arXiv subjects

ChangJae Lee

Publications and source records attributed to ChangJae Lee.

4 recordsLinked to original sources

arXiv2026

Event-Aware Loss Design for Forecasting of Convective Precipitation and Lightning

Accurate forecasting of high-impact weather, specifically extreme precipitation and lightning, remains a significant challenge in numerical weather prediction (NWP) due to the complexity of atmospheric microphysics. While deep-learning models have shown promise in large-scale forecasting, they often suffer from systematic under-prediction of rare, high-intensity events and localized convective showers when optimized with conventional loss functions like Mean Squared Error (MSE). This study proposes an Event-Aware multi-task deep-learning post-processing framework designed to improve the representation of convective processes by leveraging lightning observations. The model jointly predicts precipitation amount, rainfall probability, and lightning occurrence using a shared-backbone Patch-cGAN (Conditional Generative Adversarial Network) architecture. To address the rare event problem, we introduce a lightning-informed loss-weighting strategy that element-wisely multiplies the MSE component by a spatial weight map derived from observed lightning strikes, forcing the model to prioritize accuracy in convective regions during training. Evaluations conducted over the Korean Peninsula during the 2025 Summer demonstrate that our framework outperforms standard AI benchmarks and conventional NWP models, particularly at intense rainfall thresholds (40 mm/6 h). Furthermore, the model exhibits superior skill in predicting lightning compared to conventional lightning parameterization and instability-index-based methods. These results indicate that integrating physical event indicators into the loss formulation effectively guides models to learn the meteorological signatures of deep convection, offering a pathway toward more reliable extreme weather forecasting.

physics.ao-ph↗

arXiv2026

Enhancing the QA Model through a Multi-domain Debiasing Framework

Question-answering (QA) models have advanced significantly in machine reading comprehension but often exhibit biases that hinder their performance, particularly with complex queries in adversarial conditions. This study evaluates the ELECTRA-small model on the Stanford Question Answering Dataset (SQuAD) v1.1 and adversarial datasets AddSent and AddOneSent. By identifying errors related to lexical bias, numerical reasoning, and entity recognition, we develop a multi-domain debiasing framework incorporating knowledge distillation, debiasing techniques, and domain expansion. Our results demonstrate up to 2.6 percentage point improvements in Exact Match (EM) and F1 scores across all test sets, with gains in adversarial contexts. These findings highlight the potential of targeted bias mitigation strategies to enhance the robustness and reliability of natural language understanding systems.

cs.CL↗

arXiv2025

Exploring Multimodal AI Reasoning for Meteorological Forecasting from Skew-T Diagrams

Forecasting from atmospheric soundings is a fundamental task in operational meteorology, often requiring structured visual reasoning over Skew-T log-P diagrams by human forecasters. While recent advances in Vision-Language Models (VLMs) have shown promise in other scientific domains, their application to meteorological diagram interpretation remains largely unexplored. In this study, we present a lightweight AI assistant that interprets Skew-T diagrams using a small language model (LM) and a small VLM fine-tuned to emulate human forecasters. Using a curriculum learning framework, we first train the models to identify key atmospheric features from diagrams through visual question answering, followed by chain-of-thought reasoning tasks that estimate precipitation probability based on the derived visual groundings. Model inputs include either textual summaries or generated Skew-T diagrams derived from operational Numerical Weather Prediction (NWP) forecasts, paired with three-hour precipitation observations from South Korea's Auto Weather Stations network. Evaluation results demonstrate that the fine-tuned VLM achieves skill comparable to an operational NWP model, despite relying solely on static atmospheric profiles. Ablation studies reveal that visual grounding and reasoning supervision are critical for performance, while attention map analysis confirms that the model learns to focus on relevant meteorological features. These findings highlight the potential of compact, interpretable multimodal models to support weather forecasting tasks. The approach offers a computationally efficient alternative to large-scale systems, and future work could extend it to more complex applications.

physics.ao-ph↗

arXiv2025

Improving Post-Processing for Quantitative Precipitation Forecasting Using Deep Learning: Learning Precipitation Physics from High-Resolution Observations

Accurate quantitative precipitation forecasting (QPF) remains one of the main challenges in numerical weather prediction (NWP), primarily due to the difficulty of representing the full complexity of atmospheric microphysics through parameterization schemes. This study introduces a deep learning-based post-processing model, DL-QPF, which diagnoses precipitation fields from meteorological forecasts by learning directly from high-resolution radar estimates precipitation. The DL-QPF model is constructed using a Patch-conditional Generative Adversarial Network (Patch-cGAN) architecture combined with a U-Net generator and a discriminator. The generator learns meteorological features relevant to precipitation, while the adversarial loss from the discriminator encourages the generation of realistic rainfall patterns and distributions. Training is performed on three years of warm-season data over the Korean Peninsula, with input variables derived from ECMWF's Integrated Forecasting System High-Resolution forecast (IFS-HRES). Model verification is conducted against multiple reference models, including global (IFS-HRES, KIM), regional (KIM-Regional, KIM-LENS), and AI-based (GraphCast) forecasts. Verification across multiple rainfall thresholds shows that DL-QPF achieves a frequency bias near one and superior success ratios. Particularly for heavy and intense rainfall events, DL-QPF outperforms both conventional NWP and an AI model, demonstrating improved skill in capturing high-intensity precipitation. This study highlights the potential of observational data-driven deep learning approaches in post-processing QPF. By directly learning from observations, DL-QPF reduces systematic biases and enhances the realism of forecasted rainfall distributions. These results demonstrate the model's potential to enhance QPF realism.

physics.ao-ph↗