RainGuru
| Project: RainGuru | |
|---|---|
| Section | |
| Status | |
| Start Date | 2022-06-01 |
| End Date | |
| Principal Investigator: | User:Marc Schleiss |
| Short Description | Rainfall nowcasting using machine learning |
Related Media
| date ᵖ | description ᵖ | |
|---|---|---|
 | RainGuru logo |
Project Participants
The RainGuRu https://rainguru.tudelft.nl/ is a deep-learning rainfall forecasting model developed by TU Delft researchers and HKV (https://www.hkv.nl/) with the help of seed funding from the TU Delft Safety & Security Institute. The model uses a gated recurrent convolutional neural network known as Traj-GRU, first proposed in 2017 by Shi et al. (https://arxiv.org/abs/1706.03458) and further developed and adapted for the Netherlands by TU Delft master students Eva van der Kooij (2021), Diewertje Dekker (2022) and Yanghuan Zou (2023).
RainGuru comes in two flavors: RainGuru-live https://rainguru.tudelft.nl/, a free, open source alternative to Buienradar for visualizing the latest rainfall forecasts in the Netherlands, visualize real-time rainfall forecasts, calculating key performance metrics and comparing forecasts to past observations.
Rainguru-edu https://rainguru-edu.tudelft.nl a user-friendly platform where students and researchers can upload their own rainfall forecasting models and easily visualize their predictions on famous historical storms. For security reasons, a password is needed to upload a new model. Please contact Marc Schleiss for more information.