Carlo Albert
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Carlo Albert has a total of 59 co-authors. Here are the top co-authors by number of shared publications:
- Andreas Scheidegger
EAWAG
17 shared publications - Simone Ulzega
Institute of Computational Life Sciences, Zürich University of Applied Sciences (ZHAW), Wädenswil, Switzerland
15 shared publications - Fabrizio Fenicia
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
10 shared publications - Edward Lavender
Department of Systems Analysis, Integrated Assessment and Modelling Eawag Swiss Federal Institute of Aquatic Science and Technology Dübendorf Switzerland
8 shared publications - Marvin Höge
Swiss Federal Institute for Aquatic Science and Technology (Eawag), Dübendorf, Switzerland
8 shared publications - James Thorburn
School of Applied Sciences Edinburgh Napier University Edinburgh UK
7 shared publications
Latest Publications
- Open Access
- 15 October 2026
- Published by Wiley
How Much Catchment Information Does an LSTM Use for Streamflow Prediction?
In recent years, long short-term memory (LSTM) networks equipped with static catchment attributes have become the standard tool for prediction in gauged and ungauged basins. However, the real effectiveness of static catchment descriptors in guiding prediction in unseen basins has sparked considerable debate. In particular, it remains unclear whether catchment attributes truly allow knowledge transfer to sites where the model has not been trained. Furthermore, it is not clear how much information attributes really contain for streamflow prediction and how much they still miss, whereas recent studies show that this information is mainly related to climate. We address these questions by training an encoder-based LSTM conditioned on the weather. This way, the architecture encourages the encoder to prioritize attribute information complementary to the meteorological forcing variables. Remarkably, our findings reveal that the predictive information of commonly used attributes can be represented by two latent features in most of the catchments. Further, we relate these latent features to the topography and the vegetation of the catchment, bridging the gap between performance and interpretability.
- 15 October 2026
- Published by Wiley
Open AccessHow Much Catchment Information Does an LSTM Use for Streamflow Prediction?
In recent years, long short-term memory (LSTM) networks equipped with static catchment attributes have become the standard tool for prediction in gauged and ungauged basins. However, the real effectiveness of static catchment descriptors in guiding prediction in unseen basins has sparked considerable debate. In particular, it remains unclear whether catchment attributes truly allow knowledge transfer to sites where the model has not been trained. Furthermore, it is not clear how much information attributes really contain for streamflow prediction and how much they still miss, whereas recent studies show that this information is mainly related to climate. We address these questions by training an encoder-based LSTM conditioned on the weather. This way, the architecture encourages the encoder to prioritize attribute information complementary to the meteorological forcing variables. Remarkably, our findings reveal that the predictive information of commonly used attributes can be represented by two latent features in most of the catchments. Further, we relate these latent features to the topography and the vegetation of the catchment, bridging the gap between performance and interpretability.
- Open Access
Big data analysis of animal movements in aquatic ecosystems with acoustic telemetry
Underwater receiver networks (passive acoustic telemetry systems) are deployed to track animals in aquatic habitats all over the world, but remarkably limited attention has been given over to how we can strengthen the value of these networks through statistical and computational advances. Here, we upscale state-of-the-art methods of Bayesian inference to big animal-tracking datasets from acoustic telemetry, with the largest geolocation analysis in a sparse passive acoustic telemetry system (with non-overlapping receivers) to date. Using four years of data from 93 lake trout ( Salvelinus namaycush ) in North America's Lake Champlain (657,360 timesteps per individual), we formulate and fit state-space models to reconstruct animal movements through time. Uniquely, we directly embed biological expertise and detailed complementary datasets from fine-scale positioning systems, accelerometry, swim-tunnel experiments and field range tests in our analysis. Using simulated and real-world datasets, we map movement patterns and estimate residency in distinct management zones. We quantify array precision and deliver maps and residency estimates with a median error and precision (standard error) below 1 %. These results strengthen the evidence base for management. This work takes us a step towards robust inference of movement patterns at scale in acoustic telemetry systems across the world. We can, and should, build on prior scientific progress and extend the value of hard-earned data beyond individual studies to refine inferences for ecology and management.
Open AccessBig data analysis of animal movements in aquatic ecosystems with acoustic telemetry
Underwater receiver networks (passive acoustic telemetry systems) are deployed to track animals in aquatic habitats all over the world, but remarkably limited attention has been given over to how we can strengthen the value of these networks through statistical and computational advances. Here, we upscale state-of-the-art methods of Bayesian inference to big animal-tracking datasets from acoustic telemetry, with the largest geolocation analysis in a sparse passive acoustic telemetry system (with non-overlapping receivers) to date. Using four years of data from 93 lake trout ( Salvelinus namaycush ) in North America's Lake Champlain (657,360 timesteps per individual), we formulate and fit state-space models to reconstruct animal movements through time. Uniquely, we directly embed biological expertise and detailed complementary datasets from fine-scale positioning systems, accelerometry, swim-tunnel experiments and field range tests in our analysis. Using simulated and real-world datasets, we map movement patterns and estimate residency in distinct management zones. We quantify array precision and deliver maps and residency estimates with a median error and precision (standard error) below 1 %. These results strengthen the evidence base for management. This work takes us a step towards robust inference of movement patterns at scale in acoustic telemetry systems across the world. We can, and should, build on prior scientific progress and extend the value of hard-earned data beyond individual studies to refine inferences for ecology and management.
- Open Access
- 28 November 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (48) , eadx0255
Animal tracking with particle algorithms informs protected area design
Animal movements affect their exposure to threats and the efficacy of conservation measures, such as marine protected areas (MPAs). However, many species’ movements are difficult to reconstruct from available datasets, hampering conservation efforts. This is especially the case for aquatic species that rarely surface, for which data are often limited to observations from acoustic telemetry (detections) and ancillary sensors. Here, we pioneer the use of state-of-the-art particle algorithms to model movements, integrate datasets, and assess MPA design, leveraging a case study of a Critically Endangered elasmobranch. Our algorithms led to 5-fold improvements in space-use maps and 30-fold improvements in residency estimates compared to prevailing methods. By integrating tracking datasets, we were uniquely able to examine movements beyond acoustic receivers, MPA-scale residency, and specific habitats beyond protected areas that warrant protection. This work reveals a modeling framework that enhances the conservation value of acoustic telemetry, supporting analyses of MPA efficacy worldwide.
- 28 November 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (48) , eadx0255
Open AccessAnimal tracking with particle algorithms informs protected area design
Animal movements affect their exposure to threats and the efficacy of conservation measures, such as marine protected areas (MPAs). However, many species’ movements are difficult to reconstruct from available datasets, hampering conservation efforts. This is especially the case for aquatic species that rarely surface, for which data are often limited to observations from acoustic telemetry (detections) and ancillary sensors. Here, we pioneer the use of state-of-the-art particle algorithms to model movements, integrate datasets, and assess MPA design, leveraging a case study of a Critically Endangered elasmobranch. Our algorithms led to 5-fold improvements in space-use maps and 30-fold improvements in residency estimates compared to prevailing methods. By integrating tracking datasets, we were uniquely able to examine movements beyond acoustic receivers, MPA-scale residency, and specific habitats beyond protected areas that warrant protection. This work reveals a modeling framework that enhances the conservation value of acoustic telemetry, supporting analyses of MPA efficacy worldwide.
- Open Access
- 6 October 2025
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 992 (1) , 61
Shedding Light on the Solar Dynamo Using Data-driven Bayesian Parameter Inference
We use a zero-dimensional stochastic dynamo model describing the coupled oscillatory behavior of the toroidal and poloidal components of the solar magnetic field. The model includes a time delay, arising from the assumption that the two field components are generated in spatially segregated locations, and a quenching of the α-effect for amplitudes of the toroidal magnetic field outside a range defined by a lower and an upper threshold. We apply two distinct simulation-based Bayesian inference methods to calibrate the model parameters. The first approach employs a Simulated Annealing Approximate Bayesian Computation algorithm, based on the comparison of relevant features extracted from observed data with their counterparts from simulated data. The second one leverages a sequential neural posterior density estimation method known as Automatic Posterior Transformation. The inference was carried out using two different data sets: the observed sunspot record and a recent millennial solar activity reconstruction based on 14C cosmogenic radionuclides from tree rings. The different methods and data sets produce remarkably consistent posterior distributions for the model parameters. Bayesian inference results corroborate the hypothesis presented in a previous publication that the solar dynamo might be operating close to a critical bifurcation point characterized by the coexistence of two modes akin to normal and Grand Minima regimes. We show that the calibrated dynamo model, despite its simplicity, can be used to make predictions about solar cycles.
- 6 October 2025
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 992 (1) , 61
Open AccessShedding Light on the Solar Dynamo Using Data-driven Bayesian Parameter Inference
We use a zero-dimensional stochastic dynamo model describing the coupled oscillatory behavior of the toroidal and poloidal components of the solar magnetic field. The model includes a time delay, arising from the assumption that the two field components are generated in spatially segregated locations, and a quenching of the α-effect for amplitudes of the toroidal magnetic field outside a range defined by a lower and an upper threshold. We apply two distinct simulation-based Bayesian inference methods to calibrate the model parameters. The first approach employs a Simulated Annealing Approximate Bayesian Computation algorithm, based on the comparison of relevant features extracted from observed data with their counterparts from simulated data. The second one leverages a sequential neural posterior density estimation method known as Automatic Posterior Transformation. The inference was carried out using two different data sets: the observed sunspot record and a recent millennial solar activity reconstruction based on 14C cosmogenic radionuclides from tree rings. The different methods and data sets produce remarkably consistent posterior distributions for the model parameters. Bayesian inference results corroborate the hypothesis presented in a previous publication that the solar dynamo might be operating close to a critical bifurcation point characterized by the coexistence of two modes akin to normal and Grand Minima regimes. We show that the calibrated dynamo model, despite its simplicity, can be used to make predictions about solar cycles.
- Open Access
Animal geolocation with convolution algorithms in Julia and R via Wahoo.jl
Animal geolocation is the core of movement ecology. In aquatic ecosystems, electronic tagging and tracking technologies, such as passive acoustic telemetry systems and biologging sensors, are widely deployed. However, statistical estimation of individual locations from these datasets can be challenging and computationally expensive. Here, we introduce Wahoo.jl , a Julia package that fits state-space models to animal-tracking data via convolution algorithms. Wahoo.jl supports passive acoustic telemetry (detection/non-detection) and biologging (i.e., depth) datasets; implements grid-based filtering, smoothing and sampling of trajectories; and exploits GPU acceleration. Using simulations, we illustrate how to use Wahoo.jl from Julia and R to reconstruct movements for an example individual tagged with an acoustic transmitter and an archival depth tag. We also provide validation and sensitivity analyses. Wahoo.jl fills a key gap in the animal-tracking toolbox. The package provides an accessible, flexible and performant interface for an inference methodology that reliably handles multimodal inference problems that challenge other approaches. We discuss the approach’s pros and cons and provide guidance to readers on when to reach for Wahoo.jl .
Open AccessAnimal geolocation with convolution algorithms in Julia and R via Wahoo.jl
Animal geolocation is the core of movement ecology. In aquatic ecosystems, electronic tagging and tracking technologies, such as passive acoustic telemetry systems and biologging sensors, are widely deployed. However, statistical estimation of individual locations from these datasets can be challenging and computationally expensive. Here, we introduce Wahoo.jl , a Julia package that fits state-space models to animal-tracking data via convolution algorithms. Wahoo.jl supports passive acoustic telemetry (detection/non-detection) and biologging (i.e., depth) datasets; implements grid-based filtering, smoothing and sampling of trajectories; and exploits GPU acceleration. Using simulations, we illustrate how to use Wahoo.jl from Julia and R to reconstruct movements for an example individual tagged with an acoustic transmitter and an archival depth tag. We also provide validation and sensitivity analyses. Wahoo.jl fills a key gap in the animal-tracking toolbox. The package provides an accessible, flexible and performant interface for an inference methodology that reliably handles multimodal inference problems that challenge other approaches. We discuss the approach’s pros and cons and provide guidance to readers on when to reach for Wahoo.jl .
- Open Access
- 3 April 2025
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 16 (8) , 1609-1616
patter: Particle algorithms for animal tracking in R and Julia
State‐space models are a powerful modelling framework in movement ecology that represents individual movements and the processes connecting movements to observations. However, fitting state‐space models to animal‐tracking data can be difficult and computationally expensive. Here, we introduce patter, a package that provides particle filtering and smoothing algorithms that fit Bayesian state‐space models to tracking data, with a focus on data from aquatic animals in receiver arrays. patter is written in R, with a performant Julia backend. Package functionality supports data simulation, preparation, filtering, smoothing and mapping. In two examples, we demonstrate how to implement patter to reconstruct the movements of a tagged animal in an acoustic telemetry system from acoustic detections and ancillary observations. With perfect information, the particle filter reconstructs the true (unobserved) movement path (Example One). More generally, particle algorithms represent an individual's possible location probabilistically as a weighted series of samples (‘particles’). In our illustration, we resolve an individual's (unobserved) location every 2 min during 1 month and use particles to visualise movements, map space use and quantify residency (Example Two). patter facilitates robust, flexible and efficient analyses of animal‐tracking data. The methods are widely applicable and enable refined analyses of space use, home ranges and residency.
- 3 April 2025
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 16 (8) , 1609-1616
Open Accesspatter: Particle algorithms for animal tracking in R and Julia
State‐space models are a powerful modelling framework in movement ecology that represents individual movements and the processes connecting movements to observations. However, fitting state‐space models to animal‐tracking data can be difficult and computationally expensive. Here, we introduce patter, a package that provides particle filtering and smoothing algorithms that fit Bayesian state‐space models to tracking data, with a focus on data from aquatic animals in receiver arrays. patter is written in R, with a performant Julia backend. Package functionality supports data simulation, preparation, filtering, smoothing and mapping. In two examples, we demonstrate how to implement patter to reconstruct the movements of a tagged animal in an acoustic telemetry system from acoustic detections and ancillary observations. With perfect information, the particle filter reconstructs the true (unobserved) movement path (Example One). More generally, particle algorithms represent an individual's possible location probabilistically as a weighted series of samples (‘particles’). In our illustration, we resolve an individual's (unobserved) location every 2 min during 1 month and use particles to visualise movements, map space use and quantify residency (Example Two). patter facilitates robust, flexible and efficient analyses of animal‐tracking data. The methods are widely applicable and enable refined analyses of space use, home ranges and residency.
- Open Access
- 3 April 2025
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 16 (8) , 1808-1819
Particle algorithms for animal movement modelling in receiver arrays
Particle filters and smoothers are sequential Monte Carlo algorithms used to fit non‐linear, non‐Gaussian state‐space models. These algorithms are well placed to fit process‐oriented models to animal‐tracking data, especially in receiver arrays, but to date they have received limited attention in the ecological literature. We introduce a Bayesian filtering–smoothing algorithm that reconstructs individual movements and patterns of space use from animal‐tracking data, with a focus on passive acoustic telemetry systems. Within a sound probabilistic framework, the methodology integrates the movement process and the observation processes of disparate datasets, while correctly representing uncertainty. In a simulation‐based analysis, we compare the performance of our algorithm to the prevailing heuristic methods used to study movements and space use in passive acoustic telemetry systems and analyse algorithm sensitivity. We find the particle smoothing methodology outperforms heuristic methods across the board. Particle‐based maps represent simulated movements more accurately, even in dense receiver arrays, and are better suited to analyses of home ranges, residency and habitat preferences. This study sets a new state‐of‐the‐art for movement modelling in receiver arrays. Particle algorithms provide a robust, flexible and intuitive modelling framework with potential applications in many ecological settings.
- 3 April 2025
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 16 (8) , 1808-1819
Open AccessParticle algorithms for animal movement modelling in receiver arrays
Particle filters and smoothers are sequential Monte Carlo algorithms used to fit non‐linear, non‐Gaussian state‐space models. These algorithms are well placed to fit process‐oriented models to animal‐tracking data, especially in receiver arrays, but to date they have received limited attention in the ecological literature. We introduce a Bayesian filtering–smoothing algorithm that reconstructs individual movements and patterns of space use from animal‐tracking data, with a focus on passive acoustic telemetry systems. Within a sound probabilistic framework, the methodology integrates the movement process and the observation processes of disparate datasets, while correctly representing uncertainty. In a simulation‐based analysis, we compare the performance of our algorithm to the prevailing heuristic methods used to study movements and space use in passive acoustic telemetry systems and analyse algorithm sensitivity. We find the particle smoothing methodology outperforms heuristic methods across the board. Particle‐based maps represent simulated movements more accurately, even in dense receiver arrays, and are better suited to analyses of home ranges, residency and habitat preferences. This study sets a new state‐of‐the‐art for movement modelling in receiver arrays. Particle algorithms provide a robust, flexible and intuitive modelling framework with potential applications in many ecological settings.
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Leveraging Machine Learning to Uncover and Interpret Relevant Catchment Features
Recent advances in catchment hydrology [Kratzert et al., 2019–2021] demonstrate the superiority of LSTMs over traditional conceptual models for streamflow prediction in large-sample datasets. LSTMs achieve better streamflow accuracies by leveraging information from diverse hydrological behaviors. These models are enriched with static catchment attributes, which, when combined with meteorological drivers, play a critical role in streamflow formation. Augmenting LSTMs with these attributes further enhances their performance compared to vanilla LSTMs, underscoring the importance of these attributes for accurate streamflow predictions. Building on this, a recent study [Bassi et al., 2024] employed a conditional autoencoder to reveal that most of the relevant catchment information for streamflow prediction can be distilled into two features, with a third feature being beneficial for particularly challenging catchments. In this work, we directly derive a minimal set of catchment features from known attributes by passing them through the encoder and subsequently comparing streamflow predictions against state-of-the-art benchmarks [Kratzert et al., 2021]. Our findings indicate that while the intrinsic dimension of 26 commonly used attributes is four, only two features suffice for accurate streamflow prediction. This aligns closely with the findings of Bassi et al. (2024), suggesting that nearly all relevant information for streamflow prediction is encapsulated in known attributes. Finally, we provide an interpretation of these two machine-learning-derived features using information theory techniques.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessLeveraging Machine Learning to Uncover and Interpret Relevant Catchment Features
Recent advances in catchment hydrology [Kratzert et al., 2019–2021] demonstrate the superiority of LSTMs over traditional conceptual models for streamflow prediction in large-sample datasets. LSTMs achieve better streamflow accuracies by leveraging information from diverse hydrological behaviors. These models are enriched with static catchment attributes, which, when combined with meteorological drivers, play a critical role in streamflow formation. Augmenting LSTMs with these attributes further enhances their performance compared to vanilla LSTMs, underscoring the importance of these attributes for accurate streamflow predictions. Building on this, a recent study [Bassi et al., 2024] employed a conditional autoencoder to reveal that most of the relevant catchment information for streamflow prediction can be distilled into two features, with a third feature being beneficial for particularly challenging catchments. In this work, we directly derive a minimal set of catchment features from known attributes by passing them through the encoder and subsequently comparing streamflow predictions against state-of-the-art benchmarks [Kratzert et al., 2021]. Our findings indicate that while the intrinsic dimension of 26 commonly used attributes is four, only two features suffice for accurate streamflow prediction. This aligns closely with the findings of Bassi et al. (2024), suggesting that nearly all relevant information for streamflow prediction is encapsulated in known attributes. Finally, we provide an interpretation of these two machine-learning-derived features using information theory techniques.
- Open Access
Animal tracking with particle algorithms for conservation
The movements of aquatic animals affect their exposure to threats and the efficacy of conservation measures, such as Marine Protected Areas (MPAs). However, many species’ movements remain poorly understood and difficult to reconstruct from available datasets, hampering conservation efforts. This is especially the case for species that rarely surface, for which data are often limited to observations from acoustic telemetry (detections) and ancillary sensors, such as archival tags. Here, we pioneer the use of state-of-the-art particle algorithms to model animal movement, integrate datasets and assess MPA design, using a case study of the Critically Endangered flapper skate ( Dipturus intermedius ) in Scotland. Our algorithms led to 5-fold improvements in maps of space use and 30-fold improvements in residency estimates (lower mean error) compared to prevailing heuristic methods. By formally integrating tracking datasets, we were uniquely able to examine movements beyond receivers into fished zones, MPA-scale residency and specific habitats beyond protected areas that may warrant protection. This work showcases a probabilistically sound modelling framework that is sufficiently fast, flexible and accessible to meet the demands of modern animal-tracking datasets in acoustic telemetry systems. This represents a marked advance for analyses of animal movements and MPA efficacy worldwide.
Open AccessAnimal tracking with particle algorithms for conservation
The movements of aquatic animals affect their exposure to threats and the efficacy of conservation measures, such as Marine Protected Areas (MPAs). However, many species’ movements remain poorly understood and difficult to reconstruct from available datasets, hampering conservation efforts. This is especially the case for species that rarely surface, for which data are often limited to observations from acoustic telemetry (detections) and ancillary sensors, such as archival tags. Here, we pioneer the use of state-of-the-art particle algorithms to model animal movement, integrate datasets and assess MPA design, using a case study of the Critically Endangered flapper skate ( Dipturus intermedius ) in Scotland. Our algorithms led to 5-fold improvements in maps of space use and 30-fold improvements in residency estimates (lower mean error) compared to prevailing heuristic methods. By formally integrating tracking datasets, we were uniquely able to examine movements beyond receivers into fished zones, MPA-scale residency and specific habitats beyond protected areas that may warrant protection. This work showcases a probabilistically sound modelling framework that is sufficiently fast, flexible and accessible to meet the demands of modern animal-tracking datasets in acoustic telemetry systems. This represents a marked advance for analyses of animal movements and MPA efficacy worldwide.
- Open Access
- 20 January 2025
- Published by Qeios Ltd
Review of: "Active Sequential Posterior Estimation for Sample-Efficient Simulation-Based Inference"
This manuscript introduces Active Learning to the APT-variant of Simulation-based Inference. The proposed method iteratively selects parameter sets with the highest expected information gain in terms of the variance of the posterior density w.r.t. th...
- 20 January 2025
- Published by Qeios Ltd
Open AccessReview of: "Active Sequential Posterior Estimation for Sample-Efficient Simulation-Based Inference"
This manuscript introduces Active Learning to the APT-variant of Simulation-based Inference. The proposed method iteratively selects parameter sets with the highest expected information gain in terms of the variance of the posterior density w.r.t. th...