How can we ensemble geophysical models better?

Add to your list(s) Download to your calendar using vCal

• Matt Amos | University of Lancaster
• Tuesday 18 May 2021, 11:00-12:30
• https://zoom.us/j/6708259482?pwd=Qk03U3hxZWNJZUZpT2pVZnFtU2RRUT09.

If you have a question about this talk, please contact mwlw3.

Ensembles of geophysical models are typically used to increase the accuracy of projections and to allow for the quantification of projective uncertainties. However, the landscape of these models is complicated with model components being shared between institutions (meaning they’re not independent), compounded by the fact that models are not equally good. When ensembling models together, we need methods that can account for the complicated interconnections between models and the variable model performance.

In this talk, I’ll introduce the concept of ensembling models and how we might go about doing this better, to consider variable model skill and model dependence. I’ll present two applications of model ensembling: a process-based weighting to project ozone hole recovery, and Bayesian neural network method to infill gaps in historic ozone records using an ensemble of models. Though the applications are specific these methods are widely applicable to any user of ensembles of environmental models.

This talk is part of the AI4ER Seminar Series series.

This talk is included in these lists:

• AI4ER Seminar Series
• Cambridge Centre for Data-Driven Discovery (C2D3)
• Cambridge talks
• Chris Davis' list
• Interested Talks
• Trust & Technology Initiative - interesting events
• bld31
• https://zoom.us/j/6708259482?pwd=Qk03U3hxZWNJZUZpT2pVZnFtU2RRUT09
• ndk22's list
• ob366-ai4er
• rp587

Note that ex-directory lists are not shown.