Estimating tropical cyclone (TC) intensity is the first step in the processes of monitoring and predicting destructive TC disasters. Due to the dilemma of meteorological methods, accurate estimation of TC intensity is a long-term challenge. In recent years, while deep learning methods have been applied to TC intensity estimation, most of them fail to make full use of multi-channel satellite imageries to consider three-dimensional structure of TC. In this paper, we propose a novel deep learning model (3DAttentionTCNet) to overcome this shortcoming. The model can automatically extract three-dimensional environment information related to TC intensity from multi-channel satellite observation imageries such as infrared, water vapor and microwave rainrate satellite imageries by three-dimensional convolution. In addition, we employ the convolutional block attention module to simulate visual attention for strengthening the model’s attention to core cloud structure and important channels. The experimental results show that the root-mean-square error (RMSE) of the proposed model is 9.48 kts, which is improved by 25% compared to that of the advanced Dvorak technique (ADT) and by 9.2% over that of the traditional deep learning method of TC intensity estimation.

Tropical cyclones (TC) are among the most deadly and damaging natural disasters in coastal areas worldwide. Traditional forecasting methods face challenges as they neglect crucial spatial information related to intensity changes and require substantial human and material resources. Moreover, current deep learning approaches often rely on reanalysis data from observations far from land, making them challenging to acquire and operationalize. In response to these issues, the paper introduces the Tropical Cyclone Intensity Prediction Network (TCIP-Net), which, while maintaining interpretability, successfully extracts rich convective structural information from the infrared channel of satellite imagery. We present the spatio-temporal evolution trajectory of TC radial structural information through Hovmöller diagrams. Additionally, we construct a subnetwork with one backbone convolution and four branch convolution operations to extract asymmetric information of TC structure. The convective core reveals the distribution of convective systems around the eye, aiding in targeted attention to convective information in infrared imagery. The model aims to quantitatively explain the contributions of satellite imagery (infrared, microwave), convective structure, and key physical factors to TC intensity prediction task. We utilize multiple TC cases to assess and validate the applicability and effectiveness of the model. The results indicate that TCIP-Net achieved good performance. This approach provides practical guidance for predicting TC intensity using advanced artificial intelligence-based methods and is expected to complement operational models.

Global navigation satellite system (GNSS) timing modules from 5G base stations enable precipitable water vapor (PWV) retrieval, offering a promising approach to overcome the limitations of conventional meteorological stations in atmospheric monitoring. In this study, we retrieved PWV values (referred to as 5G-PWV) based on GNSS observations from 5G base stations in Zhoushan City, Zhejiang Province, China, using data from September to November 2024. The results were compared with ERA5 reanalysis data (ERA5-PWV) and observations from a dedicated GNSS station in Zhoushan (ZJZS-PWV). The 5G PWV data showed consistent temporal variability with both ERA5-PWV and ZJZS-PWV. After applying a bi-weight quality control algorithm, comparisons against ERA5-PWV revealed a mean bias of −1.07 mm, a root mean square error of 8.03 mm, and a correlation coefficient of 0.828 for 5G-PWV. We further conducted assimilation and forecast experiments using the CMA MESO (China Meteorological Administration Mesoscale Model) coupled with its three-dimensional variational data assimilation system for two typhoon cases: Bebinca and Krathon. Validation of precipitation forecasts—with and without assimilation of 5G PWV—showed that the assimilation of 5G-PWV improved the threat score (TS). After assimilating 5G-PWV, both false alarms and missed alarms across different precipitation intensity thresholds were slightly reduced. The improved forecast performance is attributed to effective adjustments in the initial humidity field through the assimilation of 5G-PWV, as well as subsequent enhancements in the model's dynamic conditions via precipitation feedback. Our study demonstrates that GNSS derived PWV from 5G base stations has clear potential to improve typhoon-related precipitation forecasting.

Atmospheric temperature and humidity profiles are essential for numerical weather prediction and severe weather monitoring. To effectively utilize data from the Geostationary Interferometric Infrared Sounder (GIIRS) onboard the FY-4 satellite, this study proposes a retrieval method based on a radial basis function (RBF) neural network, which integrates numerical model background profiles with GIIRS simulated radiance errors to construct a mapping from these two inputs to background profile errors. A channel selection strategy is developed using correlations between background errors and radiance errors to identify channels sensitive to temperature and humidity variations at different pressure levels. Experiments are conducted using data from land stations in Zhejiang Province, China, from August to December 2024, including 829 clear-sky and 2109 cloudy profiles. Under clear-sky conditions, the method reduces temperature and humidity root mean square error (RMSE) by approximately 39% and 22.3% compared to background profiles. Under cloudy conditions, despite severe radiation interference, RMSE reductions of 38.5% for temperature and 15.3% for humidity are achieved, with notable improvements below 900 hPa and above 750 hPa for humidity. Compared with the multilayer perceptron (MLP) method, RBF shows superior performance under all test conditions, especially in cloudy-sky humidity retrieval. The proposed approach provides an effective, physically constrained framework for operational GIIRS data application in temperature and humidity retrieval.

This study systematically evaluates the performance of GPM_IMERG and GSMaP_Gauge hourly precipitation products in typhoon rainfall over Mainland China. Using hourly gauge observations from the China Meteorological Administration (CMA)’s national basic meteorological stations as reference, 32 landfalling typhoons during 2021–2025 are analyzed. A multi-layered evaluation framework is established based on 50 km annular stratification from 0 to 500 km relative to typhoon centers, multiple statistical metrics, and dual thresholds for light rain and extreme precipitation. Results indicate systematic underestimation of typhoon rainfall by both products, with GSMaP_Gauge exhibiting more severe negative bias that intensifies nonlinearly with increasing rainfall intensity. Spatially, widespread overestimation occurs in North China, while underestimation dominates elsewhere, with large negative biases concentrated in high-observation regions. Monthly variations show predominantly negative deviations across most months, with GSMaP_Gauge demonstrating persistent negative anomalies except for sporadic positive outliers. Regarding precipitation detection capability, both products perform adequately for light rain, but their capability to capture extreme precipitation remains rather limited, as evidenced by sharply declining Critical Success Index (CSI) across all distance ranges and omission of over 60% extreme precipitation events. GPM_IMERG shows only sporadic high CSI values in the inner-core region during June and October. Error distributions exhibit significant spatiotemporal non-stationarity: errors attenuate markedly within 0–100 km of typhoon centers; seasonally, June and September show higher correlation coefficients but larger RMSE, whereas August presents lower correlation yet smaller errors; diurnally, the 0–50 km zone displays a “three-peak–two-valley” pattern with error maxima in the afternoon, early morning, and evening. In conclusion, both products estimate light typhoon precipitation with reasonable accuracy but still have considerable room for improvement in estimating heavy and extreme rainfall. Dynamic error models based on three-dimensional stratification of distance–season–diurnal phase, coupled with bias correction, are imperative before their application to hydrometeorological modeling, disaster investigation, and climate research.