DIRECTORY
TEAM MEMBERS
- Dr. Wei Zhang, Postdoctoral researcher at Department of Geosciences, Princeton University, and NOAA’s Geophysical Fluid Dynamics Laboratory
- Dr. Jia Geng, PHD at Rosenstiel School of Marine and Atmospheric Science, University of Miami
- Dr. Junfei Xia, PHD at Rosenstiel School of Marine and Atmospheric Science, University of Miami
- Dr. Ben Kirtman, Professor at Department of Atmospheric Science, University of Miami; Director of NOAA Cooperative Institute for Marine and Atmospheric Studies; Fellow of the American Meteorological Society
PROJECT SUMMARY
The El Niño-Southern Oscillation (ENSO) significantly influences Earth’s climate, ecosystems, and human societies; Yet, forecasting ENSO at lead times of more than a year remains a challenge for the whole climate community. This project aims at developing a deep learning system by training a convolutional neural network to improve multi-year ENSO prediction, which will potentially supplement current dynamical forecast systems based on global climate models.
PROJECT BACKGROUND
ENSO is a dominant climate patterns involving changes of sea surface temperature in the tropical Pacific. The recurrence of warm and cold phases of ENSO, commonly referred to as El Niño and La Niña events, substantially impacts regional and global climate, water cycle, agriculture, and ecosystems. Climate scientists have started to utilize climate models to forecast ENSO since the 1980s. After decades of effort, current climate prediction systems (e.g., ensemble forecasting using coupled models from the Coupled Model Intercomparison Project Phase 5 and Phase 6; hereafter, CMIP5/6) can skillfully forecast ENSO with lead-times of 6-12 months. Long-lead prediction of ENSO nevertheless particularly with lead-times over one year remains problematic in practice.
PROJECT MILESTONES
May 2021 - Project was initiated!
Jun. 2021 - Project proposal finished and submitted!
July 2021 - Finished data collection and preparation. Collected data included CMIP5/6 historical simulations and observational datasets for sea surface temperature and ocean heat content.
Coming soon…
- Collect and annotate more data.
- Start to develop and train a CNN model.
- Try all kinds of things to optimize the deep learning model.
- Test the reliability of the deep learning ENSO prediction model.
- More ideas later …
BLOG
BLOG#1 - What you need know about ENSO?
BLOG#2 - A Summary of ENSO Prediction Skills
Coming soon…