资源环境科技发展态势分析平台

There is growing public awareness of global risks that are related to land degradation, poverty, food security, migration flows, natural disasters and levels of violence and conflict. In the past decades, a wealth of performance databases has become available, and these are used to quantify those risks and to influence governance globally. We name the monitoring of the 17 Sustainable Development Goals (SDGs), the establishing of priorities in humanitarian aid programs and the design of early warning forecasting systems. This article addresses a question that underlies the social and political application of risk indicators, namely: how reliable are such data that can be accessed or downloaded 'in a few mouse clicks'? Reliability is an important issue for users of these data since poor data will lead to poor inferences. In addition, flawed data are usually related to poor and fragile countries, countries that need humanitarian aid and financial investments the most. In order to get a grip on this reliability issue, we explore the possible uncertainties attached to global risk-related indicators. In this article we (i) provide an overview of available data sources, (ii) briefly describe the way institutes aggregate risk indicators from an underlying set of basic indicators to form composites, and (iii) identify various sources of uncertainty related to global risk indicators and their composites. Furthermore, we give solutions for coping with uncertainties in the partial or complete absence of such information. We acknowledge that these solutions are insufficient to quantify all (cascading) uncertainties concerning global indicators, especially those related to 'Campbell's law'. Therefore, we applied a 'ringtest across data from leading institutes as for five open access risk indicators: governance, impacts of natural disasters, conflicts, vulnerability/coping capacity, and all security risks combined. We find that the coherence between indicators from different organisations but with identical definitions varies enormously. We find that indicators denoted as 'impacts of natural disasters' are almost uncorrelated across four organisations. However, indicators denoting 'governance' or 'all security risks combined' show remarkable high correlations.