Precipitation nowcasting plays a vital role in preventing meteorological disasters, and Doppler radar data act as an important input for nowcasting models. When using the traditional extrapolation method it is difficult to model highly nonlinear echo movements. The key challenge of the nowcasting mission lies in achieving high-precision radar echo extrapolation. In recent years, machine learning has made great progress in the extrapolation of weather radar echoes. However, most of models neglect the multi-modal characteristics of radar echo data, resulting in blurred and unrealistic prediction images. This paper aims to solve this problem by utilizing the features of a generative adversarial network (GAN), which can enhance multi-modal distribution modeling, and design the radar echo extrapolation model GAN–argcPredNet v1.0. The model is composed of an argcPredNet generator and a convolutional neural network discriminator. In the generator, a gate controlling the memory and output is designed in the rgcLSTM component, thereby reducing the loss of spatiotemporal information. In the discriminator, the model uses a dual-channel input method, which enables it to strictly score according to the true echo distribution, and it thus has a more powerful discrimination ability. Through experiments on a radar dataset from Shenzhen, China, the results show that the radar echo hit rate (probability of detection; POD) and critical success index (CSI) have an average increase of 21.4 % and 19 %, respectively, compared with rgcPredNet under different intensity rainfall thresholds, and the false alarm rate (FAR) has decreased by an average of 17.9 %. We also found a problem during the comparison of the result graph and the evaluation index. The recursive prediction method will produce the phenomenon that the prediction result will gradually deviate from the true value over time. In addition, the accuracy of high-intensity echo extrapolation is relatively low. This is a question worthy of further investigation. In the future, we will continue to conduct research from these two directions.

In 2016, 192 cities, including Wuhan, Nanjing, and Jiujiang, suffered from severe flooding, which raised social and government concerns in China. This paper proposes a method based on the underlying surface to estimate the urban flood-carrying capacity. First, water is extended into the underlying surface vegetation-impervious surface-soil (VIS) model to form the vegetation-impervious surface-soil and water (VIS-W) model. Second, the watershed is delimitated using a digital elevation model (DEM). Third, the natural water system’s storage capacity, including rivers, lakes, and puddles, is estimated. Using the rainfall–runoff hydrology model and the VIS-W model, the storing ability, the receiving ability, the discharge potential, and the emergency of pumping outward are assessed to derive the final flood-carrying capacity. Finally, the result is compared with data on the waterlogging points collected in July 2016 during a flood and waterlogging event in Wuhan. It is found that 84% of waterlogging points are located in the “weak” or “normal” areas, and 16% are located in the “strong” areas. Additionally, 99% of total waterlogging points are located upstream, based on the stream extracted by the DEM. This phenomenon indicates that Wuhan can mitigate flood disasters by fully utilizing the natural water system storage capacity to corporate the city’s drainage schedule.

We present a variational autoencoder (VAE) learning framework with introspective training for conditional image synthesis, and explore conditional capsule encoder by class-wise mask label insertion for this framework. Our model only consists of encoder (E), generator (G) and classifier (C), where E and G can be adversarially optimized, and C helps to boost conditional generation, improve authenticity and provide generation measures for E and G. Discriminator is not necessary in our framework and its absence makes our model more concise with fewer artifacts and pattern collapse problems. To compensate for the blurry weakness of VAE-like models, feature matching is introduced into loss functions by means of C to offer more reasonable measures between real and synthesized images. Moreover, in consideration of the key role of E in autoencoders as well as the interesting characteristics of capsule structure, conditional capsule encoder is preliminary explored in the image synthesis model. Class labels participate conditional encoding by masking high-level capsules of other categories, and capsule loss for the encoder is added to facilitate conditional synthesis. Experiments on MNIST and Fashion-MNIST data sets show that our model achieves real conditional synthesis performances with better diversity and fewer artifacts. And conditional capsule encoder also reveals interesting synthesis effects.

Precipitation nowcasting has important implications for urban operation and flood prevention. Radar echo extrapolation is a common method in precipitation nowcasting. Using deep learning models to extrapolate radar echo data has great potential. The increase of lead time leads to a weaker correlation between the real rainfall evolution and the generated images. The evolution information is easily lost during extrapolation, which is reflected as echo attenuation. Existing models, including generative adversarial network (GAN)-based models, have difficulty curbing attenuation, resulting in insufficient accuracy in rainfall prediction. To solve this issue, a spatiotemporal process enhancement network (GAN-argcPredNet v2.0) based on GAN-argcPredNet v1.0 has been designed. GAN-argcPredNet v2.0 curbs attenuation by avoiding blurring or maintaining the intensity. A spatiotemporal image correlation (STIC) prediction network is designed as the generator. By suppressing the blurring effect of rain distribution and reducing the negative bias by STIC attention, the generator generates more accurate images. Furthermore, the discriminator is a channel–spatial (CS) convolution network. The discriminator enhances the discrimination of echo information and provides better guidance to the generator in image generation by CS attention. The experiments are based on the radar dataset of southern China. The results show that GAN-argcPredNet v2.0 performs better than other models. In heavy rainfall prediction, compared with the baseline, the probability of detection (POD), the critical success index (CSI), the Heidke skill score (HSS) and bias score increase by 18.8 %, 17.0 %, 17.2 % and 26.3 %, respectively. The false alarm ratio (FAR) decreases by 3.0 %.

Watershed delimitation is very important in flood control management. The traditional sub-watersheds delimitated by a filling digital elevation model (DEM) may change the real sink area, such that it may not be the best choice in studies sensitive to sub-watershed storage. This paper proposes a dynamical watershed delimitation method using a no-fill DEM and precipitation. It considers a closed sink area containing cells that fully flow into a large special cell, which can flow out when its water level is “higher than outlet”. We took Wuhan City as a study area and defined the precipitation in return periods of 1, 5, 20, or 100 years to derive the sub-watersheds. It is found that, in the four delimitations, the ratio of isolated basic units which could not flow outside were 27%, 9%, 5%, and 1%, respectively, as the precipitation increased. The results show that the provided method satisfies the assumption that the sink area might overflow with increased precipitation. The sub-watershed delimitated by the proposed method has higher correlation with the distribution of waterlogging points than those delimitated according to the D8 algorithm. These findings indicate that the proposed method can derive reasonable sub-watershed delimitation and that it may be helpful in the practice of urban flood control management.

Accurate rainfall nowcasting is crucial for protecting lives and property. Multi-source meteorological data can reflect rainfall evolution from different perspectives. Making full use of multi-source meteorological data is an effective method to improve the performance of rainfall nowcasting. For rainfall nowcasting task, we propose a cross-modal spatiotemporal joint predictive network, referred to as CM-STJointNet. This is a nonlinear rainfall nowcasting model with radar and satellite data as inputs. Multi-task learning strategy is used to implement radar echo extrapolation and satellite imagery prediction, and this constraint realizes that the model forecasting follows the motion of clouds. We experiments conducted on the publicly available meteorological dataset (SEVIR), CM-STJointNet uses radar and infrared bright temperature observations as input data and realizes two hours of rainfall nowcasting. Compared with four unimodal methods (ConvLSTM, PredRNN, Rainformer, and Earthformer), four multimodal methods (Rain-F, RN-Net, MetNet-2, and S2DIR) demonstrated superior prediction capabilities. The experimental results affirm the high efficiency of CM-STJointNet and provide more insights for future advances in rainfall nowcasting.