Sensor Web Enablement based Model Web Implementation for Climate Change Applications

Abstract

Interoperable Web-enabled environmental models are essential part of the service oriented environmental applications. Concepts, architectures, standards and software allowing seamless integration of models in distributed webenabled applications (Model Web) are therefore a logical extension of the quest for interoperable service infrastructures allowing visualization and data access that is currently implemented within the scope of Infrastructure for Spatial Information in Europe directive (INSPIRE), and necessary for further development of the Global Monitoring for Environment and Security (GMES), Shared Environmental Information System (SEIS), and the Global Earth Observation System of Systems (GEOSS) initiatives. SANY Sensor Service Architecture (SensorSA) proposed the Sensor Web Enablement (SWE) suite of standards developed by Open Geospatial Consortium (OGC) as a base for web-enabling the “sensor-like models”. Sensor-like in this context means that the output data model contains observations, that is a set of values with (at least) the associated units, spatial and temporal context. This is true for a great majority of environmental models, and for many other sources of information including e.g. cadastres and the Volunteered Geographic Information (VGI). This versatility of SWE, as well as its focus on processes rather than on the physical sensors is often confusing for end users. We therefore recommend use of “observation web” as a technology-neutral synonymous for sensor web in all discussions with end users and decision makers, and SWE-based Model Web for web-enabled models with SWE service interfaces and data models. Our experiences in SANY and, more recently, in SUDPLAN research project show that SWE indeed provides most of the Model Web functionality for environmental and climate change applications. The Sensor Observation Service (SOS) allows us to access both the underlying data required for the model run, as well as to expose the model results; the Sensor Planning Service to configure, schedule and control the model runs; and the Sensor Modelling Language (SensorML) and Observations & Measurements (O&M) modelling languages to describe the process and the result set respectively. This paper presents our experiences with SWE-based Model Web in SUDPLAN, and discusses the already achieved results, as well as the planned model web developments. Special attention shall be given to (1) process description in both SOS and SPS context; (2) various possibilities for presenting the time series of 2D coverages in O&M; (3) model parameterization, execution and monitoring; (4) input-, output- and processing uncertainties. Keywords: Environmental modelling; Open Geospatial Consortium; Sensor Web enablement; OGC SWE; Model Web; Observation Web; Time Series Toolbox; Sensor Service Architecture; SensorSA;