arXiv subjects

Ankur Mahesh

Publications and source records attributed to Ankur Mahesh.

12 recordsLinked to original sources

arXiv2026

Surface temperature extremes produced by huge machine learning hindcasts of summer 2023

The summer of 2023 brought record-breaking heat extremes across the globe. In this work, we investigate how much worse these extremes might have become, given identical large-scale conditions. Numerical weather prediction (NWP) ensembles simulate alternate heatwave storylines, but their limited sample size diminishes their statistical coverage of observed heatwaves and ability to represent worst-case events. Using the Spherical Fourier Neural Operator machine learning (ML) weather model, we generated a huge ensemble of 7,424 storyline forecasts of summer temperature extremes. Despite being constrained by the same initial and boundary conditions and using training data from the same modeling system, the ML ensemble produced extreme heatwave scenarios exceeding temperatures from NWP ensembles. For 30% of global land areas, ML-generated extreme events were outside the prediction envelope from extrapolating smaller NWP ensembles with extreme value theory. This research demonstrates that very large ensembles substantially improve upon NWP-based characterization of worst-case heat extremes.

stat.AP↗

arXiv2026

Anomalous Diffusion of Tropical Cyclones Observed in Huge Ensembles of Hindcasts

We examine whether tropical cyclones (TCs) obey ordinary Brownian or anomalous diffusion using a huge ensemble (HENS) of hindcasts for summer 2023. Anomalous diffusion has been inferred for actual TCs from the fluctuations in their tracks from the shortest paths between the initiation and termination of each cyclone. We reproduce the same anomalous diffusion power laws connecting spatial position and time using HENS. In addition, we show that the variance in the position of a single TC across HENS since initiation follows a scaling law with time that, in some cases, corresponds to ballistic motion of the TC through the background atmospheric flow. This determination was enabled by the exceptional statistics determined from thousands of plausible yet counterfactual recreations of 34 individual TCs. HENS consists of 7424 15-day hindcasts initiated from observed atmospheric conditions each day from June 1, 2023 to August 31, 2023 using the ECMWF ERA5 meteorological reanalysis. The hindcasts were generated using NVIDIA's Spherical Fourier Neural Operator (SFNO) machine-learning-based weather and climate emulator. We identify tropical cyclones in HENS using a variant of the Tempest Extremes detection and tracking frameworks for TCs with adjustments to the disposable parameters to minimize the numbers of false positives and negatives relative to the International Best Track Archive for Climate Stewardship (IBTrACS) records for TCs observed in summer 2023. We conclude with the implications of our findings for the predictability of TC tracks and landfall locations on lead times of days to weeks.

physics.ao-ph↗

arXiv2026

Examining Fast Radiatively Driven Responses Using Machine-Learning Weather Emulators

The response of the climate system to increased greenhouse gases and other radiative perturbations is governed by a combination of fast and slow feedbacks. Slow feedbacks are typically activated in response to changes in ocean temperatures on decadal timescales and manifest as changes in climatic state with no recent historical analogue. However, fast feedbacks are activated in response to rapid atmospheric physical processes on weekly timescales, and they are already operative in the present-day climate. This distinction implies that the physics of fast radiative feedbacks is present in the historical meteorological reanalyses used to train many recent successful machine-learning-based (ML) emulators of weather and climate. In addition, these feedbacks are functional under the historical boundary conditions pertaining to the top-of-atmosphere radiative balance and sea-surface temperatures. Together, these factors imply that we can use historically trained ML weather emulators to study the response of radiative-convective equilibrium (RCE), and hence the global hydrological cycle, to perturbations in carbon dioxide and other well-mixed greenhouse gases. Without retraining on prospective Earth system conditions, we use ML weather emulators to quantify the fast precipitation response to reduced and elevated carbon dioxed concentrations with no recent historical precedent. We show that the responses from historically trained emulators agree with those produced by full-physics Earth System Models (ESMs). In conclusion, we discuss the prospects for and advantages from using ESMs and ML emulators to study fast processes in global climate.

physics.ao-ph↗

arXiv2026

Watch an AI Weather Model Learn (and Unlearn) Tropical Cyclones

In a changing climate, artificial intelligence (AI) weather models have the potential to provide cheaper, faster, and more accurate forecasts of high-impact weather events. To realize this potential and gauge trustworthiness, there is a need for more research on how models learn extreme events and how that learning might be improved. Here, we investigate how a Spherical Fourier Neural Operator (SFNO) learns tropical cyclones (TCs) by saving every checkpoint from training and analyzing storm specific metrics. We find evidence that for some storms the SFNO learns information about TC intensity that it loses later in training. This unlearning pattern is associated with anomalously moist environments and may be due to the model unlearning the relationship between moisture and TC intensity. This work provides a first example of leveraging task-specific training dynamics to further our understanding of how AI weather models learn extreme events.

physics.ao-ph↗

arXiv2025

Constraining Atmospheric River Uncertainty Using Instantaneous Poleward Latent Heat Transport

Atmospheric rivers (ARs) are extreme weather events that play a crucial role in the global hydrological cycle. As a key mechanism of latent heat transport (LHT), they help maintain energy balance in the climate system. While an AR is characterized by a long, narrow corridor of water vapor associated with a low-level jet stream, there is no unambiguous definition of an AR grounded in geophysical fluid dynamics. AR identification is currently performed by a variety of threshold-based algorithms, which has introduced uncertainty in the estimated contribution of ARs to LHT. We calculate the instantaneous eddy LHT from moist, poleward anomalies. Based on the dynamics of the large-scale atmospheric circulation, this quantity is a physics-based upper bound that constrains AR projections from the variety of detection algorithms. We quantify the contribution of ARs to transient eddies, stationary eddies, and transient-stationary eddy interactions, and we show the relative contributions of ARs vs. other processes, such as dry, equatorward transport. We use this upper bound as a reference to quantify ARs' frequency, intensity, and temporal variability. In the historical climate, the AR reference transport is ~2.21 PW at the latitude of peak transport in Northern Hemisphere winter, with a temporal standard deviation of approximately 0.47 PW. In a future climate projection, at this latitude, AR-induced LHT will increase by 0.5 PW and the corresponding temporal variability will increase by 0.14 PW. The future change in the AR reference transport is larger than the changes associated with other types of atmospheric anomalies.

physics.ao-ph↗

arXiv2025

How does an AI Weather Model Learn to Forecast Extreme Weather?

In a warming climate with more frequent severe weather, artificial intelligence (AI) weather models have the potential to provide cheaper, faster, and more accurate forecasts of high-impact weather events. To realize this potential, there is a need for more research on how models learn extreme events and how that learning might be improved. We investigate how a spherical Fourier neural operator model (SFNO) learns extreme weather by saving every checkpoint throughout training and analyzing a collection of 9 extreme weather events including heatwaves, atmospheric rivers, and tropical cyclones. The SFNO learns heatwaves similarly to other weather days, but we find evidence that the model learns information about atmospheric river and tropical cyclone forecasts that it loses later in training. We propose a possible training strategy to improve the forecasting of extreme events by retaining information from earlier training checkpoints, and provide initial evidence of its utility.

physics.ao-ph↗

arXiv2025

FourCastNet 3: A geometric approach to probabilistic machine-learning weather forecasting at scale

FourCastNet 3 advances global weather modeling by implementing a scalable, geometric machine learning (ML) approach to probabilistic ensemble forecasting. The approach is designed to respect spherical geometry and to accurately model the spatially correlated probabilistic nature of the problem, resulting in stable spectra and realistic dynamics across multiple scales. FourCastNet 3 delivers forecasting accuracy that surpasses leading conventional ensemble models and rivals the best diffusion-based methods, while producing forecasts 8 to 60 times faster than these approaches. In contrast to other ML approaches, FourCastNet 3 demonstrates excellent probabilistic calibration and retains realistic spectra, even at extended lead times of up to 60 days. All of these advances are realized using a purely convolutional neural network architecture tailored for spherical geometry. Scalable and efficient large-scale training on 1024 GPUs and more is enabled by a novel training paradigm for combined model- and data-parallelism, inspired by domain decomposition methods in classical numerical models. Additionally, FourCastNet 3 enables rapid inference on a single GPU, producing a 60-day global forecast at 0.25°, 6-hourly resolution in under 4 minutes. Its computational efficiency, medium-range probabilistic skill, spectral fidelity, and rollout stability at subseasonal timescales make it a strong candidate for improving meteorological forecasting and early warning systems through large ensemble predictions.

cs.LG↗

arXiv2025

Huge Ensembles Part II: Properties of a Huge Ensemble of Hindcasts Generated with Spherical Fourier Neural Operators

In Part I, we created an ensemble based on Spherical Fourier Neural Operators. As initial condition perturbations, we used bred vectors, and as model perturbations, we used multiple checkpoints trained independently from scratch. Based on diagnostics that assess the ensemble's physical fidelity, our ensemble has comparable performance to operational weather forecasting systems. However, it requires orders of magnitude fewer computational resources. Here in Part II, we generate a huge ensemble (HENS), with 7,424 members initialized each day of summer 2023. We enumerate the technical requirements for running huge ensembles at this scale. HENS precisely samples the tails of the forecast distribution and presents a detailed sampling of internal variability. HENS has two primary applications: (1) as a large dataset with which to study the statistics and drivers of extreme weather and (2) as a weather forecasting system. For extreme climate statistics, HENS samples events 4$σ$ away from the ensemble mean. At each grid cell, HENS increases the skill of the most accurate ensemble member and enhances coverage of possible future trajectories. As a weather forecasting model, HENS issues extreme weather forecasts with better uncertainty quantification. It also reduces the probability of outlier events, in which the verification value lies outside the ensemble forecast distribution.

cs.LG↗

arXiv2025

Huge Ensembles Part I: Design of Ensemble Weather Forecasts using Spherical Fourier Neural Operators

Studying low-likelihood high-impact extreme weather events in a warming world is a significant and challenging task for current ensemble forecasting systems. While these systems presently use up to 100 members, larger ensembles could enrich the sampling of internal variability. They may capture the long tails associated with climate hazards better than traditional ensemble sizes. Due to computational constraints, it is infeasible to generate huge ensembles (comprised of 1,000-10,000 members) with traditional, physics-based numerical models. In this two-part paper, we replace traditional numerical simulations with machine learning (ML) to generate hindcasts of huge ensembles. In Part I, we construct an ensemble weather forecasting system based on Spherical Fourier Neural Operators (SFNO), and we discuss important design decisions for constructing such an ensemble. The ensemble represents model uncertainty through perturbed-parameter techniques, and it represents initial condition uncertainty through bred vectors, which sample the fastest growing modes of the forecast. Using the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System (IFS) as a baseline, we develop an evaluation pipeline composed of mean, spectral, and extreme diagnostics. Using large-scale, distributed SFNOs with 1.1 billion learned parameters, we achieve calibrated probabilistic forecasts. As the trajectories of the individual members diverge, the ML ensemble mean spectra degrade with lead time, consistent with physical expectations. However, the individual ensemble members' spectra stay constant with lead time. Therefore, these members simulate realistic weather states, and the ML ensemble thus passes a crucial spectral test in the literature. The IFS and ML ensembles have similar Extreme Forecast Indices, and we show that the ML extreme weather forecasts are reliable and discriminating.

physics.ao-ph↗

arXiv2024

A Practical Probabilistic Benchmark for AI Weather Models

Since the weather is chaotic, forecasts aim to predict the distribution of future states rather than make a single prediction. Recently, multiple data driven weather models have emerged claiming breakthroughs in skill. However, these have mostly been benchmarked using deterministic skill scores, and little is known about their probabilistic skill. Unfortunately, it is hard to fairly compare AI weather models in a probabilistic sense, since variations in choice of ensemble initialization, definition of state, and noise injection methodology become confounding. Moreover, even obtaining ensemble forecast baselines is a substantial engineering challenge given the data volumes involved. We sidestep both problems by applying a decades-old idea -- lagged ensembles -- whereby an ensemble can be constructed from a moderately-sized library of deterministic forecasts. This allows the first parameter-free intercomparison of leading AI weather models' probabilistic skill against an operational baseline. The results reveal that two leading AI weather models, i.e. GraphCast and Pangu, are tied on the probabilistic CRPS metric even though the former outperforms the latter in deterministic scoring. We also reveal how multiple time-step loss functions, which many data-driven weather models have employed, are counter-productive: they improve deterministic metrics at the cost of increased dissipation, deteriorating probabilistic skill. This is confirmed through ablations applied to a spherical Fourier Neural Operator (SFNO) approach to AI weather forecasting. Separate SFNO ablations modulating effective resolution reveal it has a useful effect on ensemble dispersion relevant to achieving good ensemble calibration. We hope these and forthcoming insights from lagged ensembles can help guide the development of AI weather forecasts and have thus shared the diagnostic code.

physics.ao-ph↗

arXiv2024

Analyzing and Exploring Training Recipes for Large-Scale Transformer-Based Weather Prediction

The rapid rise of deep learning (DL) in numerical weather prediction (NWP) has led to a proliferation of models which forecast atmospheric variables with comparable or superior skill than traditional physics-based NWP. However, among these leading DL models, there is a wide variance in both the training settings and architecture used. Further, the lack of thorough ablation studies makes it hard to discern which components are most critical to success. In this work, we show that it is possible to attain high forecast skill even with relatively off-the-shelf architectures, simple training procedures, and moderate compute budgets. Specifically, we train a minimally modified SwinV2 transformer on ERA5 data, and find that it attains superior forecast skill when compared against IFS. We present some ablations on key aspects of the training pipeline, exploring different loss functions, model sizes and depths, and multi-step fine-tuning to investigate their effect. We also examine the model performance with metrics beyond the typical ACC and RMSE, and investigate how the performance scales with model size.

cs.LG↗

arXiv2018

Exascale Deep Learning for Climate Analytics

We extract pixel-level masks of extreme weather patterns using variants of Tiramisu and DeepLabv3+ neural networks. We describe improvements to the software frameworks, input pipeline, and the network training algorithms necessary to efficiently scale deep learning on the Piz Daint and Summit systems. The Tiramisu network scales to 5300 P100 GPUs with a sustained throughput of 21.0 PF/s and parallel efficiency of 79.0%. DeepLabv3+ scales up to 27360 V100 GPUs with a sustained throughput of 325.8 PF/s and a parallel efficiency of 90.7% in single precision. By taking advantage of the FP16 Tensor Cores, a half-precision version of the DeepLabv3+ network achieves a peak and sustained throughput of 1.13 EF/s and 999.0 PF/s respectively.

cs.DC↗