arXiv subjects

Naveen Sudharsan

Publications and source records attributed to Naveen Sudharsan.

11 recordsLinked to original sources

arXiv2026

AFDBench: A Reasoning-First AI Scientist for NationalWeather Service Forecast Discussions

Large language models (LLMs) hallucinate numerical values when generating high-stakes meteorological text, posing risks for weather communication. We present AFDBench, an AI meteorologist that generates professional Area Forecast Discussions (AFDs) by reasoning through structured AI weather forecast data from Google's WeatherNext 2. We introduce AFDBench, the first benchmark for evaluating generative meteorological reasoning, comprising 7,732 expert written discussions from 13 National Weather Service (NWS) offices paired with real AI weather forecast inputs, and three complementary metrics: Met-Align (numerical accuracy), Style-Align (professional dialect adherence), and Input-Grounding (fidelity to source weather data). Zero-shot evaluations reveal that open-source LLMs achieve low Style-Align (~0.33) and moderate Input-Grounding (~0.88), failing to write in the professional NWS register or faithfully use their input data. We apply Group Relative Policy Optimization (GRPO) with domain-specific rewards targeting temperature accuracy, synoptic correctness, and format compliance. On 1,033 held-out samples from two unseen NWS offices, GRPO nearly doubles Style-Align from 0.318 to 0.619 and improves Input-Grounding from 0.881 to 0.940, demonstrating that reinforcement learning teaches a 7B-parameter model to write like a professional meteorologist and faithfully interpret AI weather data.

cs.LG↗

arXiv2026

Aircast-Mars: A Mars Foundation Model for Global Weather Forecasting with HEALPix-Aware Convolutions

Foundation models for planetary atmospheres promise fast, lightweight surrogates of expensive general circulation models (GCMs) for mission planning and scientific inquiry. Here we present Aircast-Mars, a deep-learning weather prediction system for Mars trained on the Ensemble Mars Atmosphere Reanalysis System (EMARS) v1.0. We regrid temperature, zonal wind, and meridional wind fields across 28 vertical levels onto a hierarchical equal-area isolatitude pixelization (HEALPix) mesh at Nside = 64 (~110 km resolution) and train a HEALPix-aware 2D U-Net inspired by the DLESyM architecture to predict the next hourly atmospheric state. The model employs custom inter-face padding that respects the topology of the 12-face HEALPix sphere and modern ConvNeXt residual blocks with capped Gaussian Error Linear Unit (GELU) activations. While containing 4.3 million trainable parameters, a compact size compared to terrestrial weather foundation models, the network achieves a best validation Mean Squared Error (MSE) of 1.58e-5 in normalized units. Recursive autoregressive rollouts remain stable and physically coherent for 25 hours (one Martian sol), with Root Mean Square Error (RMSE) growing monotonically from ~0.004 at t + 1 h to ~0.031 at t + 25 h without divergence. Compared to a baseline 3D U-Net, the HEALPix-aware architecture reduces validation loss by more than an order of magnitude while using fewer parameters. The model generates a one-hour forecast in approximately 0.5 seconds on a single GPU, offering several orders-of-magnitude speedup over traditional numerical GCMs. These results demonstrate that parsimonious, geometry-respecting neural architectures can capture synoptic-scale Martian atmospheric dynamics and provide a foundation for planetary-scale weather forecasting.

physics.ao-ph↗

arXiv2026

AirCast-SR: A Foundation Model for Kilometer-Scale Atmospheric Super-Resolution via Latent Consistency Diffusion

Operational weather prediction at kilometer scales remains computationally prohibitive for traditional numerical weather prediction (NWP) models, limiting forecast access for applications in energy, agriculture, and disaster management that require fine-grained spatiotemporal detail. Here we introduce AirCast-SR, a foundation model for atmospheric super-resolution that downscales global AI weather forecasts from 0.25 degree (~28 km) to 1 km horizontal resolution at hourly temporal resolution, producing 67-hour forecasts of eight coupled surface variables simultaneously. EarthMind-SR employs a three-dimensional U-Net conditioned within a Latent Consistency Model (LCM) diffusion framework, trained on patch-based samples over the contiguous United States (CONUS) using GraphCast forecasts as input and NOAA's Analysis of Record for Calibration (AORC) as the target. The model achieves near-zero bias across all variables and lead times, and its radial power spectral density analysis demonstrates preservation of fine-scale atmospheric structure at wavelengths of 10 km to 100 km where coarser models lose spectral power. We validate EarthMind-SR across three CONUS case studies spanning winter, summer, and spring seasons, and demonstrate zero-shot global transferability over India and Germany using independent surface station observations without any retraining or fine-tuning. As an open-weights foundation model, EarthMind-SR establishes a new paradigm for kilometer-scale AI weather prediction and provides a platform for regional fine-tuning, distillation, and downstream applications in climate services and hazard forecasting.

cs.LG↗

arXiv2026

CoolPath Tool: A Thermal Comfort Path Planning Tool for Urban Mobility

Extreme heat poses a growing challenge for active transportation in cities where conventional weather reporting (e.g. limited air temperature measurement for the whole city) fails to capture the large microclimate variations that pedestrians and cyclists experience. We present a novel walking and biking route planning tool (''CoolPath Tool'') that selects paths based on thermal comfort using the Universal Thermal Climate Index (UTCI) rather than just distance or travel time. This system combines high-resolution thermal modeling with real-time route mapping. We generate city-scale UTCI maps using GPU version of Solar and LongWave Environmental Irradiance Geometry (SOLWEIG) model, to account for urban features (buildings, trees,) and weather conditions. The urban features are pre-mapped using satellite data products and the routes are the roadways. For any given origin and destination, our tool calculates the average UTCI along each possible route and recommends the ''coolest'' route, i.e. the path with the lowest heat stress (often the most shaded or otherwise thermally comfortable), while still being reasonably direct. We demonstrate this in a case study for Austin, Texas. The approach identifies routes that significantly reduce pedestrians' heat exposure (often recommending routes with a much larger proportion of shade). Such thermally-informed route planning has important public well-being, and economic implications: by helping people avoid dangerous heat hotspots and sun-exposed areas, it can reduce the risk of heat-related illness and make walking or biking a safer choice even on hot days. This tool is part of the Austin Digital Twin efforts developed as part the UT-City CoLab needs. The work while demonstrated for Austin, TX is scalable, and transferrable to other cities globally.

physics.soc-ph↗

arXiv2025

UrbanDIFF: A Denoising Diffusion Model for Spatial Gap Filling of Urban Land Surface Temperature Under Dense Cloud Cover

Satellite-derived Land Surface Temperature (LST) products are central to surface urban heat island (SUHI) monitoring due to their consistent grid-based coverage over large metropolitan regions. However, cloud contamination frequently obscures LST observations, limiting their usability for continuous SUHI analysis. Most existing LST reconstruction methods rely on multitemporal information or multisensor data fusion, requiring auxiliary observations that may be unavailable or unreliable under persistent cloud cover. Purely spatial gap-filling approaches offer an alternative, but traditional statistical methods degrade under large or spatially contiguous gaps, while many deep learning based spatial models deteriorate rapidly with increasing missingness. Recent advances in denoising diffusion based image inpainting models have demonstrated improved robustness under high missingness, motivating their adoption for spatial LST reconstruction. In this work, we introduce UrbanDIFF, a purely spatial denoising diffusion model for reconstructing cloud contaminated urban LST imagery. The model is conditioned on static urban structure information, including built-up surface data and a digital elevation model, and enforces strict consistency with revealed cloud free pixels through a supervised pixel guided refinement step during inference. UrbanDIFF is trained and evaluated using NASA MODIS Terra LST data from seven major United States metropolitan areas spanning 2002 to 2025. Experiments using synthetic cloud masks with 20 to 85 percent coverage show that UrbanDIFF consistently outperforms an interpolation baseline, particularly under dense cloud occlusion, achieving SSIM of 0.89, RMSE of 1.2 K, and R2 of 0.84 at 85 percent cloud coverage, while exhibiting slower performance degradation as cloud density increases.

cs.CV↗

arXiv2025

Exploring the design space of diffusion and flow models for data fusion

Data fusion is an essential task in various domains, enabling the integration of multi-source information to enhance data quality and insights. One key application is in satellite remote sensing, where fusing multi-sensor observations can improve spatial and temporal resolution. In this study, we explore the design space of diffusion and flow models for data fusion, focusing on the integration of Defense Meteorological Satellite Program's Operational Linescan System (DMSP-OLS) and Visible Infrared Imaging Radiometer Suite (VIIRS) nighttime lights data. Our approach leverages a diverse set of 2D image-to-image generative models, including UNET, diffusion, and flow modeling architectures. We evaluate the effectiveness of these architectures in satellite remote sensing data fusion, identifying diffusion models based on UNet as particularly adept at preserving fine-grained spatial details and generating high-fidelity fused images. We also provide guidance on the selection of noise schedulers in diffusion-based models, highlighting the trade-offs between iterative solvers for faster inference and discrete schedulers for higher-quality reconstructions. Additionally, we explore quantization techniques to optimize memory efficiency and computational cost without compromising performance. Our findings offer practical insights into selecting the most effective diffusion and flow model architectures for data fusion tasks, particularly in remote sensing applications, and provide recommendations for leveraging noise scheduling strategies to enhance fusion quality.

cs.CV↗

arXiv2025

Towards NoahMP-AI: Enhancing Land Surface Model Prediction with Deep Learning

Accurate soil moisture prediction during extreme events remains a critical challenge for earth system modeling, with profound implications for drought monitoring, flood forecasting, and climate adaptation strategies. While land surface models (LSMs) provide physically-based predictions, they exhibit systematic biases during extreme conditions when their parameterizations operate outside calibrated ranges. Here we present NoahMP-AI, a physics-guided deep learning framework that addresses this challenge by leveraging the complete Noah-MP land surface model as a comprehensive physics-based feature generator while using machine learning to correct structural limitations against satellite observations. We employ a 3D U-Net architecture that processes Noah-MP outputs (soil moisture, latent heat flux, and sensible heat flux) to predict SMAP soil moisture across two contrasting extreme events: a prolonged drought (March-September 2022) and Hurricane Beryl (July 2024) over Texas. When comparing NoahMP-AI with NoahMP, our results demonstrate an increase in R-squared values from -0.7 to 0.5 during drought conditions, while maintaining physical consistency and spatial coherence. The framework's ability to preserve Noah-MP's physical relationships while learning observation-based corrections represents a significant advance in hybrid earth system modeling. This work establishes both a practical tool for operational forecasting and a benchmark for investigating the optimal integration of physics-based understanding with data-driven learning in environmental prediction systems.

physics.ao-ph↗

arXiv2025

UT-GraphCast Hindcast Dataset: A Global AI Forecast Archive from UT Austin for Weather and Climate Applications

The UT GraphCast Hindcast Dataset from 1979 to 2024 is a comprehensive global weather forecast archive generated using the Google DeepMind GraphCast Operational model. Developed by researchers at The University of Texas at Austin under the WCRP umbrella, this dataset provides daily 15 day deterministic forecasts at 00UTC on an approximately 25 km global grid for a 45 year period. GraphCast is a physics informed graph neural network that was trained on ECMWF ERA5 reanalysis. It predicts more than a dozen key atmospheric and surface variables on 37 vertical levels, delivering a full medium range forecast in under one minute on modern hardware.

physics.geo-ph↗

arXiv2025

Enhancing Near Real Time AI-NWP Hurricane Forecasts: Improving Explainability and Performance Through Physics-Based Models and Land Surface Feedback

Hurricane track forecasting remains a significant challenge due to the complex interactions between the atmosphere, land, and ocean. Although AI-based numerical weather prediction models, such as Google Graphcast operation, have significantly improved hurricane track forecasts, they currently function as atmosphere-only models, omitting critical land and ocean interactions. To investigate the impact of land feedback, we conducted independent simulations using the physics-based Hurricane WRF experimental model to assess how soil moisture variations influence storm trajectories. Our results show that land surface conditions significantly alter storm paths, demonstrating the importance of land-atmosphere coupling in hurricane prediction. Although recent advances have introduced AI-based atmosphere-ocean coupled models, a fully functional AI-driven atmosphere-land-ocean model does not yet exist. Our findings suggest that AI-NWP models could be further improved by incorporating land surface interactions, improving both forecast accuracy and explainability. Developing a fully coupled AI-based weather model would mark a critical step toward more reliable and physically consistent hurricane forecasting, with direct applications for disaster preparedness and risk mitigation.

physics.ao-ph↗

arXiv2025

Increasing risk of oppressive heatwaves over India in the future warming

This study examines the increasing frequency of heatwaves, particularly focusing on extreme (high temperature, low humidity) and oppressive (high temperature, high humidity) heatwaves, and their impacts on human mortality. We find that both types of heatwaves are increasing, with oppressive heatwaves showing a faster rate of growth. Importantly, oppressive heatwaves are more strongly correlated with heat-stress-related human deaths than extreme heatwaves, indicating they pose a greater health risk. Using climate model simulations, we project a significant increase in the number of oppressive heatwave days under future warming scenarios. Under 1.5°C global warming, oppressive heatwaves will increase five-fold by the end of the century (2070-2100), relative to the historical period (1975-2005). Under 2°C warming, this increase rises to eight-fold, with an almost two-fold increase in oppressive heatwaves compared to the 1.5°C scenario. Extreme heatwave days, in contrast, remain relatively constant. Limiting warming to 1.5°C could reduce the likelihood of oppressive and extreme heatwaves by 44% and 25%, respectively, compared to a 2°C warming world. These findings highlight the urgent need for adaptation strategies, particularly in densely populated regions, to mitigate the health risks of rising heatwave intensity and frequency.

physics.geo-ph↗

arXiv2024

NDUI+: A fused DMSP-VIIRS based global normalized difference urban index dataset

Urbanization is advancing rapidly, covering less than 2% of Earth's surface yet profoundly influencing global environments and experiencing disproportionate impacts from extreme weather events. Effective urban management and planning require high-resolution, temporally consistent datasets that capture the complexity of urban growth and dynamics. This study presents NDUI+, a novel global urban dataset addressing critical gaps in urban data continuity and quality. NDUI+ integrates data from the Defense Meteorological Satellite Program's Operational Linescan System (DMSP-OLS), VIIRS Nighttime Light, and Landsat 7 NDVI using advanced remote sensing and deep learning techniques. The dataset resolves sensor discontinuity challenges, offering a seamless 30-meter spatial and annual temporal resolution time series from 1999 to the present. NDUI+ demonstrates high precision and granularity, aligning closely with high-resolution satellite data and capturing urban dynamics effectively. The dataset provides valuable insights for urban climate studies, IPCC assessments, and urbanization research, complementing resources like UT-GLOBUS for urban modeling.

physics.soc-ph↗