arXiv subjects

Dhamma Kimpara

Publications and source records attributed to Dhamma Kimpara.

9 recordsLinked to original sources

arXiv2026

GOES-East full-disk AI nowcasting of cloud evolution in observation space

Clouds affect aviation, solar energy, remote sensing, and storm prediction, yet they remain among the hardest atmospheric features to forecast, particularly at convective scales. Because clouds are shaped by processes spanning a wide range of space and time scales, numerical weather prediction, extrapolation methods, and existing machine learning (ML) approaches are each limited by some combination of accuracy, domain size, and temporal resolution. We present DOP+, an ML approach for clouds that forecasts GOES-East full-disk infrared brightness temperatures by extending direct observation prediction (DOP) with conditioning on meteorological fields. The domain covers tropical, midlatitude, and marine regimes across $\sim 10^8 ~ km^2$, roughly a fifth of Earth's surface. DOP+ forecasts cloud evolution at 10-minute resolution and outperforms persistence, synoptic-scale NWP, and a pure DOP baseline across 0-6 h lead times in fractions skill score and mean absolute error skill score. Convective structure is retained out to 2-3 h. DOP+ thus achieves a state-of-the-art combination of accuracy, temporal resolution, and spatial coverage. Our work lays the foundation for fully global cloud nowcasting at convective timescales.

physics.ao-ph↗

arXiv2025

Controllable Probabilistic Forecasting with Stochastic Decomposition Layers

AI weather prediction ensembles with latent noise injection and optimized with the continuous ranked probability score (CRPS) have produced both accurate and well-calibrated predictions with far less computational cost compared with diffusion-based methods. However, current CRPS ensemble approaches vary in their training strategies and noise injection mechanisms, with most injecting noise globally throughout the network via conditional normalization. This structure increases training expense and limits the physical interpretability of the stochastic perturbations. We introduce Stochastic Decomposition Layers (SDL) for converting deterministic machine learning weather models into probabilistic ensemble systems. Adapted from StyleGAN's hierarchical noise injection, SDL applies learned perturbations at three decoder scales through latent-driven modulation, per-pixel noise, and channel scaling. When applied to WXFormer via transfer learning, SDL requires less than 2\% of the computational cost needed to train the baseline model. Each ensemble member is generated from a compact latent tensor (5 MB), enabling perfect reproducibility and post-inference spread adjustment through latent rescaling. Evaluation on 2022 ERA5 reanalysis shows ensembles with spread-skill ratios approaching unity and rank histograms that progressively flatten toward uniformity through medium-range forecasts, achieving calibration competitive with operational IFS-ENS. Multi-scale experiments reveal hierarchical uncertainty: coarse layers modulate synoptic patterns while fine layers control mesoscale variability. The explicit latent parameterization provides interpretable uncertainty quantification for operational forecasting and climate applications.

cs.LG↗

arXiv2025

Winter Precipitation Type Diagnosis and Uncertainty Quantification with a Physically Consistent Machine Learning Method

Accurately forecasting winter precipitation type and its transitions is critical for high-impact decision making. However, existing methods struggle in thermodynamically ambiguous regimes, and most do not quantify forecast uncertainty from a single model run. We developed an evidential neural network that predicts calibrated probabilities for four winter precipitation types (rain, snow, freezing rain, and ice pellets) along with epistemic uncertainty estimates at the computational cost of a standard neural network. The model was trained on quality-controlled and curated observations from the crowd-sourced mPING dataset paired with vertical thermodynamic profiles from the NOAA Rapid Refresh model analyses. Rigorous physical quality control removed thermodynamically implausible reports. Bulk evaluation against held-out mPING observations from June 2020 through June 2022 shows the ML model outperforms area-based deterministic methods in success ratio for freezing rain and ice pellets while maintaining comparable or better performance for rain and snow. A reduced freezing rain probability of detection reflects genuinely ambiguous thermodynamic environments rather than a uniform model deficiency and is more robustly represented through the full probability distribution than through the dominant predicted class alone. Thermodynamic regime analysis demonstrates that model prediction errors are physically structured and concentrated in interpretable regions of diagnostic space consistent with the known difficulty of freezing rain and ice pellet discrimination. We further demonstrate the model's physical consistency and operational utility through two contrasting mid-western U.S. winter storm case studies and an interactive visualization tool that enables dynamic interrogation of model predictions and uncertainty in real time.

physics.ao-ph↗

arXiv2025

Consistency Conditions for Differentiable Surrogate Losses

The statistical consistency of surrogate losses for discrete prediction tasks is often checked via the condition of calibration. However, directly verifying calibration can be arduous. Recent work shows that for polyhedral surrogates, a less arduous condition, indirect elicitation (IE), is still equivalent to calibration. We give the first results of this type for non-polyhedral surrogates, specifically the class of convex differentiable losses. We first prove that under mild conditions, IE and calibration are equivalent for one-dimensional losses in this class. We construct a counter-example that shows that this equivalence fails in higher dimensions. This motivates the introduction of strong IE, a strengthened form of IE that is equally easy to verify. We establish that strong IE implies calibration for differentiable surrogates and is both necessary and sufficient for strongly convex, differentiable surrogates. Finally, we apply these results to a range of problems to demonstrate the power of IE and strong IE for designing and analyzing consistent differentiable surrogates.

cs.LG↗

arXiv2025

CAMulator: Fast Emulation of the Community Atmosphere Model

We introduce CAMulator version 1, an auto-regressive machine-learned (ML) emulator of the Community Atmosphere Model version 6 (CAM6) that simulates the next atmospheric state given the prescribed sea surface temperatures and incoming solar radiation. CAMulator explicitly conserves global dry air mass, moisture, and total atmospheric energy while remaining numerically stable over indefinite climate integrations. It successfully reproduces the annual CAM6 climatology and key modes of climate variability, including the El Ni\~no-Southern Oscillation, the North Atlantic Oscillation, and the Pacific-North American pattern, with slightly muted variability. When forced with sea surface temperature (SST) outside the training distribution, CAMulator exhibits a systematic cold bias in high-latitude regions, particularly in boreal winter, likely due to the absence of interactive land and sea ice. Nonetheless, CAMulator achieves these results with a 350 times speedup over CAM6, making it an efficient alternative for generating large ensembles.

physics.ao-ph↗

arXiv2024

Community Research Earth Digital Intelligence Twin (CREDIT)

Recent advancements in artificial intelligence (AI) for numerical weather prediction (NWP) have significantly transformed atmospheric modeling. AI NWP models outperform traditional physics-based systems, such as the Integrated Forecast System (IFS), across several global metrics while requiring fewer computational resources. However, existing AI NWP models face limitations related to training datasets and timestep choices, often resulting in artifacts that reduce model performance. To address these challenges, we introduce the Community Research Earth Digital Intelligence Twin (CREDIT) framework, developed at NSF NCAR. CREDIT provides a flexible, scalable, and user-friendly platform for training and deploying AI-based atmospheric models on high-performance computing systems. It offers an end-to-end pipeline for data preprocessing, model training, and evaluation, democratizing access to advanced AI NWP capabilities. We demonstrate CREDIT's potential through WXFormer, a novel deterministic vision transformer designed to predict atmospheric states autoregressively, addressing common AI NWP issues like compounding error growth with techniques such as spectral normalization, padding, and multi-step training. Additionally, to illustrate CREDIT's flexibility and state-of-the-art model comparisons, we train the FUXI architecture within this framework. Our findings show that both FUXI and WXFormer, trained on six-hourly ERA5 hybrid sigma-pressure levels, generally outperform IFS HRES in 10-day forecasts, offering potential improvements in efficiency and forecast accuracy. CREDIT's modular design enables researchers to explore various models, datasets, and training configurations, fostering innovation within the scientific community.

cs.AI↗

arXiv2024

Trading off Consistency and Dimensionality of Convex Surrogates for the Mode

In multiclass classification over $n$ outcomes, the outcomes must be embedded into the reals with dimension at least $n-1$ in order to design a consistent surrogate loss that leads to the "correct" classification, regardless of the data distribution. For large $n$, such as in information retrieval and structured prediction tasks, optimizing a surrogate in $n-1$ dimensions is often intractable. We investigate ways to trade off surrogate loss dimension, the number of problem instances, and restricting the region of consistency in the simplex for multiclass classification. Following past work, we examine an intuitive embedding procedure that maps outcomes into the vertices of convex polytopes in a low-dimensional surrogate space. We show that full-dimensional subsets of the simplex exist around each point mass distribution for which consistency holds, but also, with less than $n-1$ dimensions, there exist distributions for which a phenomenon called hallucination occurs, which is when the optimal report under the surrogate loss is an outcome with zero probability. Looking towards application, we derive a result to check if consistency holds under a given polytope embedding and low-noise assumption, providing insight into when to use a particular embedding. We provide examples of embedding $n = 2^{d}$ outcomes into the $d$-dimensional unit cube and $n = d!$ outcomes into the $d$-dimensional permutahedron under low-noise assumptions. Finally, we demonstrate that with multiple problem instances, we can learn the mode with $\frac{n}{2}$ dimensions over the whole simplex.

cs.LG↗

arXiv2023

Evidential Deep Learning: Enhancing Predictive Uncertainty Estimation for Earth System Science Applications

Robust quantification of predictive uncertainty is critical for understanding factors that drive weather and climate outcomes. Ensembles provide predictive uncertainty estimates and can be decomposed physically, but both physics and machine learning ensembles are computationally expensive. Parametric deep learning can estimate uncertainty with one model by predicting the parameters of a probability distribution but do not account for epistemic uncertainty.. Evidential deep learning, a technique that extends parametric deep learning to higher-order distributions, can account for both aleatoric and epistemic uncertainty with one model. This study compares the uncertainty derived from evidential neural networks to those obtained from ensembles. Through applications of classification of winter precipitation type and regression of surface layer fluxes, we show evidential deep learning models attaining predictive accuracy rivaling standard methods, while robustly quantifying both sources of uncertainty. We evaluate the uncertainty in terms of how well the predictions are calibrated and how well the uncertainty correlates with prediction error. Analyses of uncertainty in the context of the inputs reveal sensitivities to underlying meteorological processes, facilitating interpretation of the models. The conceptual simplicity, interpretability, and computational efficiency of evidential neural networks make them highly extensible, offering a promising approach for reliable and practical uncertainty quantification in Earth system science modeling. In order to encourage broader adoption of evidential deep learning in Earth System Science, we have developed a new Python package, MILES-GUESS (https://github.com/ai2es/miles-guess), that enables users to train and evaluate both evidential and ensemble deep learning.

cs.LG↗

arXiv2022

Proper losses for discrete generative models

We initiate the study of proper losses for evaluating generative models in the discrete setting. Unlike traditional proper losses, we treat both the generative model and the target distribution as black-boxes, only assuming ability to draw i.i.d. samples. We define a loss to be black-box proper if the generative distribution that minimizes expected loss is equal to the target distribution. Using techniques from statistical estimation theory, we give a general construction and characterization of black-box proper losses: they must take a polynomial form, and the number of draws from the model and target distribution must exceed the degree of the polynomial. The characterization rules out a loss whose expectation is the cross-entropy between the target distribution and the model. By extending the construction to arbitrary sampling schemes such as Poisson sampling, however, we show that one can construct such a loss.

cs.LG↗