AI

Amortized Localized Posterior Inference for Model Calibration with Application to an Ocean Turbulence Closure

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 30/10/2026, 06:48:24 UTC.

Briefings

A new method uses localized posteriors to improve model calibration by weighting observations based on their proximity to a target context. This approach reduces predictive errors compared to global methods, demonstrated in twin experiments with the Branin-Hoo model and an ocean turbulence closure. The study applies this to a single-column ocean model using atmospheric surface forcing as context.

Key points

• The method constructs a family of localized posteriors by weighting observation likelihoods according to their proximity to a target context.
• Localization reduces reported aggregate predictive errors relative to global posterior calibration in both the Branin-Hoo model and the k-$\epsilon$ turbulence closure.

Context

Model calibration often fails when a single parameter cannot reproduce physical responses across different contexts due to observation and context uncertainties. This research addresses these issues by quantifying and accounting for uncertainties through a new inference framework.

Limitations

The study relies on two twin experiments constructed to admit local compensation. Results are limited to the Branin-Hoo model and the k-$\epsilon$ turbulence closure in a single-column ocean model. The specific city or region where the ocean model was applied is not identified in the evidence.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 30/10/2026, 12:10:42 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeifx36odzcjut72nzcswh6d6drepgcbntsjkvpebomnpj324rfrxhi