Computational analysis to support flood risk management often depends on the coupling of independently developed software codes into a coherent overall analysis. This is currently achieved by a mixture of ad-hoc code and manual intervention. Ad-hoc code is expensive to write and error-prone. Especially when written for a relatively "one off" analysis, it is often underdocumented. Manual intervention is likewise error-prone; it also makes repeating an analysis difficult and severely limits the number of runs which can be practically undertaken.

It is quite impractical, for example, to apply computational uncertainty or sensitivity analysis tools over an analysis enacted by manually running component codes in sequence, since these techniques depend on running very large numbers of simulations.

In recent years a number of software frameworks have been developed which support, to some degree, the coupling of independently developed simulation codes. A high profile example is OpenMI?, the "open modelling interface", developed by the HarmonIT? project. This project was funded under the European Commission's Fifth Framework research programme, itself part of a cluster of projects focused on supporting the implementation of the Water Framework Directive.

Computational tools to support decision making in flood risk management almost invariably build on simulation, and simulator coupling of this type is therefore an important capability. They generally also involve several layers of higher-level analysis, however, with each layer becoming a component over which the next is applied (see Layered analysis). Each layer may introduce any number of additional dimensions of analysis.

The metamodels underlying existing simulator coupling frameworks like OpenMI? do not recognise this structure, and thus provide no particular support for uncertainty and risk analysis. Developing a metamodel which does has been central to the development of the Reframe software.