0.25
0.5
0.75
1.25
1.5
1.75
2
Supermodels: Interactive Ensembles of Imperfect Models
Published on May 27, 20133800 Views
At a dozen or so institutes around the world, comprehensive climate models are being developed and improved. Each model provides reasonable simulations of the observed climate, each with its own stren
Related categories
Chapter list
Supermodels: Interactive Ensembles of Imperfect Models00:00
Contributors to this presentation00:10
SUMO project00:31
Overview00:56
Climate models - 101:47
Climate models - 203:07
Sources of prediction uncertainty - 103:48
Sources of prediction uncertainty - 204:25
Sources of prediction uncertainty - 304:43
Sources of prediction uncertainty - 404:59
Coupled Model Intercomparison Project05:24
Motivation - 106:47
Dynamical systems10:26
Motivation - 210:32
Supermodeling as multimodel interactive ensemble methods12:44
Supermodel proposals17:59
Synchronization in chaotic systems18:04
Strong coupling leads to synchronization18:23
Large connection limit20:38
Supermodel proposals: weighted SUMO24:01
SUMO learning27:40
SUMO learning: Cost function of short integration error29:19
Vector field learning in weighted SUMO 30:48
SUMO in low dimensional perfect model experiments32:35
Lorenz 63 set up33:27
SUMO in Lorenz 6333:59
Imperfect model setting: chaotically driven Lorenz 6335:17
Overfitting in vectorfield learning versus attractor learning36:26
Atmospheric GCMs coupled to the same ocean (Shen e.a. 2013)39:02
SUMO of two atmospheric GCMs coupled to the same ocean (Shen e.a. 2013)41:24
Synchronization of the atmospheres42:34
Bias in mean SST44:20
Bias in mean SST with optimized SUMO47:10
Bias SST SUMO test set47:18
Conclusions47:46
Thank you49:18