The DeepQPF is a deep learning Quantitative Precipitation Nowcasting (QPN) technique that provides the hourly-accumulated rainmaps for the next three hours. This page provides real-time DeepQPF-PONI and DeepQPF++ rainmaps, including the current prediction and previous predictions of 1-h and 2-h ago. Update schedule: every 10 minutes
Ashesh, C.-T. Chang, B.-F. Chen, H.-T. Lin, B. Chen, and T.-S. Huang, 2022: Accurate and Clear
Quantitative Precipitation Nowcasting based on a Deep Learning Model with Consecutive Attention
and Rainmap Discrimination. Artificial Intelligence for the Earth Systems. 1(3), e210005.
Chen, D.-Y., C.-T. Chang, B.-F. Chen*, & T.-S. Huang, 2025: Integrating Heterogeneous Data for
Deep Learning Precipitation Nowcasting. Weather and Forecasting. 40(10)
, 1977-2001.
DeepQPF 0-3-h QPN Verification
This page offers a comprehensive 72-hour, real-time verification of DeepQPF-PONI and DeepQPF++ 0-1h and 0-3h rainmaps, comparing their performances against the Central Weather Administration QPESUMS observations. Update schedule: hourly
Thunderstorm Warning Guidance
This page provides real-time automated thunderstorm warning guidance products of DeepQPF-PONI and DeepQPF++, integrating QPESUMS observation and DeepQPF prediction to display hourly rainmaps and township-level thunderstorm warning for time windows of −30~+30, −20~+40, -10~+50, and 0~+60 minutes. Update schedule: every 10 minutes