Causal discovery for dynamical systems poses a major challenge in fields where active interventions are infeasible. Most methods used to investigate these systems and their associated benchmarks are tailored to deterministic, low-dimensional and weakly nonlinear time-series data. To address these limitations, we present CausalDynamics, a large-scale benchmark and extensible data generation framework to advance the structural discovery of dynamical causal models. Our benchmark consists of true causal graphs derived from thousands of both linearly and nonlinearly coupled ordinary and stochastic differential equations as well as two idealized climate models. We perform a comprehensive evaluation of state-of-the-art causal discovery algorithms for graph reconstruction on systems with noisy, confounded, and lagged dynamics. CausalDynamics consists of a plug-and-play, build-your-own coupling workflow that enables the construction of a hierarchy of physical systems. We anticipate that our framework will facilitate the development of robust causal discovery algorithms that are broadly applicable across domains while addressing their unique challenges. We provide a user-friendly implementation and documentation on https://kausable.github.io/CausalDynamics.

Data assimilation (DA) aims to estimate the full state of a dynamical system by combining partial and noisy observations with a prior model forecast, commonly referred to as the background. In atmospheric applications, this problem is fundamentally ill-posed due to the sparsity of observations relative to the high-dimensional state space. Traditional methods address this challenge by simplifying background priors to regularize the solution, which are empirical and require continual tuning for application. Inspired by alignment techniques in text-to-image diffusion models, we propose Align-DA, which formulates DA as a generative process and uses reward signals to guide background priors—replacing manual tuning with data-driven alignment. Specifically, we train a score-based model in the latent space to approximate the background-conditioned prior, and align it using three complementary reward signals for DA: (1) assimilation accuracy, (2) forecast skill initialized from the assimilated state, and (3) physical adherence of the analysis fields. Experiments with multiple reward signals demonstrate consistent improvements in analysis quality across different evaluation metrics and observation-guidance strategies. These results show that preference alignment, implemented as a soft constraint, can automatically adapt complex background priors tailored to DA, offering a promising new direction for advancing the field.

Modern climate projections lack adequate spatial and temporal resolution due to computational constraints, leading to inaccuracies in representing critical processes like thunderstorms that occur on the sub-resolution scale. Hybrid methods combining physics with machine learning (ML) offer faster, higher fidelity climate simulations by outsourcing compute-hungry, high-resolution simulations to ML emulators. However, these hybrid physics-ML simulations require domain-specific data and workflows that have been inaccessible to many ML experts. This paper is an extended version of our NeurIPS award-winning ClimSim dataset paper. The ClimSim dataset includes 5.7 billion pairs of multivariate input/output vectors spanning ten years at high temporal resolution, capturing the influence of high-resolution, high-fidelity physics on a host climate simulator's macro-scale state. In this extended version, we introduce a significant new contribution in Section 5, which provides a cross-platform, containerized pipeline to integrate ML models into operational climate simulators for hybrid testing. We also implement various baselines of ML models and hybrid simulators to highlight the ML challenges of building stable, skillful emulators. The data (https://huggingface.co/datasets/LEAP/ClimSim_high-res, also in a low-resolution version at https://huggingface.co/datasets/LEAP/ClimSim_low-res and an aquaplanet version at https://huggingface.co/datasets/LEAP/ClimSim_low-res_aqua-planet) and code (https://leap-stc.github.io/ClimSim and https://github.com/leap-stc/climsim-online) are publicly released to support the development of hybrid physics-ML and high-fidelity climate simulations.

Accurate prediction of climate in the subseasonal-to-seasonal scale is crucial for disaster preparedness and robust decision making amidst climate change. Yet, forecasting beyond the weather timescale is challenging because it deals with problems other than initial condition, including boundary interaction, butterfly effect, and our inherent lack of physical understanding. At present, existing benchmarks tend to have shorter forecasting range of up-to 15 days, do not include a wide range of operational baselines, and lack physics-based constraints for explainability. Thus, we propose ChaosBench, a challenging benchmark to extend the predictability range of data-driven weather emulators to S2S timescale. First, ChaosBench is comprised of variables beyond the typical surface-atmospheric ERA5 to also include ocean, ice, and land reanalysis products that span over 45 years to allow for full Earth system emulation that respects boundary conditions. We also propose physics-based, in addition to deterministic and probabilistic metrics, to ensure a physically-consistent ensemble that accounts for butterfly effect. Furthermore, we evaluate on a diverse set of physics-based forecasts from four national weather agencies as baselines to our data-driven counterpart such as ViT/ClimaX, PanguWeather, GraphCast, and FourCastNetV2. Overall, we find methods originally developed for weather-scale applications fail on S2S task: their performance simply collapse to an unskilled climatology. Nonetheless, we outline and demonstrate several strategies that can extend the predictability range of existing weather emulators, including the use of ensembles, robust control of error propagation, and the use of physics-informed models. Our benchmark, datasets, and instructions are available at https://leap-stc.github.io/ChaosBench.

Modern climate projections lack adequate spatial and temporal resolution due to computational constraints. A consequence is inaccurate and imprecise predictions of critical processes such as storms. Hybrid methods that combine physics with machine learning (ML) have introduced a new generation of higher fidelity climate simulators that can sidestep Moore's Law by outsourcing compute-hungry, short, high-resolution simulations to ML emulators. However, this hybrid ML-physics simulation approach requires domain-specific treatment and has been inaccessible to ML experts because of lack of training data and relevant, easy-to-use workflows. We present ClimSim, the largest-ever dataset designed for hybrid ML-physics research. It comprises multi-scale climate simulations, developed by a consortium of climate scientists and ML researchers. It consists of 5.7 billion pairs of multivariate input and output vectors that isolate the influence of locally-nested, high-resolution, high-fidelity physics on a host climate simulator's macro-scale physical state.The dataset is global in coverage, spans multiple years at high sampling frequency, and is designed such that resulting emulators are compatible with downstream coupling into operational climate simulators. We implement a range of deterministic and stochastic regression baselines to highlight the ML challenges and their scoring. The data (https://huggingface.co/datasets/LEAP/ClimSim_high-res) and code (https://leap-stc.github.io/ClimSim) are released openly to support the development of hybrid ML-physics and high-fidelity climate simulations for the benefit of science and society.

Clouds play a key role in regulating climate change but are difficult to simulate within Earth system models (ESMs). Improving the representation of clouds is one of the key tasks toward more robust climate change projections. This study introduces a new machine-learning-based framework relying on satellite observations to improve understanding of the representation of clouds and their relevant processes in climate models. The proposed method is capable of assigning distributions of established cloud types to coarse data. It facilitates a more objective evaluation of clouds in ESMs and improves the consistency of cloud process analysis. The method is built on satellite data from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument labeled by deep neural networks with cloud types defined by the World Meteorological Organization (WMO), using cloud-type labels from CloudSat as ground truth. The method is applicable to datasets with information about physical cloud variables comparable to MODIS satellite data and at sufficiently high temporal resolution. We apply the method to alternative satellite data from the Cloud_cci project (ESA Climate Change Initiative), coarse-grained to typical resolutions of climate models. The resulting cloud-type distributions are physically consistent and the horizontal resolutions typical of ESMs are sufficient to apply our method. We recommend outputting crucial variables required by our method for future ESM data evaluation. This will enable the use of labeled satellite data for a more systematic evaluation of clouds in climate models.

Data-driven algorithms, in particular neural networks, can emulate the effect of sub-grid scale processes in coarse-resolution climate models if trained on high-resolution climate simulations. However, they may violate key physical constraints and lack the ability to generalize outside of their training set. Here, we show that physical constraints can be enforced in neural networks, either approximately by adapting the loss function or to within machine precision by adapting the architecture. As these physical constraints are insufficient to guarantee generalizability, we additionally propose to physically rescale the training and validation data to improve the ability of neural networks to generalize to unseen climates. Code-github.com/theucler/CBRAIN-CAM.

OBJECTIVE: Birth month and climate impact lifetime disease risk, while the underlying exposures remain largely elusive. We seek to uncover distal risk factors underlying these relationships by probing the relationship between global exposure variance and disease risk variance by birth season. MATERIAL AND METHODS: This study utilizes electronic health record data from 6 sites representing 10.5 million individuals in 3 countries (United States, South Korea, and Taiwan). We obtained birth month-disease risk curves from each site in a case-control manner. Next, we correlated each birth month-disease risk curve with each exposure. A meta-analysis was then performed of correlations across sites. This allowed us to identify the most significant birth month-exposure relationships supported by all 6 sites while adjusting for multiplicity. We also successfully distinguish relative age effects (a cultural effect) from environmental exposures. RESULTS: Attention deficit hyperactivity disorder was the only identified relative age association. Our methods identified several culprit exposures that correspond well with the literature in the field. These include a link between first-trimester exposure to carbon monoxide and increased risk of depressive disorder (R = 0.725, confidence interval [95% CI], 0.529-0.847), first-trimester exposure to fine air particulates and increased risk of atrial fibrillation (R = 0.564, 95% CI, 0.363-0.715), and decreased exposure to sunlight during the third trimester and increased risk of type 2 diabetes mellitus (R = -0.816, 95% CI, -0.5767, -0.929). CONCLUSION: A global study of birth month-disease relationships reveals distal risk factors involved in causal biological pathways that underlie them.

Global soil moisture estimates at fine spatial resolutions is necessary for many applications. However, current spaceborne instruments have coarse resolution. In this study, we develop a new Artificial Neural Network (ANN) based disaggregation algorithm to downscale soil moisture observations from Soil Moisture Active Passive (SMAP) mission to a fine resolution of ~2km using ancillary data from visible/infrared frequencies. We use soil moisture estimates from SMAP at two different spatial resolutions to train the downscaling algorithm. Results show that ANN can successfully capture the complex relationship between the soil moisture estimates at two different spatial resolutions using the ancillary data provided.

Plant's photosynthetic activity and transpiration are constrained by the amount of water available to them through roots (i.e. root zone soil moisture) as well as nutrient and atmospheric conditions. Therefore, to better understand the response of plants to different stress conditions, knowledge of root zone soil moisture is essential. However, current global satellites dedicated to soil moisture monitoring are limited to L-band frequencies that have a low (<; 5cm) penetration depth. In this study, we implement a new root zone soil moisture retrieval algorithm that takes advantage of multi-frequency microwave observations to infer root soil moisture from L-band measurements and inspired by plant hydraulics. The algorithm is a statistical retrieval that uses a set of target data to train an artificial neural network. Results of applying the retrieval algorithm to one year of observations along with future validation measures is presented.

The SUD-MED project aims at monitoring water resources over Mediterranean regions. As part of the project, this paper presents a method we developed for estimating cereal water requirement. The method consists in driving the simple model developed by the FAO with remotely-sensed data. It was tested on an little area cultivated with wheat in the semi-arid Marrakech plain (Morocco). We use a time series of high spatial resolution images acquired by SPOT-4/HRVIR during the 2001/2002 agricultural season. The method outlines the spatio-temporal patterns of crop cycles. The associated maps of phenological variables and seasonal evapotranspiration appear consistent with regional rainfall and irrigation features. Perspectives of improvement are finally discussed.

DART (Discrete Anisotropic Radiative Transfer) was developed in 1996 for simulating radiative transfer in 3D scenes. Since then, it was greatly improved to make it more accurate, comprehensive and operational (e.g., simulation of thermal infrared and atmospheric radiative transfer). Presently, a single DART simulation gives 2 major products. (1) 3-D radiation budget of the Earth-Atmosphere system. (2) Optical remote sensing images at any altitude from bottom up to top of the atmosphere, for many view directions, simultaneously in several spectral bands, from the visible up to thermal infrared. DART works with natural landscapes (i.e., forests, field mosaics, etc.) made of trees, grass, rivers, etc. and urban landscapes made of buildings, roads, etc. Topography is simulated with digital elevation models. Atmosphere (vertical profiles, etc.) and Earth surface (spectral reflectance, etc.) databases can be used, sensor characteristics can be accounted for, etc. Moreover, a Graphic User Interface (GUI) is used to input scene parameters and to display scene and DART simulations. Recent improvements of DART (patent (PCT/FR 02/01181)) are presented here.

This paper presents a new approach to aggregate spatially distributed land surface heat fluxes and landscape characteristics, adopting a probabilistic point of view and using remote sensing data. This method is tested in the Marrakech plain (Morocco), with different remote sensing sensors, and different probabilistic approaches, in order to reach the best evaluation of large scale heat fluxes and characteristics probability distributions.