Accurate short-term precipitation forecast is extremely important for urban flood warning and natural disaster prevention. In this paper, we present an innovative deep learning model named ISA-PredRNN (improved self-attention PredRNN) for precipitation nowcasting based on radar echoes on the basis of the advanced PredRNN-V2. We introduce the self-attention mechanism and the long-term memory state into the model and design a new set of gating mechanisms. To better capture different intensities of precipitation, the loss function with weights was designed. We further train the model using a combination of reverse scheduled sampling and scheduled sampling to learn the long-term dynamics from the radar echo sequences. Experimental results show that the new model (ISA-PredRNN) can effectively extract the spatiotemporal features of radar echo maps and obtain radar echo prediction results with a small gap from the ground truths. From the comparison with the other six models, the new ISA-PredRNN model has the most accurate prediction results with a critical success index (CSI) of 0.7001, 0.5812 and 0.3052 under the radar echo thresholds of 10 dBZ, 20 dBZ and 30 dBZ, respectively.

Understanding the influence of the Antarctic on the global climate is crucial for the prediction of global warming. However, due to very few observation sites, it is difficult to reconstruct the rational spatial pattern by filling in the missing values from the limited site observations. To tackle this challenge, regional spatial gap-filling methods, such as Kriging and inverse distance weighted (IDW), are regularly used in geoscience. Nevertheless, the reconstructing credibility of these methods is undesirable when the spatial structure has massive missing pieces. Inspired by image inpainting, we propose a novel deep learning method that demonstrates a good effect by embedding the physics-aware initialization of deep learning methods for rapid learning and capturing the spatial dependence for the high-fidelity imputation of missing areas. We create the benchmark dataset that artificially masks the Antarctic region with ratios of 30%, 50% and 70%. The reconstructing monthly mean surface temperature using the deep learning image inpainting method RFR (Recurrent Feature Reasoning) exhibits an average of 63% and 71% improvement of accuracy over Kriging and IDW under different missing rates. With regard to wind speed, there are still 36% and 50% improvements. In particular, the achieved improvement is even better for the larger missing ratio, such as under the 70% missing rate, where the accuracy of RFR is 68% and 74% higher than Kriging and IDW for temperature and also 38% and 46% higher for wind speed. In addition, the PI-RFR (Physics-Informed Recurrent Feature Reasoning) method we proposed is initialized using the spatial pattern data simulated by the numerical climate model instead of the unified average. Compared with RFR, PI-RFR has an average accuracy improvement of 10% for temperature and 9% for wind speed. When applied to reconstruct the spatial pattern based on the Antarctic site observations, where the missing rate is over 90%, the proposed method exhibits more spatial characteristics than Kriging and IDW.

Meteorological prediction is crucial for various sectors, including agriculture, navigation, daily life, disaster prevention, and scientific research. However, traditional numerical weather prediction (NWP) models are constrained by their high computational resource requirements, while the accuracy of deep learning models remains suboptimal. In response to these challenges, we propose a novel deep learning-based model, the Spatiotemporal Fusion Model (STFM), designed to enhance the accuracy of meteorological predictions. Our model leverages Fifth-Generation ECMWF Reanalysis (ERA5) data and introduces two key components: a spatiotemporal encoder module and a spatiotemporal fusion module. The spatiotemporal encoder integrates the strengths of convolutional neural networks (CNNs) and recurrent neural networks (RNNs), effectively capturing both spatial and temporal dependencies. Meanwhile, the spatiotemporal fusion module employs a dual attention mechanism, decomposing spatial attention into global static attention and channel dynamic attention. This approach ensures comprehensive extraction of spatial features from meteorological data. The combination of these modules significantly improves prediction performance. Experimental results demonstrate that STFM excels in extracting spatiotemporal features from reanalysis data, yielding predictions that closely align with observed values. In comparative studies, STFM outperformed other models, achieving a 7% improvement in ground and high-altitude temperature predictions, a 5% enhancement in the prediction of the u/v components of 10 m wind speed, and an increase in the accuracy of potential height and relative humidity predictions by 3% and 1%, respectively. This enhanced performance highlights STFM’s potential to advance the accuracy and reliability of meteorological forecasting.

Weather forecasting is a classical problem in remote sensing, in which precipitation is difficult to predict accurately because of its complex physical motion. Precipitation significantly impacts human life, work, and the ecological environment. Precise precipitation forecasting is vital for water resource management, ecological protection, and disaster mitigation through precise precipitation forecasting. This study introduces an innovative deep learning-based precipitation-forecasting method DFMM-Precip that integrates reanalysis of precipitation data and satellite data using a multi-modal fusion layer and predicts future precipitation details through a global–local joint temporal-spatial attention mechanism. By effectively combining satellite infrared data with reanalysis data, the approach enhances the accuracy of precipitation forecasting. Experimental results for 24 h precipitation forecasts show that DFMM-Precip’s multi-modal fusion layer successfully integrates multi-modal data related to precipitation, leading to improved forecast accuracy. In particular, the global–local joint temporal-spatial attention mechanism provides precise, detailed forecasting of spatial and temporal precipitation patterns, outperforming other state-of-the-art models. The MSE of the forecasting results is 10 times lower than that of the advanced RNN model and 2.4 times lower than that of the advanced CNN model with single-modal data input. The probability of successful rainfall prediction is improved by more than 10%.

Uncertain parameters in physical parameterizations of general circulation models (GCMs) greatly impact model performance. In recent years, automatic parameter optimization has been introduced for tuning model performance of GCMs, but most of the optimization methods are unconstrained optimization methods under a given performance indicator. Therefore, the calibrated model may break through essential constraints that models have to keep, such as the radiation balance at the top of the model. The radiation balance is known for its importance in the conservation of model energy. In this study, an automated and efficient parameter optimization with the radiation balance constraint is presented and applied in the Community Atmospheric Model (CAM5) in terms of a synthesized performance metric using normalized mean square error of radiation, precipitation, relative humidity, and temperature. The tuned parameters are from the parameterization schemes of convection and cloud. The radiation constraint is defined as the absolute difference of the net longwave flux at the top of the model (FLNT) and the net solar flux at the top of the model (FSNT) of less than 1 W m ⁻² . Results show that the synthesized performance under the optimal parameters is 6.3 % better than the control run (CNTL) and the radiation imbalance is as low as 0.1 W m ⁻² . The proposed method provides an insight for physics-guided optimization, and it can be easily applied to optimization problems with other prerequisite constraints in GCMs.