Journal of Atmospheric Science Research

The Warm-Sector Thunderstorms Triggered by Mesoscale Boundary-Layer Convergence over the DPRK

2026-03-26T11:45:24+08:00

Warm-sector thunderstorms (WSTs), characterized by weak synoptic forcing and extreme precipitation rates, pose a major global forecasting challenge. This study investigates the mesoscale processes initiating WSTs over the complex terrain of the Democratic People's Republic of Korea (DPRK), a region where triggering mechanisms remain poorly understood. We analyze three extreme rainfall events (Hoichang 2016, Unpa 2017, Pyongyang 2018), each producing rainfall rates exceeding 60 mm h⁻1 under the weak forcing typical of the northwestern periphery of the West Pacific Subtropical High. While operational global models failed to predict these events, the convection-permitting WRF model skillfully replicated the initiating mechanisms and subsequent convection. Model performance was quantitatively assessed using multiple verification metrics, including Probability of Detection (POD), False Alarm Ratio (FAR), Bias, and Critical Success Index (CSI). Integrated analysis of observations and high-resolution (3 km) Weather Research and Forecasting (WRF) model simulations reveals a consistent trigger: mesoscale boundary-layer convergence lines. These zones formed through the interaction of synoptic southwesterlies with localized, terrain-modulated flows and were collocated with horizontal moisture gradients. Crucially, the three-dimensional structure of Convective Available Potential Energy (CAPE) manifested as narrow, vertical towers of high instability, delineating regions of deep convection initiation 2–4 h in advance. A pre-convective, deep moist layer (relative humidity >80% in the 850–700 hPa layer) was identified as a necessary precondition. This study establishes terrain-forced boundary-layer convergence as a primary trigger for WSTs over the DPRK, providing a valuable framework for improving prediction in other monsoonal regions with complex topography.

Hybrid Deep Learning and Bayesian Framework for Long-Term Fog Forecasting: A Case Study at Lucknow Airport in the Indo-Gangetic Plains

2026-04-20T15:34:58+08:00

Fog significantly affects aviation, transportation, agriculture, and energy in North India during winter. Accurate prediction is difficult due to nonlinear meteorological interactions and uncertainty. Traditional models provide point estimates without uncertainty, while Bayesian methods offer probabilistic outputs but struggle to capture complex temporal dependencies in atmospheric time-series data effectively. To address these limitations, this study proposes a hybrid long-term forecasting framework that integrates a three-layer Bidirectional Long Short-Term Memory (BiLSTM) network with a Modified Bayesian Beta Regression model. The model is trained and evaluated using 24 years (2000–2023) of meteorological observations from Lucknow Airport, representing a fog-prone region within the Indo-Gangetic Plains. Fog behaviour was represented through a normalized Fog Index (0–1) that incorporates both intensity and persistence, offering a more stable and informative forecast target than raw visibility. The BiLSTM component captures temporal dependencies and produces accurate point forecasts, achieving a root mean square error (RMSE) of 0.125 for a three-day prediction horizon. The Bayesian layer enhances reliability by generating calibrated uncertainty intervals. The resulting model achieved an average interval width of approximately 0.59 and a prediction interval coverage close to 80%, effectively representing both aleatoric and epistemic uncertainty. Additional techniques, including Min-Max normalization, sequence windowing, Fast Fourier Transform (FFT) feature augmentation, learning-rate scheduling, and Monte Carlo dropout, improved generalization and model stability. The proposed hybrid framework outperforms standalone models and shows strong potential for operational fog forecasting, offering both accurate predictions and uncertainty-aware confidence estimates for aviation, transportation, and early-warning systems.

Diagnosing Seasonal Structures and Short-Term Forecasting of Tropospheric Ozone in a Tropical City Using Singular Spectrum Analysis and Linear Recurrent Formula

2026-04-24T09:20:44+08:00

Tropospheric ozone (O₃) is a secondary pollutant whose variability in tropical urban environments is strongly controlled by seasonal meteorology, photochemistry, and episodic emissions such as biomass burning. This study applies Singular Spectrum Analysis (SSA) combined with the Linear Recurrent Formula (LRF) to analyze and forecast daily tropospheric ozone in Campo Grande, Brazil, using SISAM/INPE satellite data from 2000 to 2018. In contrast to previous SSA-based applications, this work introduces a systematic evaluation of embedding window length (L = 30, 60, and 90) to assess the robustness of the decomposition and component separability. In addition, the spectral consistency of reconstructed components is examined to support the identification of dominant temporal modes. For forecasting, a strict out-of-sample framework is adopted, using 2000–2017 for training and 2018 for independent validation, ensuring no information leakage. The LRF model achieved RMSE = 0.79 ppb, MAE = 0.64 ppb, and MAPE = 3.8%, outperforming persistence and seasonal naïve benchmarks. Results indicate that ozone variability is predominantly seasonal, with weak long-term trends and relevant intra-seasonal fluctuations. The proposed framework provides a transparent, computationally efficient, and reproducible approach for diagnosing and forecasting ozone variability in tropical environments.

Correction to: Deep Learning-based Flood Risk Prediction for Climate Resilience Planning in Malawi

2026-04-10T11:39:39+08:00

Data Availability Statement Correction

In the originally published version of this article, the Data Availability Statement did not provide sufficient detail regarding the specific data sources and access information.

To improve transparency and reproducibility, the Data Availability Statement has been updated as follows:

The data supporting the findings of this study are derived from a combination of publicly available datasets and primary research data.

Precipitation (rainfall) data were obtained from the NASA POWER Data Access Viewer, available at: https://power.larc.nasa.gov/data-access-viewer/.

Sea Surface Temperature (SST) data were retrieved via Google Earth Engine from the NOAA Optimum Interpolation Sea Surface Temperature Climate Data Record (NOAA/CDR/OISST/V2.1), covering the period 1990–2024.

Flood vulnerability data were derived from primary survey data collected by the corresponding author as part of doctoral research on flood vulnerability assessment in Malawi, supplemented by regional environmental and hydrological observational data.

The primary survey dataset is not fully publicly available due to ethical and data-sharing considerations; however, it can be made available from the authors upon reasonable request, subject to applicable conditions.

All publicly available datasets can be accessed through the links provided above and are sufficient to support the reproducibility of the study.

This correction does not affect the results or conclusions of the article. The original publication has also been updated.

DOI of original article: https://doi.org/10.30564/jasr.v8i2.10377

Correction Date: 13 April 2026

Refer to: Mwalwimba, I. K., Kalonjeka, B., Msadala, V., Katonda, V., Chisenga, C., Ngongondo, C., & Manda, M. (2025). Deep Learning-based Flood Risk Prediction for Climate Resilience Planning in Malawi. Journal of Atmospheric Science Research, 8(2), 37–50.