reference viewed: 41 (4 ULiège)
Full Text
See detailDIVA & DIVAnd interpolation tools: All you need to know about them
Troupin, Charles ULiege; Barth, Alexander ULiege; Watelet, Sylvain ULiege et al

Conference (2019, April 11)

Detailed reference viewed: 56 (5 ULiège)