Wildfire prediction has become increasingly crucial due to the escalating impacts of climate change. Traditional CNN-based wildfire prediction models struggle with handling missing oceanic data and addressing the long-range dependencies across distant regions in meteorological data. In this paper, we introduce an innovative Graph Neural Network (GNN)-based model for global wildfire prediction. We propose a hybrid model that combines the spatial prowess of Graph Convolutional Networks (GCNs) with the temporal depth of Long Short-Term Memory (LSTM) networks. Our approach uniquely transforms global climate and wildfire data into a graph representation, addressing challenges such as null oceanic data locations and long-range dependencies inherent in traditional models. Benchmarking against established architectures using an unseen ensemble of JULES-INFERNO simulations, our model demonstrates superior predictive accuracy. Furthermore, we emphasise the model's explainability, unveiling potential wildfire correlation clusters through community detection and elucidating feature importance via Integrated Gradient analysis. Our findings not only advance the methodological domain of wildfire prediction but also underscore the importance of model transparency, offering valuable insights for stakeholders in wildfire management.

The Koopman operator provides a powerful framework for modeling dynamical systems and has attracted growing interest from the machine learning community. However, its infinite-dimensional nature makes identifying suitable finite-dimensional subspaces challenging, especially for deep architectures. We argue that these difficulties come from suboptimal representation learning, where latent variables fail to balance expressivity and simplicity. This tension is closely related to the information bottleneck (IB) dilemma: constructing compressed representations that are both compact and predictive. Rethinking Koopman learning through this lens, we demonstrate that latent mutual information promotes simplicity, yet an overemphasis on simplicity may cause latent space to collapse onto a few dominant modes. In contrast, expressiveness is sustained by the von Neumann entropy, which prevents such collapse and encourages mode diversity. This insight leads us to propose an information-theoretic Lagrangian formulation that explicitly balances this tradeoff. Furthermore, we propose a new algorithm based on the Lagrangian formulation that encourages both simplicity and expressiveness, leading to a stable and interpretable Koopman representation. Beyond quantitative evaluations, we further visualize the learned manifolds under our representations, observing empirical results consistent with our theoretical predictions. Finally, we validate our approach across a diverse range of dynamical systems, demonstrating improved performance over existing Koopman learning methods.

Real-time rooftop wind-speed distribution is essential for safe operation of drones or urban air mobilities, wind control systems, and rooftop utilization. However, rooftop flows exhibit strong nonlinearity, separation, and cross-direction variability, which challenge flow field reconstruction from sparse sensors. This study develops a learning-from-observation framework using wind-tunnel experiment data obtained by Particle Image Velocimetry (PIV) and benchmarks Kriging interpolation against three deep learning models-UNet, Vision Transformer Autoencoder (ViTAE), and Conditional Wasserstein GAN (CWGAN). We evaluate two training strategies-single wind-direction training (SDT) and mixed wind-direction training (MDT)-across sensor densities of 5-30, test robustness under +/- 1-grid sensor position perturbations, and optimize sensor placement via Proper Orthogonal Decomposition with QR decomposition. Results indicate the potential of deep learning methods to reconstruct rooftop wind fields from sparse sensor data. Compared with traditional Kriging interpolation, the deep learning models achieved improvements of up to 32.7% in SSIM, 24.2% in FAC2, and 27.8% in NMSE. In addition, training with mixed wind directions was found to be essential for achieving better model performance, resulting in further improvements of up to 173.7% in SSIM, 16.7% in FAC2, and 98.3% in MG compared with single-direction train ing. Furthermore, the comparison among methods indicates that sensor configuration, optimization, and training strategy should be jointly considered to ensure reliable deployment. For example, QR-based optimization en hances robustness by up to 27.8% under sensor perturbations, although with metric-dependent performance trade-offs. By training on experimental rather than simulated data, the proposed framework also offers practical guidance for method selection and sensor placement across different scenarios. The source code for this project is publicly available as open source on Github (https://github.com/Yng314/windreconstruction).

Full-field reconstruction of air pollution is essential for evaluating pollution exposure and supporting public health decision-making. However, the complex interactions among pollutants, hard-to-predict weather patterns, and limited monitoring station coverage make this a complex task. We apply deep learning techniques to provide fast and accurate reconstructions from sparse observations of four key pollutants: NO2, O3, PM2.5 and PM10. Models are trained on full-field simulation data and evaluated on real-world observations collected from 9 to 28 monitoring stations in the city of Paris. We introduce a diffusion-based generative framework for multi-pollutant reconstruction and benchmark its performance against deterministic deep learning models. Despite noisy observations and strong spatial variability, the models achieve high structural similarity on simulated validation data and produce realistic spatial patterns on real-world observations, as indicated by power-spectrum analysis. We introduce data augmentation methods that enable transfer to real-world observations without retraining, allowing the models to generalise beyond the training period. These findings highlight the potential of ML models for reliable real-world deployment in air pollution reconstruction tasks.

Computer models are indispensable tools for understanding the Earth system. While high-resolution operational models have achieved many successes, they exhibit persistent biases, particularly in simulating extreme events and statistical distributions. In contrast, coarse-grained idealized models isolate fundamental processes and can be precisely calibrated to excel in characterizing specific dynamical and statistical features. However, different models remain siloed by disciplinary boundaries. By leveraging the complementary strengths of models of varying complexity, we develop an explainable AI framework for Earth system emulators. It bridges the model hierarchy through a reconfigured latent data assimilation technique, uniquely suited to exploit the sparse output from the idealized models. The resulting bridging model inherits the high resolution and comprehensive variables of operational models while achieving global accuracy enhancements through targeted improvements from idealized models. Crucially, the mechanism of AI provides a clear rationale for these advancements, moving beyond black-box correction to physically insightful understanding in a computationally efficient framework that enables effective physics-assisted digital twins and uncertainty quantification. We demonstrate its power by significantly correcting biases in CMIP6 simulations of El Niño spatiotemporal patterns, leveraging statistically accurate idealized models. This work also highlights the importance of pushing idealized model development and advancing communication between modeling communities.

Predicting high-dimensional dynamical systems with irregular time steps presents significant challenges for current data-driven algorithms. These irregularities arise from missing data, sparse observations, or adaptive computational techniques, reducing prediction accuracy. To address these limitations, we propose a novel method: a Physics-Spatiotemporal Masked Autoencoder. This method integrates convolutional autoencoders for spatial feature extraction with masked autoencoders optimised for irregular time series, leveraging attention mechanisms to reconstruct the entire physical sequence in a single prediction pass. The model avoids the need for data imputation while preserving the physical integrity of the system. Here, 'physics' refers to high-dimensional fields generated from underlying dynamical systems, rather than enforcing explicit physical constraints or PDE residuals. We evaluate this approach on multiple simulated datasets and real-world ocean temperature data. The results demonstrate that our method achieves significant improvements in prediction accuracy, robustness to nonlinearities, and computational efficiency over traditional convolutional and recurrent network methods. The model shows potential for capturing complex spatiotemporal patterns without requiring domain-specific knowledge, with applications in climate modelling, fluid dynamics, ocean forecasting, environmental monitoring, and scientific computing.

In recent decades, the intensification of wildfire activity in western Canada has resulted in substantial socio-economic and environmental losses. Accurate wildfire risk prediction is hindered by the intrinsic stochasticity of ignition and spread and by nonlinear interactions among fuel conditions, meteorology, climate variability, topography, and human activities, challenging the reliability and interpretability of purely data-driven models. We propose a trustworthy data-driven wildfire risk prediction framework based on long-sequence, multi-scale temporal modeling, which integrates heterogeneous drivers while explicitly quantifying predictive uncertainty and enabling process-level interpretation. Evaluated over western Canada during the record-breaking 2023 and 2024 fire seasons, the proposed model outperforms existing time-series approaches, achieving an F1 score of 0.90 and a PR-AUC of 0.98 with low computational cost. Uncertainty-aware analysis reveals structured spatial and seasonal patterns in predictive confidence, highlighting increased uncertainty associated with ambiguous predictions and spatiotemporal decision boundaries. SHAP-based interpretation provides mechanistic understanding of wildfire controls, showing that temperature-related drivers dominate wildfire risk in both years, while moisture-related constraints play a stronger role in shaping spatial and land-cover-specific contrasts in 2024 compared to the widespread hot and dry conditions of 2023. Data and code are available at https://github.com/SynUW/mmFire.

Transformer-based neural operators have shown remarkable performance for approximating solution operators of partial differential equations on complex geometries. However, existing approaches implicitly assume a fixed domain size, which limits their ability to generalize at inference. In this work, we investigate domain extension, namely zero-shot inference on spatial domains that are significantly larger than those encountered during training. We argue that this setting fundamentally requires spatial locality and translation equivariance. We propose to implement this locality via a decomposable bias in the attention logits computation, enabling finely controllable locality while remaining fully decomposable into query-key inner products and directly compatible with optimized attention kernels. Combined with rotary positional embeddings, it enables expressive embeddings with controllable spatial support without altering the transformer architecture. We empirically show that our approach substantially improves zero-shot generalization to larger domains across two PDE benchmarks and a 3D industrial atmospheric flow application. Our code and datasets are available at https://github.com/cerea-daml/domain-extension.

Air flow modeling at a local scale is essential for applications such as pollutant dispersion modeling or wind farm modeling. To circumvent costly Computational Fluid Dynamics (CFD) computations, deep learning surrogate models have recently emerged as promising alternatives. However, in the context of urban air flow, deep learning models struggle to adapt to the high variations of the urban geometry and to large mesh sizes. To tackle these challenges, we introduce Anchored Branched Steady-state WInd Flow Transformer (AB-SWIFT), a transformer-based model with an internal branched structure uniquely designed for atmospheric flow modeling. We train our model on a specially designed database of atmospheric simulations around randomised urban geometries and with a mixture of unstable, neutral, and stable atmospheric stratifications. Our model reaches the best accuracy on all predicted fields compared to state-of-the-art transformers and graph-based models. Our code and data is available at https://github.com/cerea-daml/abswift.

The growing frequency of wildfires globally has highlighted the importance of immediate fire forecasting. Traditional high-accuracy fire spread simulations, like cellular automata and computational fluid dynamics, are detailed but require extensive computational resources and time. Consequently, there has been a significant push towards developing machine learning-based fire prediction models. These models, while effective, tend to be specific to certain regions and demand a large volume of simulation data for training, leading to considerable computational demands across various ecoregions. In response, this study introduces a generative approach using three-dimensional Vector-Quantized Variational Autoencoders. This method is designed to create spatial-temporal sequences predicting the progression of future wildfires in specific ecoregions. The effectiveness of this model was evaluated in the context of the Chimney fire, a notable recent wildfire in California. The results demonstrate that the model effectively produces realistic and structured fire scenarios, incorporating influential geophysical factors like vegetation and terrain slope. Additionally, the data generated by this model were used to develop and train a surrogate model for wildfire spread prediction. This surrogate model was successfully validated using both simulated data and actual data from the Chimney fire incident.

Data-driven approaches have emerged as a powerful alternative to traditional numerical methods for forecasting physical systems, offering fast inference and reduced computational costs. However, for complex systems and those without prior knowledge, the accuracy of long-term predictions frequently deteriorates due to error accumulation. Existing solutions often adopt an autoregressive approach that unrolls multiple time steps during each training iteration; although effective for long-term forecasting, this method requires storing entire unrolling sequences in GPU memory, leading to high resource demands. Moreover, optimizing for long-term accuracy in autoregressive frameworks can compromise short-term performance. To address these challenges, we introduce the Stochastic PushForward (SPF) training framework in this paper. SPF preserves the one-step-ahead training paradigm while still enabling multi-step-ahead learning. It dynamically constructs a supplementary dataset from the model's predictions and uses this dataset in combination with the original training data. By drawing inputs from both the ground truth and model-generated predictions through a stochastic acquisition strategy, SPF naturally balances short-and long-term predictive performance and further reduces overfitting and improves generalization. Furthermore, the training process is executed in a one-step-ahead manner, with multi-step-ahead predictions precomputed between epochs-thus eliminating the need to retain entire unrolling sequences in memory, thus keeping memory usage stable. We demonstrate the effectiveness of SPF on the Burgers' equation and the Shallow Water benchmark. Experimental results demonstrated that SPF delivers superior long-term accuracy compared to autoregressive approaches while reducing memory consumption. This positions SPF as a promising solution for resource-constrained environments and complex physical simulations.

The lattice Boltzmann equation (LBE), rooted in kinetic theory, provides a powerful framework for capturing complex flow behaviour by describing the evolution of single-particle distribution functions (PDFs). Despite its success, solving the LBE numerically remains computationally intensive due to strict time-step restrictions imposed by collision kernels. Here, we introduce a physics-informed neural operator framework for the LBE that enables prediction over large time horizons without step-by-step integration, effectively bypassing the need to explicitly solve the collision kernel. We incorporate intrinsic moment-matching constraints of the LBE, along with global equivariance of the full distribution field, enabling the model to capture the complex dynamics of the underlying kinetic system. Our framework is discretization-invariant, enabling models trained on coarse lattices to generalise to finer ones (kinetic super-resolution). In addition, it is agnostic to the specific form of the underlying collision model, which makes it naturally applicable across different kinetic datasets regardless of the governing dynamics. Our results demonstrate robustness across complex flow scenarios, including von Karman vortex shedding, ligament breakup, and bubble adhesion. This establishes a new data-driven pathway for modelling kinetic systems.

The rapid growth of data-driven applications is ubiquitous across virtually all scientific domains, led to an increasing demand for effective methods to handle data deficiencies and mitigate the imperfect data. This paper presents a guide for researchers encountering real-world data-driven applications, and the respective challenges associated with this. This article proposes the concept of the Data Paradigm, combining the principles of machine learning, data science and data assimilation to tackle challenges in data-driven applications. Models area product of the data upon which they are trained, data collected from real world scenarios is perfect due to natural limitations of sensing and collection. computational modelling of real world systems is intrinsically limited by the various deficiencies encountered in real data. The Data Learning Paradigm aims to leverage the strengths of data improvement to the accuracy, reliability, and interpretability of data-driven models. We outline a range of methods are currently being implemented in the field of Data Learning involving machine learning and data methods, and discuss how these mitigate the various problems associated with data-driven models, illustrating improved results in a multitude of real world applications. We highlight examples where these methods led to significant advancements infields such as environmental monitoring, planetary exploration, healthcare analytics, linguistic analysis, social networks, and smart manufacturing. We offer a guide to how these may be implemented to deal with general types of limitations in data, alongside their current and applications.

Predicting the extent of massive wildfires once ignited is essential to reduce the subsequent socioeconomic losses and environmental damage, but challenging because of the complexity of fire behaviour. Existing physics-based models are limited in predicting large or long-duration wildfire events. Here, we develop a deep-learning-based predictive model, Fire-Image-DenseNet (FIDN), that uses spatial features derived from both near real-time and reanalysis data on the environmental and meteorological drivers of wildfire. We trained and tested this model using more than 300 individual wildfires that occurred between 2012 and 2019 in the western US. In contrast to existing models, the performance of FIDN does not degrade with fire size or duration. Furthermore, it predicts final burnt area accurately even in very heterogeneous landscapes in terms of fuel density and flammability. The FIDN model showed higher accuracy, with a mean squared error (MSE) about 82% and 67% lower than those of the predictive models based on cellular automata (CA) and the minimum travel time (MTT) approaches, respectively. Its structural similarity index measure (SSIM) averages 97%, outperforming the CA and FlamMap MTT models by 6% and 2%, respectively. Additionally, FIDN is approximately three orders of magnitude faster than both CA and MTT models. The enhanced computational efficiency and accuracy advancements offer vital insights for strategic planning and resource allocation for firefighting operations.

Reliable and accurate reconstruction for large-scale and complex physical fields in real-time from limited observations has been a longstanding challenge. In recent years, sensors have been increasingly deployed in numerous physical systems. However, the locations of these sensors can shift over time, such as with mobile sensors, or when sensors are deployed and removed. These sparse and randomly located sensors further exacerbate the difficulty of reconstructing the physical field. In this paper, we present a new deep learning model called Vision Transformer-based Autoencoder (ViTAE) for reconstructing large-scale and complex fields. The proposed network structure is based on a novel core design: vision transformer encoder and Convolutional Neural Network (CNN) decoder. First, we split a two-dimensional field into patches and developed a vision transformer encoder to transfer patches into latent representations. We then reshape the linear latent representations to patches before concatenation, along with a CNN decoder, to reconstruct the field. The proposed model is tested in four different numerical experiments, using generated synthetic data, spatially distributed PM2.5 data, Computational Fluid Dynamics (CFD) simulation data and National Oceanic and Atmospheric Administration (NOAA) sea surface temperature data. The numerical results highlight the strength of ViTAE-SL compared to Kriging and state-of-the-art deep-learning models with significantly higher reconstruction accuracy, computational efficiency, and robust scaling behavior.

Electrocardiography (ECG) signals can be considered as multivariable time series (TS). The state-of-the-art ECG data classification approaches, based on either feature engineering or deep learning techniques, treat separately spectral and time domains in machine learning systems. No spectral-time domain communication mechanism inside the classifier model can be found in current approaches, leading to difficulties in identifying complex ECG forms. In this article, we proposed a novel deep learning model named spectral cross-domain neural network (SCDNN) with a new block called soft-adaptive threshold spectral enhancement (SATSE), to simultaneously reveal the key information embedded in spectral and time domains inside the neural network. More precisely, the domain-cross information is captured by a general convolutional neural network (CNN) backbone, and different information sources are merged by a self-adaptive mechanism to mine the connection between time and spectral domains. In SATSE, the knowledge from time and spectral domains is extracted via the fast Fourier transformation (FFT) with soft trainable thresholds in modified sigmoid functions. The proposed SCDNN is tested with several classification tasks implemented on the public ECG databases PTB-XL and CPSC2018. SCDNN outperforms the state-of-the-art approaches with a low computational cost regarding a variety of metrics in all classification tasks on both databases, by finding appropriate domains from the infinite spectral mapping. The convergence of the trainable thresholds in the spectral domain is also numerically investigated in this article. The robust performance of SCDNN provides a new perspective to exploit knowledge across deep learning models from time and spectral domains. The code repository can be found: https://github.com/DL-WG/SCDNN-TS.

Vision-Language Pretraining (VLP) has demonstrated remarkable capabilities in learning visual representations from textual descriptions of images without annotations. Yet, effective VLP demands large-scale image-text pairs, a resource that suffers scarcity in the medical domain. Moreover, conventional VLP is limited to 2D images while medical images encompass diverse modalities, often in 3D, making the learning process more challenging. To address these challenges, we present Generative Text-Guided 3D Vision-Language Pretraining for Unified Medical Image Segmentation (GTGM), a framework that extends of VLP to 3D medical images without relying on paired textual descriptions. Specifically, GTGM utilizes large language models (LLM) to generate medical-style text from 3D medical images. This synthetic text is then used to supervise 3D visual representation learning. Furthermore, a negative-free contrastive learning objective strategy is introduced to cultivate consistent visual representations between augmented 3D medical image patches, which effectively mitigates the biases associated with strict positive-negative sample pairings. We evaluate GTGM on three imaging modalities - Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and electron microscopy (EM) over 13 datasets. GTGM's superior performance across various medical image segmentation tasks underscores its effectiveness and versatility, by enabling VLP extension into 3D medical imagery while bypassing the need for paired text.

Wildfires pose a significant threat to ecosystems, wildlife, and human communities, leading to habitat destruction, pollutant emissions, and biodiversity loss. Accurate wildfire risk prediction is crucial for mitigating these impacts and safeguarding both environmental and human health. This paper provides a comprehensive review of wildfire risk prediction methodologies, with a particular focus on deep learning approaches combined with remote sensing. We begin by defining wildfire risk and summarizing the geographical distribution of related studies. In terms of data, we analyze key predictive features, including fuel characteristics, meteorological and climatic conditions, socioeconomic factors, topography, and hydrology, while also reviewing publicly available wildfire prediction datasets derived from remote sensing. Additionally, we emphasize the importance of feature collinearity assessment and model interpretability to improve the understanding of prediction outcomes. Regarding methodology, we classify deep learning models into three primary categories: time-series forecasting, image segmentation, and spatiotemporal prediction, and further discuss methods for converting model outputs into risk classifications or probability-adjusted predictions. Finally, we identify the key challenges and limitations of current wildfire-risk prediction models and outline several research opportunities. These include integrating diverse remote sensing data, developing multimodal models, designing more computationally efficient architectures, and incorporating cross-disciplinary methods—such as coupling with numerical weather-prediction models—to enhance the accuracy and robustness of wildfire-risk assessments.

Accurate and rapid prediction of wildfire trends is crucial for effective management and mitigation. However, the stochastic nature of fire propagation poses significant challenges in developing reliable simulators. In this paper, we introduce PyTorchFire, an open-access, PyTorch-based software that leverages GPU acceleration. With our redesigned differentiable wildfire Cellular Automata (CA) model, we achieve millisecond-level computational efficiency, significantly outperforming traditional CPU-based wildfire simulators on real-world-scale fires at high resolution. Real-time parameter calibration is made possible through gradient descent on our model, aligning simulations closely with observed wildfire behavior both temporally and spatially, thereby enhancing the realism of the simulations. Our PyTorchFire simulator, combined with real-world environmental data, demonstrates superior generalizability compared to supervised learning surrogate models. Its ability to predict and calibrate wildfire behavior in real-time ensures accuracy, stability, and efficiency. PyTorchFire has the potential to revolutionize wildfire simulation, serving as a powerful tool for wildfire prediction and management.

Unstructured grid data are essential for modelling complex geometries and dynamics in computational physics. Yet, their inherent irregularity presents significant challenges for conventional machine learning (ML) techniques. This paper provides a comprehensive review of advanced ML methodologies designed to handle unstructured grid data in high-dimensional dynamical systems. Key approaches discussed include graph neural networks, transformer models with spatial attention mechanisms, interpolation-integrated ML methods, and meshless techniques such as physics-informed neural networks. These methodologies have proven effective across diverse fields, including fluid dynamics and environmental simulations. This review is intended as a guidebook for computational scientists seeking to apply ML approaches to unstructured grid data in their domains, as well as for ML researchers looking to address challenges in computational physics. It places special focus on how ML methods can overcome the inherent limitations of traditional numerical techniques and, conversely, how insights from computational physics can inform ML development. For this purpose, we mainly focus in this review on recent papers from the past decade that reflect strong interactions between computational physics and deep learning methods. To support benchmarking, this review also provides a summary of open-access datasets of unstructured grid data in computational physics. Finally, emerging directions such as generative models with unstructured data, reinforcement learning for mesh generation, and hybrid physics-data-driven paradigms are discussed to inspire future advancements in this evolving field.