Uncertainty propagation in chained web based modelling services: the case of eHabitat Jon Olav Skøien 1 , Michael Andreas Schulz 1 , Gregoire Dubois 1 , Richard Jones 3 , Gerard B. M. Heuvelink 2 , Dan Cornford 3 Abstract eHabitat is a Web Processing Service (WPS) designed to compute the likelihood of finding ecosystems with similar conditions. Starting from a reference area, typically a protected area, one can compute for each pixel of a region of interest the probability to find a combination of a set of predefined environmental indicators that is similar to the one observed in the reference area using the Mahalanobis distances to the mean and covariance of these indicators. Inputs to the WPS are thus the reference polygon and a set of environmental indicators, typically thematic geospatial “lay- ers”, which can be discovered using standardised catalogues. The outputs can be tailored to specific end user needs in terms of data format and data resolution. Because these input layers can range from geophysical data captured through remote sensing to socio-economical indicators, eHabitat is exposed to a broad range of different types and levels of uncertainties which are inevitably propagated through the service (see e.g. Heuvelink, 1998). Potentially chained to other services to perform ecological forecasting, for example, eHabitat would be an additional component further propagating uncertainties from a potentially long chain of model services. Such a configuration of distributed data and model services as envisaged by initiatives such as the “Model Web” from the Group on Earth Observations, to be of any use to policy or decision makers, requires from users clear information on data uncertainties. The devel- opment of such an Uncertainty-Enabled Model Web is the scope of the UncertWEB project which is promoting in- teroperability between data and models with quantified uncertainty and building a framework on existing open, in- ternational standards. It is the objective of this paper to illustrate a few key ideas behind UncertWeb using eHabitat to discuss the main types of uncertainties the WPS has to deal with and to present the benefits of the use of the Un- certWeb framework. 1. Introduction Among the range of species distribution models (SDM) used in ecology [Guisan and Zimmerman, 2000], a relatively common method is based on the use of the Mahalanobis distance to create Environmen- tal suitability maps (ESM) [Clark et al., 1993; Knick and Dyer, 1997; Rotenberry et al., 2002]. These maps are based on presence only observations, and have traditionally been used to estimate the potential habitats of single species based on a set of environmental indicators. Despite its simplicity, the Mahalano- bis distance has the potential to perform equally well as more complex models [Johnson and Gillingham, 2005; Tsoar et al., 2007] provided that suitable environmental variables are chosen as indicators. SDMs tend to predict the potential habitat of a species rather than the realized habitat, which in some cases can be considerably smaller. The Mahalanobis distance is also one of the methods that can be used to forecast the suitability of habi- tats, comparing the forecasted climate of a given habitat with the current climate. Some caution has been advised for this use of SDMs, in particular for single species [Hijmans and Graham, 2006; Sinclair et al., 2010]. There are several reasons for this: first of all, these models do not take into account incomplete 1 European Commission, Joint Research Centre, Institute for Environment and Sustainability, Ispra, 21027(VA), Italy, email: (jon.skoien, michael.schulz, gregoire.dubois) @jrc.ec.europa,eu 2 Wageningen University, Land Dynamics Group, Wageningen, the Netherlands, email: gerard.heuvelink@wur.nl 3 NCRG and Computer Science, Aston University, Birmingham, UK – (jonesrm1, d.cornford) @aston.ac.uk EnviroInfo 2011: Innovations in Sharing Environmental Observations and Information Copyright 2011 Shaker Verlag Aachen, ISBN: 978-3-8440-0451-9