Cloud services offer the potential to re-configure how environmental models are developed and used, breaking models out of their scientific environment and make them available for ubiquitous processing and analysis of environmental information. But this won't be straightforward. Here Wouter Buytaert, leader of the EVO Modelling task discusses some of the issues and ideas that will have to be addressed.
Data availability in environmental sciences is expanding at a rapid pace. From the constant stream of high-resolution satellite images to the local efforts of “citizen scientists”, there is an increasing need to process the growing stream of heterogeneous data and turn it into useful information for decision-making. Environmental models, ranging from simple rainfall – runoff relations to complex climate models, can be very useful tools to process data, identify patterns and help predict the potential impact of management scenarios. Therefore, we want to break models out of their scientific environment and make them available for ubiquitous processing and analysis of scientific information. This is not straightforward. Models consist of complex algorithms, assumption, and simplifications that make the interpretation of their results a challenging task. However, new methods of presentation, visualisation and interaction may make this possible.
Historically, the complexity of most environmental models has typically confined them to scientific laboratories and academic computer clusters, where they are harnessed by researchers that understand the algorithms, assumptions, simplifications and potential errors of these models. As a result, decision-makers often have high and unrealistic expectations regarding scientific knowledge and environmental models, as if they will provide clear-cut cases for particular policies. Or they may lack confidence in models and results they cannot reproduce themselves. The integration of information provided by environmental models into policy formulation is therefore quite a challenge.
Recent technological innovations in networking and computing and sensor development enable building a new generation of interactive models plugged into virtual environments. Driven by technological developments, environmental sensors are becoming smaller, cheaper, and increasingly automated. They can be employed in pervasive sensor networks and connected to the internet where they provide constant streams of data. Similarly, environment agencies worldwide are starting to put historical and recent data online, helped by emerging standards for data formatting and access. At the same time, the internet provides exciting technological developments for processing these data. In particular, the advent of cloud computing facilities can tremendously speed up processing, especially for tasks that involve large, parallel calculations, as is typical for uncertainty analyses. Another advantage is that model components can be hosted online, using cloud services. For instance, rather than having to download and install a rainfall-runoff model, it may be made available as a web service by a commercial company or research group, allowing users to send off the input data and get the output data in return.
Technologies for the deployment of model components online are currently under development. For instance, the Web Processing Service being developed by the Open Geospatial Consortium provides rules for standardizing how inputs and outputs (requests and responses) for geospatial processing services need to be formulated. The standard also defines how a client can request the execution of a process, and how the output from the process is handled. Similarly, ontologies or langauges are being defined to facilitate the unambiguous exchange of information between different models, as well as metadata including uncertainties.
These developments provide far-reaching opportunities for environmental models. The availability of online databases and models as web services allows for their inclusion into tools ranging from professional workflows to interactive decision support systems for data querying and simulation aimed at decision-making.
This evolution has the potential to turn the typical top-down flow of information from scientists to users into a much more direct, interactive approach. It opens perspectives to speed up the dissemination of environmental information to a larger community of users, to harvest feedback, and to widen the opportunity to evaluate simulations and predictions from different perspectives. But care should be taken that model outputs are interpreted correctly and not used in an inappropriate context.