ISSN 1000-0526
CN 11-2282/P
CN 11-2282/P
CastNet: Deep Learning-Based Model for Quantitative Precipitation Nowcasting
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
To improve the accuracy of precipitation nowcasting, this paper proposes an adversarial neural network model named CastNet that combines deep neural networks. This model utilizes the Recurrent Neural Network (RNN) to capture the spatio-temporal features of radar echo data, employs the Adversarial Neural Network (GAN) to simulate the generation and dissipation changes of cloud clusters, and then integrates the optical flow constraint into the neural network to guide the model training, which accelerates the learning process of the neural network and enhances the spatio-temporal consistency of the model, effectively solving the problem of forecast ambiguity and significantly improving the accuracy of precipitation intensity and location. Tests were conducted on 9 major precipitation processes in Guangxi and its surrounding areas (104.41° - 112.08°E, 20.9° - 26.49°N) from May to October 2023. The results show that under various precipitation intensities (≥0.1, ≥2, ≥7, ≥15, ≥25 and≥40 mmh?1), the average TS scores of SWAN 2.0 are 0.458, 0.27, 0.085, 0.034, 0.014 and 0.003 respectively; the average TS scores of SWAN 3.0 are 0.452, 0.402, 0.225, 0.129, 0.085 and 0.048 respectively; and the average TS scores of the CastNet model are 0.439, 0.397, 0.225, 0.139, 0.104 and 0.073 respectively. Its scores are higher than those of SWAN 2.0 and SWAN 3.0 under high-intensity precipitation of ≥7 mmh?1 and above. In addition, as the forecast lead time extends, the relative advantage of CastNet becomes more obvious.
Keywords:
Project Supported:
Guangxi Natural Science Foundation Project（2022GXNSFAA035482）
Clc Number:
Mobile website
