Integrated natural resources management is an emerging discipline that aims to move beyond piecemeal approaches and promote integrated actions over a wide range of disciplines and actors. One of the technical challenges of this new discipline is the integration of information from different sources and with different levels of detail. Geographic information systems (GIS) have often been used for this purpose, but they are often inadequate to represent the structure and dynamics of human and natural systems. This thesis aims to provide a new framework for modelling information in the field of integrated natural resources management. This framework relies on the development of a modelling language (called SYSMOD) that encodes the information required by the integrated management process. This language is based on the principles of systems approach and semantic integration that allow integrating, organising and sharing data from multiple knowledge domains. To use the SYSMOD language, an information system was developed in the lab of EPFL Ecohydrology. This information system – known as "Combined Water Information System (CWIS)" because of its application to water management – consists of a database and a Web application. Its user interface provides a set of tools to create: (1) systemic views based on the SYSMOD language, (2) geographical views similar to those of GIS and (3) "report" views used to display and edit data such as text, files, images or numerical values. CWIS was subsequently enhanced to manage scenario data and therefore provide support for scenario planning. Scenarios in strategic planning help to identify the impact of the most uncertain and important factors, and thus promotes the development of strategies adapted to several possible futures. In this context, CWIS has been extended to integrate the concepts of uncertainty and data provenance, to characterise the information validity and trace the sources of data. Most of the time, the creation of scenarios and strategies requires data that can be provided only through simulations. Consequently, a modelling approach has been developed to enable the integration of mathematical models with CWIS. The particularity of this approach is to combine in a single method the concepts of model integration, uncertainty and data provenance to provide detailed results that highlight the risks and uncertainties associated with the scenarios and strategies. CWIS was applied in a series of case studies, in particular the management of water resources in Alexandria (Egypt). CWIS, and in general the framework that was used in its development, have proved to be powerful tools to handle the different types of information in a holistic and transdisciplinary way. This thesis proved that scenario planning, systems thinking, semantic integration and model integration are essential aspects to be considered in the development of tools for integrated natural resources management.