Nowcasting and early warning of severe convective weather play crucial roles in heavy rainfall warning, flood mitigation, and water resource management. However, achieving effective temporal‐spatial resolution nowcasting is a very challenging task owing to the complex dynamics and chaos. Recently, an increasing amount of research has focused on utilizing deep learning approaches for this task because of their powerful abilities in learning spatiotemporal feature representation in an end‐to‐end manner. In this paper, we present convolutional long short‐term memory with a layer called star‐shape bridge to transfer features across time steps. We build an end‐to‐end trainable model for the nowcasting problem using the radar echo data set. Furthermore, we propose a raining‐oriented loss function inspired by the critical success index and utilize the group normalization technique to refine the convergence performance in optimizing our deep network. Experiments indicate that our model outperforms convolutional long short‐term memory with the cross entropy loss function and the conventional extrapolation method.

A novel tropical cyclone (TC) intensity classification and estimation model (TCICENet) is proposed using infrared geostationary satellite images from the northwest Pacific Ocean basin in combination with a cascading deep convolutional neural network (CNN). The proposed model consists of two CNN network modules: a TC intensity classification (TCIC) module and a TC intensity estimation (TCIE) module. First, the TCIC module is utilized to divide TC intensity into three categories using infrared satellite images. Next, three TCIE models based on the CNN regression network that combine different intensity types of infrared satellite images with the TC best track data are presented. The three TCIE models consider classification error with the TCIC module in order to improve TCIE accuracy. A total of 1001 TCs from 1981-2019 were used to verify the proposed TCICENet model, with 844 TCs from 1981-2013 employed as training samples, 76 TCs from 2014-2016 used as validation samples, and 81 TCs from 2017-2019 used as testing samples. In order to reduce the computation burden of training the TCICENet model, various input image sizes were explored. An image size of 170 × 170 pixels achieved the best performance, with an overall root mean square error of 8.60 kt and a mean absolute error of 6.67 kt compared to the best track.

The properties of aerosols are highly uncertain owing to the complex changes in their composition in different regions. The radiative properties of different aerosol types differ considerably and are vital for studying aerosol regional and/or global climate effects. Traditional aerosol-type identification algorithms, generally based on cluster or empirical analysis methods, are often inaccurate and time-consuming. In response, our study aimed to develop a new aerosol-type classification model using an innovative hybrid algorithm to improve the precision and efficiency of aerosol-type identification. This novel algorithm incorporates an optical database, constructed using the Mie scattering model, and employs a random forest algorithm to classify different aerosol types based on the optical data from the database. The complex refractive index was used as a baseline to assess the performance of our hybrid algorithm against the traditional Gaussian kernel density clustering method for aerosol-type identification. The hybrid algorithm demonstrated impressive consistency rates of 90 %, 85 %, 84 %, 84 %, and 100 % for dust, mixed-coarse (mixed, course-mode aerosol), mixed-fine (mixed, fine-mode aerosol), urban/industrial, and biomass burning aerosols, respectively. Moreover, it achieved remarkable precision, with evaluation metric indexes for micro-precision, micro-recall, micro-F1-score, and accuracy of 95 %, 89 %, 91 %, and 89 %, respectively. Lastly, a global map of aerosol types was generated using the new hybrid algorithm to characterize aerosol types across the five continents. This study, utilizing a novel approach for the classification of aerosol, will help improve the accuracy of aerosol inversion and determine the sources of aerosol pollution.

Weather radar echo extrapolation, which predicts future echoes based on historical observations, is one of the complicated spatial–temporal sequence prediction tasks and plays a prominent role in severe convection and precipitation nowcasting. However, existing extrapolation methods mainly focus on a defective echo-motion extrapolation paradigm based on finite observational dynamics, neglecting that the actual echo sequence has a more complicated evolution process that contains both nonlinear motions and the lifecycle from initiation to decay, resulting in poor prediction precision and limited application ability. To complement this paradigm, we propose to incorporate a novel long-term evolution regularity memory (LERM) module into the network, which can memorize long-term echo-evolution regularities during training and be recalled for guiding extrapolation. Moreover, to resolve the blurry prediction problem and improve forecast accuracy, we also adopt a coarse–fine hierarchical extrapolation strategy and compositive loss function. We separate the extrapolation task into coarse and fine two levels which can reduce the downsampling loss and retain echo fine details. Except for the average reconstruction loss, we additionally employ adversarial loss and perceptual similarity loss to further improve the visual quality. Experimental results from two real radar echo datasets demonstrate the effectiveness of our methodology and show that it can accurately extrapolate the echo evolution while ensuring the echo details are realistic enough, even for the long term. Our method can further be improved in the future by integrating multimodal radar variables or introducing certain domain prior knowledge of physical mechanisms. It can also be applied to other spatial–temporal sequence prediction tasks, such as the prediction of satellite cloud images and wind field figures.

By utilizing physical models of the atmosphere collected from the current weather conditions, the numerical weather prediction model developed by the European Centre for Medium-range Weather Forecasts (ECMWF) can provide the indicators of severe weather such as heavy precipitation for an early-warning system. However, the performance of precipitation forecasts from ECMWF often suffers from considerable prediction biases due to the high complexity and uncertainty for the formation of precipitation. The bias correcting on precipitation (BCoP) was thus utilized for correcting these biases via forecasting variables, including the historical observations and variables of precipitation, and these variables, as predictors, from ECMWF are highly relevant to precipitation. The existing BCoP methods, such as model output statistics and ordinal boosting autoencoder, do not take advantage of both spatiotemporal (ST) dependencies of precipitation and scales of related predictors that can change with different precipitation. We propose an end-to-end deep-learning BCoP model, called the ST scale adaptive selection (SSAS) model, to automatically select the ST scales of the predictors via ST Scale-Selection Modules (S3M/TS2M) for acquiring the optimal high-level ST representations. Qualitative and quantitative experiments carried out on two benchmark datasets indicate that SSAS can achieve state-of-the-art performance, compared with 11 published BCoP methods, especially on heavy precipitation.

This paper proposes an infrared and water vapor channel satellite image fusion model (TCCL-DenseFuse) based on DenseNet. The infrared channel satellite image reflects the ground and cloud top infrared radiation or the distribution of temperature, and the water vapor channel satellite image reflects the spatial distribution of water vapor in the upper atmosphere. Studies have shown that infrared brightness temperature gradient and water vapor transport are closely related to TC generation and evolution. In order to facilitate the fusion image obtained by the proposed fusion model to have a positive effect on tropical cyclone monitoring and warning, the brightness temperature gradient and multi-scale structural similarity in the satellite image are used to construct loss function of the proposed TCCL-DenseFuse model. Quality of the fused images are evaluated by seven objective quantitative indicators. In order to further verify the real application value of the proposed TCCL-DenseFuse model, fused images are also used to TC center location. Experimental results show that the proposed TCCL-DenseFuse fused satellite image not only contains rich information from both infrared and water vapor channels but also improves the accuracy of TC center positioning. The comprehensive performance of the proposed fusion model has certain advantages compared with similar fusion methods and can provide a reference for typhoon prevention and disaster mitigation.