Environmental decision support systems (EDSS) are often constituted by a combination of several heterogeneous tools for data and information management, visualization, modelling and simulation, data analysis, reporting and so forth. EDSS need to be tailored towards specific decisions by human decisions makers and often require highly interactive components based on good human factors engineering. They are often expensive to build. This paper presents early results of project SUDPLAN, an EU funded R&D project in the 7th EU Framework Program. SUDPLAN pursues several objectives. From an environmental management perspective, the project aims at combining (larger scale) regional climate models (RCMs) with (smaller scale) local models in order to support urban management decisions related to climate change. From an end user perspective, SUDPLAN aims at supporting urban management decisions with an easy to use Scenario Management System (SMS), which supports end users in the assessment of decision alternatives. This end user support shall in particularly include the automation of recurring tasks. From an informatics perspective, the R& D in SUDPLAN aims at building a generalized framework to build EDSS which make heavy use of models and interactive visualization (part of this, not discussed in this paper, is the extension of geo-spatial towards 3D interaction). This paper focuses on the latter perspective. Over recent year, generic approaches in environmental software systems have concentrated on generic ways of integrating data, and have lead to data infrastructures. Often the term service infrastructure is used in the literature but the majority of generalized or standardized service types are at face value data access services. However, at least the data integration part of EDSS has become much more efficient than it used to be. For highly interactive end user applications including use of models, the actual design and implementation of the graphical user interface (GUI) is still a heavy engineering and development task. This is due to the fact that a) end user applications (including the GUI) are still often developed from scratch, without systematic approach and b) the integration of models into workflows is done in a similarly unsystematic fashion. The SUDPLAN SMS aims at overcoming these software engineering deficiencies. The combination of two aspects constitutes the core of the approach. First, a generalized data model to cope with managing models in the application is used to support the integration of models with the EDSS. This approach is not new and has been proposed by other researchers before. Secondly, an interactive geo-spatial development environment (named cids) is used, which allows to incorporate specialized application themes into a general framework which supports a wide range of applications. This development environment is "receptive" in the sense that it does not only generalize data integration but also the integration of GUI and visualization elements which interact with the underlying information management platform. This is done through a number of software patterns. In the SUDPLAN SMS both approaches are combined and extended towards models. This means that to integrate a model into an EDSS workflow means to instantiate a number of model managers, which support the interaction with the models, and to implement a number of renderers and editors which constitute the end user's GUI. This paper suggests that a combined systematic approach to both model integration and interactivity engineering leads to better engineered software and reduction of development efforts.