Effective communication of model uncertainty: Moving towards decision-relevant communications

Abstract

Disaster risk management requires effective communication of complex technical information: in pre-event planning and mitigation, in response activities, and during recovery processes. These range from technical risk assessments to impact projections, and simulations of the outcomes of recovery and adaptation decisions. Each relies on a suite of numerical models, from models of the hazard (e.g., tsunami) to economic projections of decision impacts. Many novel processes exist to communicate these models to both the public and decision makers in policy and practice settings. However, how we communicate the uncertainty inherent to these models remains a challenge. Non-communication of uncertainties is problematic: interdependencies between event characteristics over time create evolving uncertainties that can eclipse any simulated outcome uncertainties. We review the literature covering the many challenges of communicating uncertainty, and present the findings of a metasynthesis literature review for effective communication of model uncertainties. Themes identified include: a) clear typologies to identify and communicate uncertainties, b) effective engagement with users to identify uncertainties to focus on and when, c) managing ensembles, confidence, bias, consensus and dissensus, d) methods for communicating uncertainties., e.g., maps, graphs, time, and e) the lack of evaluation of many approaches currently in use. We propose communicators move from a one-way dissemination of advice, towards two-way and participatory approaches that identify decision-relevant uncertainty and data information needs pre-event, via a shared uncertainty management scheme. This will help identify what communication efforts should focus on during a crisis, and thus enhance situation awareness and data sharing throughout the disaster cycle.