Geoweb Services for Sharing Modelling Results in Biodiversity Networks

Karla Donato Fook, Antônio Miguel Vieira Monteiro, Gilberto Câmara, Marco Antônio Casanova, Silvana Amaral
2009 Transactions on GIS  
Biodiversity researchers in different institutions deal with predictive models for species distribution. These models are useful for biodiversity conservation policies. Species distribution modelling tools need large datasets from different sources and use many algorithms. To improve biodiversity science, scientists need to share models, data and results, and should be able to reproduce experiments from others. This article presents a geoweb service architecture that supports sharing of
more » ... g results and enables researchers to perform new modelling experiments. We show the feasibility of the proposed architecture by developing a set of prototype services, called Web Biodiversity Collaborative Modelling Services -WBCMS. They provide a set of geospatial web services that support the sharing of species distribution models. The article includes an example of a model instance that explains the WBCMS prototype. We believe that WBCMS shows how to set up a cooperative research network on biodiversity research. Biodiversity research needs measurements or inferences about species location and abundance. Since comprehensive surveys are unaffordable for large areas, species distribution models are used as indicators of species diversity. These models combine in situ data with environmental layers to predict the species distribution over a geographic area. They estimate species potential niches by comparing known occurrences and known absences with ecological limits, also called environmental variables, such as precipitation and temperature (Soberón and Peterson 2004) . Their results support biodiversity protection policies, are useful to forecast the impacts of climate change, and help to detect problems related to invasive species. Scientists working with predictive species distribution modelling need access to large sets of geospatial data such as climate, vegetation, topography, and land use (Giovanni 2005) . Since such datasets may be archived by different institutions, a scientist needs to locate them and make them interoperate. This creates a technical challenge of representing, managing, storing, and accessing distributed geospatial data. Accessing distributed geospatial data is more complex than accessing conventional data, given its large semantic and geometric variation (Breitman et al. 2006) . In addition, the scientist needs algorithms, which may also be available elsewhere. After s/he produces a result, s/he can share it with her (his) community and compare it with similar work. This scenario points out the need for a computational infrastructure that supports collaborative biodiversity studies, allowing sharing of data, models, and results (Ramamurthy 2006) . Sharing data needs information about location of repositories, archival formats, and semantic information. Sharing models needs understanding of the applicability of each algorithm to the species being modelled; it also needs good documentation about the explicit and implicit assumptions of each model. For sharing results, the scientist needs to publish the species distribution maps in a way that allows exchanging of reports, comments and ideas. Collaboration among researchers is not only about exchanging data but also about comparisons between scientific models and experimental results. To perform comparisons between models and results, provenance information is critical (Simmhan et al. 2005) . "Provenance data are essential if experiments are to be validated and verified by others, or even by those who originally performed them. It is also important in assessing the quality, and timeliness of results" . Therefore, provenance data needs to be available when models are shared. This article proposes a geoweb service architecture to support collaboration for species distribution modelling networks. We show the feasibility of the proposed architecture by developing prototype services: the Web Biodiversity Collaborative Modelling Services -WBCMS. These services provide a set of geospatial web services that support sharing of species distribution models. WBCMS protocols allow sharing of data, modelling results and information about data and results provenance. They also enable biodiversity researchers to conduct new experiments using existing models. For an early discussion of WBCMS, see Fook et al. (2007) . The WBCMS architecture is part of the OpenModeller Project, a framework for collaborative building of biodiversity models (Muñoz 2004 , Giovanni 2005 , OpenModeller 2005. This article is structured as follows. Section 2 provides a general discussion on species distribution models, and related work. Section 3 describes the WBCMS
doi:10.1111/j.1467-9671.2009.01170.x fatcat:vg5stjciszcwjcl4dckim3ieum