Geoscience explanations: Identifying what is needed for generating scientific narratives from data models

As models of geoscience systems grow in number and size, so grows the need for tools to help express the output of those models in usable forms. In this paper the utility of the output of a model is defined as its ability to support scientific explanation. Commonly the output of a model might include statistics, graphs, maps, images and animations, which require expert interpretation and evaluation in the context of the model setup or implementation. Here the narrative is presented as a type of output from a model that can present the results of a model in a form that is more useful for the non-expert. Narratives provide a rich medium for expressing causal chains of events that form the basis for explanation and its future use in policy and decision making. This paper reviews research on narratives and their role in scientific explanation. The principles of narrative construction for the Geosciences are identified, which forms the basis for determining the key components needed in explanatory statements for communicating the output of geoscience models. The potential of existing data models used in model output for generating narratives are explored, followed by the conceptual presentation of an extended data model, which supports the narrative unit and has the potential to automate aspects of the generation of scientific explanation in narrative form. (C) 2009 Elsevier Ltd. All rights reserved.