Intensification of short‐duration rainfall extremes contributes to increased urban flood risk. Yet, it remains unclear how upper‐tail rainfall statistics could change with regional warming. Here, we characterize the non‐stationarity of rainfall extremes over durations of 1–24 hr for the rapidly developing coastal megalopolis of the Greater Bay Area, China. Using high‐resolution, multi‐source, merged and gridded data we observe greater increases in rainfall intensities over the north‐central part of the region compared with the southern coastal region. Our results show, for the first time, that urbanization nonlinearly increases rainfall intensities at different durations and return periods. Over short durations (≤3‐hr) and short return periods (2‐yr), urban areas have the greatest scaling rates (≥19.9%/°C). However, over longer durations (≥9‐hr) rural areas have greater scaling rates, with a lower degree of dependency on both durations and return periods.

The mainstream of the existing approaches for video prediction builds up their models based on a Single-In-Single-Out (SISO) architecture, which takes the current frame as input to predict the next frame in a recursive manner. This way often leads to severe performance degradation when they try to extrapolate a longer period of future, thus limiting the practical use of the prediction model. Alternatively, a Multi-In-Multi-Out (MIMO) architecture that outputs all the future frames at one shot naturally breaks the recursive manner and therefore prevents error accumulation. However, only a few MIMO models for video prediction are proposed and they only achieve inferior performance due to the date. The real strength of the MIMO model in this area is not well noticed and is largely under-explored. Motivated by that, we conduct a comprehensive investigation in this paper to thoroughly exploit how far a simple MIMO architecture can go. Surprisingly, our empirical studies reveal that a simple MIMO model can outperform the state-of-the-art work with a large margin much more than expected, especially in dealing with long-term error accumulation. After exploring a number of ways and designs, we propose a new MIMO architecture based on extending the pure Transformer with local spatio-temporal blocks and a new multi-output decoder, namely MIMO-VP, to establish a new standard in video prediction. We evaluate our model in four highly competitive benchmarks. Extensive experiments show that our model wins 1st place on all the benchmarks with remarkable performance gains and surpasses the best SISO model in all aspects including efficiency, quantity, and quality. A dramatic error reduction is achieved when predicting 10 frames on Moving MNIST and Weather datasets respectively. We believe our model can serve as a new baseline to facilitate the future research of video prediction tasks. The code will be released.

Precipitation nowcasting is an important task in weather forecast. The key challenge of the task lies at radar echo map extrapolation. Recent studies show that a convolutional recurrent neural network (ConvRNN) is a promising direction to solve the problem. However, the extrapolation results of the existing ConvRNN methods tend to be blurring and unrealistic. Recent studies show that generative adversarial network (GAN) is a promising tool to address the drawback, while it suffers from the instability for training. In this letter, we build a novel ConvRNN model based on the energy-based GAN for radar echo map extrapolation. The method can alleviate the blurring and unrealistic issues and is more stable. We have conducted experiments on a real-world data set, and the results show that the proposed method outperforms several existing models, including optical flow, convolution gated recurrent unit (ConvGRU), and generative adversarial ConvGRU (GA-ConvGRU).

Precipitation nowcasting refers to the prediction of small-scale precipitation events at minute and kilometer scales within the upcoming 0 to 2 hours, which significantly impacts both human activities and daily life. However, prevailing deep learning models have primarily focused on a single radar echo data source, limiting their ability to effectively capture intricate and rapidly evolving precipitation patterns. Thus, meteorological satellite is considered to supplement radar echo data. To achieve a comprehensive integration of multi-source data with enhanced details, a two-stage fusion satellite and radar GAN-based prediction network (named FsrGAN) is proposed. In the first stage, we design a satellite-radar fusion prediction network known as FsrNet. This network employs an encoder-fusion-decoder architecture, where a novel spatial-channel attention (SCA) is proposed to enhance the filtering and fusion of multi-source and multi-scale features. In the second stage, we introduce a GAN-based network (FusionGAN) that also mines the complementary information of satellite images to sharpen the first-stage predicted radar maps with more details. Experiments are conducted on meteorological dataset in the Yangtze River Delta (YRD) region. The test results exhibit the notably superior performance of our model in terms of image quality and precipitation forecasting metrics in comparison to traditional optical flow-based methods and some well-known deep learning methods (ConvLSTM, ConvGRU, TrajGRU and PredRNN++). More importantly, our fusion model using satellite and radar data demonstrates the ability to predict convective initiation.