The initial field has a crucial influence on numerical weather prediction (NWP). Data assimilation (DA) is a reliable method to obtain the initial field of the forecast model. At the same time, data are the carriers of information. Observational data are a concrete representation of information. DA is also the process of sorting observation data, during which entropy gradually decreases. Four-dimensional variational assimilation (4D-Var) is the most popular approach. However, due to the complexity of the physical model, the tangent linear and adjoint models, and other processes, the realization of a 4D-Var system is complicated, and the computational efficiency is expensive. Machine learning (ML) is a method of gaining simulation results by training a large amount of data. It achieves remarkable success in various applications, and operational NWP and DA are no exception. In this work, we synthesize insights and techniques from previous studies to design a pure data-driven 4D-Var implementation framework named ML-4DVAR based on the bilinear neural network (BNN). The framework replaces the traditional physical model with the BNN model for prediction. Moreover, it directly makes use of the ML model obtained from the simulation data to implement the primary process of 4D-Var, including the realization of the short-term forecast process and the tangent linear and adjoint models. We test a strong-constraint 4D-Var system with the Lorenz-96 model, and we compared the traditional 4D-Var system with ML-4DVAR. The experimental results demonstrate that the ML-4DVAR framework can achieve better assimilation results and significantly improve computational efficiency.

The prediction of chaotic time series systems has remained a challenging problem in recent decades. A hybrid method using Hankel Alternative View Of Koopman (HAVOK) analysis and machine learning (HAVOK-ML) is developed to predict chaotic time series. HAVOK-ML simulates the time series by reconstructing a closed linear model so as to achieve the purpose of prediction. It decomposes chaotic dynamics into intermittently forced linear systems by HAVOK analysis and estimates the external intermittently forcing term using machine learning. The prediction performance evaluations confirm that the proposed method has superior forecasting skills compared with existing prediction methods.

Precise center-detection of tropical cyclones (TCs) is critical for dynamic analysis in high resolution model data. The existence of both smaller scale perturbations and larger scale circulations could reduce the accuracy of center positioning. In this study, an objective center-finding algorithm is developed based on a two-dimensional Fourier filter and a vorticity centroid algorithm. This proposed algorithm is able to automatically adjust its parameters according to the scale of the target vortex instead of using artificially prescribed parameters in previous research. What’s more, this new algorithm has been optimized and validated by a hundred idealized vortexes with different sizes and small-scale perturbations. A high-resolution simulation of Typhoon Soudelor (2015) was used to evaluate the performance of the new algorithm, and the proposed objective center-finding algorithm was found able to detect a precise and reliable center.

The Evaporation duct (ED) is a special type of atmospheric layer that can trap electromagnetic waves of certain frequencies, enabling over-the-horizon transmission. Accurate diagnosis and prediction of evaporation duct height have a significant impact on the performance of the radar and communication equipment. This paper proposes a hybrid Temporal Convolutional Network-Gated Recurrent Unit (TCN-GRU) EDH diagnostic model that incorporates the height parameters and reconstructs field campaign data distributions by random sampling. It provides a new research perspective and supplements the shortcomings of conventional approaches, which rely solely on time series data. The model’s performance is evaluated through comparative experiments with traditional physical theory models (BYC, MGB, NPS, and LiuLi2.0) and artificial intelligence models (XGB, LSTM) across multiple field campaign datasets. The results demonstrate that the proposed TCN-GRU EDH diagnostic model has improved EDH diagnostic accuracy.