Satellite cloud images play an important role in weather analysis and forecast. High‐resolution satellite images play a significant role in the study of mesoscale weather systems such as typhoons. With the increasing demands of locating and tracking techniques, the resolution of satellite images is no longer satisfactory. Enhancing their resolution with superresolution (SR) methods can help in identifying and locating weather systems. In this paper, we propose a multipath network model, called SRCloudNet, that involves joint training of a back‐projection network and a local residual network. SRCloudNet integrates features extracted from back‐projection units and residual dense blocks to achieve more accurate image reconstruction. We also developed a novel natural‐color cloud and contrail image data set, constituting the first‐ever satellite cloud image data set established for SR research. Because of the special features, contrail images were first used to test the performance of SRCloudNet. Extensive experiments demonstrated that SRCloudNet achieves superior performance.
Precipitation with high spatial and temporal resolution can improve the defense capability of meteorological disasters and provide indispensable instruction and early warning for social public services, such as agriculture, forestry, and transportation. Therefore, a deep learning-based algorithm entitled precipitation retrieval from satellite observations based on Transformer (PRSOT) is proposed to fill the observation gap of ground rain gauges and weather radars in deserts, oceans, and other regions. In this algorithm, the multispectral infrared brightness temperatures from Himawari-8, the new-generation geostationary satellite, have been used as predictor variables and the Global Precipitation Measurement (GPM) precipitation product has been employed to train the retrieval model. We utilized two data normalization schemes, area-based and pixel-based normalization, and conducted comparative experiments. Comparing the estimated results with the GPM product on the test set, PRSOT_Pixel_based model achieved a Probability Of Detection (POD) of 0.74, a False Alarm Ratio (FAR) of 0.44 and a Critical Success Index (CSI) of 0.47 for two-class metrics, and an Accuracy (ACC) of 0.75 for multi-class metrics. Pixel-based normalization is more suitable for meteorological data, highlighting the precipitation characteristics and obtaining better comprehensive retrieval performance in visualization and evaluation metrics. In conclusion, the proposed PRSOT model has made a remarkable and essential contribution to precipitation retrieval and outperforms the benchmark machine learning model Random Forests.