Lucy Bastina,b*, Daniel McInerneyb, Goncalo Revezc, Carlos Figueiredoc, DÌÁrio Simonetti, Jose Barredob, Frederic Achardb, Jesus San-Miguel-Ayanzb
a Aston University, Birmingham, United Kingdom
b Institute for Environment & Sustainability, European Commission Joint Research Centre, Italy
c Edisoft SA, Lisbon, Portugal
Forests play a crucial role in timber production, maintenance and development of biodiversity and in carbon sequestration and storage. However, the reliable information on forest extent, type and change, which is so important to policy makers and managers, is frequently inconsistent, incomplete and out of sync between countries and continents. In this article, we describe components developed in line with the GEOSS and INSPIRE frameworks, which access and use distributed forest information at different spatial scales, and demonstrate how these can be integrated into geographic information infrastructures. The components include an INSPIRE-compliant metadata catalogue for discovery of forest information, Web map viewers and Web-based forest analysis and modeling tools. The aim is to provide access to forest resources data at local, regional and global spatial scales, and to provide analytical capabilities for monitoring and validating forest change, using standardized Web services to facilitate accessibility, interoperability and data transfer. In this context, we also introduce concepts related to Open Geospatial Web Standards.
GEO and GEOSS
The Global Earth Observation System of Systems (GEOSS) is organized by the Group on Earth Observations (GEO), which includes more than 40 international organizations, 62 nations and the European Union. The aim of GEOSS is to build a public infrastructure to ÛÏlink together existing and planned observing systems around the world and support the development of new systems where gaps currently exist,Û (GEO, 2009). The infrastructure that coordinates access to the systems, applications, models and products is the GEOSS Common Infrastructure (GCI).
GEOSS is structured into nine societal benefit areas (SBAs), with forestry included explicitly within the agricultural benefit area. However, forest information is vital to all nine SBAs; for example, it is used to define land cover and vegetation type within the agriculture, ecosystems and biodiversity SBAs, to calculate biomass and carbon storage potential in the climate and energy areas, and as a factor in hydrological modeling for the water area. Changes in forest structure and composition thus influence all SBAs, creating a widespread need for readily accessible and usable forest data at a range of scales, from local through to national and global.
There is clearly a huge theoretical potential for forest data from a variety of scales and disciplines to be collated and aggregated into fuller, richer datasets which permit consistency checking, better analysis of multi-scale phenomena and filling of important data gaps. However, in practice there is frequently a distinct lack of harmonized forest information available at regional or continental scales and in many cases data are scattered, incomplete or not readily comparable (MEA, 2005). On the one hand, this can be caused by differences in forest definitions between countries (Vidal et al., 2008), but often it is a technical issue because the systems and data structures at local, regional and global levels cannot interoperate. Initiatives such as the INSPIRE Directive (INSPIRE, 2011), GEOSS and the EuroGEOSS project are making advances to standardise access to spatial information as well as providing common infrastructures allowing data to be discovered, shared and reused.
Web-based Forest Data
Forest data and statistics are typically compiled by forest authorities through national forest inventory (NFI) programs, which collect in-situ information, including estimates of forest area, species composition and growing stock (Tomppo et al. 2010). These data are used for strategic planning and production forecasting at national and regional levels, but they are also used to generate indicators for compliance with international reporting requirements. The data gathered at an international scale is typically managed by bodies such as the United Nations Food and Agriculture Organization (FAO) Forest Resource Assessment, the Ministerial Conference on the Protection of Forests in Europe and the United Nations Framework Convention on Climate Change (UNFCCC) Land use, Land-use Change and Forestry (LULUCF).
Pan-European forest data are compiled and maintained by the EC’s Joint Research Centre through the European Forest Data Centre (EFDAC), which incorporates the European Forest Fire Information System (EFFIS) and the e-Forest platform. These systems maintain continental scale information on the condition and health of forest resources, the spatial distribution, composition and fragmentation of forests as well as modeled species distributions based on climate change scenarios. Furthermore, EFFIS provides real-time information to Member States on the fire risk, extent of burnt areas and aggregated forest fire statistics.
At a global scale, the TREES-3 Action of the Joint Research Centre supplies data services which focus specifically on the mapping and monitoring of land cover change over time, using medium-resolution satellite imagery. In order to document forest cover change, the TREES-3 project gathers land cover information for sample sites at more than 4,000 latitude and longitude degree confluence points in Eurasian boreal forests and the tropical zones of Africa, Asia, and South America (Achard et al., 2010). This exercise constitutes a part of the Global Forest Resources Assessment program, coordinated by the Food and Agricultural Organization of the UN (Lindquist et al 2012). While the overall scope of the project is global, there are important thematic and geographic overlaps with the data services mentioned above. TREES-3 also addresses forest cover and cover change issues related to EU commitments to Multilateral Environmental Agreements, in particular UN conventions such as the UNFCCC, UNCCD, UNCBD, and the UN Forest Forum, and Action Plans such as Forest Law Enforcement Governance and Trade (FLEGT).
Historically, data from systems such as those mentioned above would be supplied as tables, maps and statistical summaries in printed reports, or made available via FTP or CD. However, many of the above systems now publish datasets as standard OpenGeoSpatial Web Services (OWS); the nature of these services is more fully discussed in the following section.
Web-based Data dissemination mechanisms
Increasingly, traditional means of data dissemination are being supplemented and extended by Web services that permit data to be discovered, combined and re-used. Key to this process are the geo-spatial Web-based standards defined by the Open GeoSpatial Consortium (OGC), which have enabled systems to become far more interoperable in general, and have greatly assisted in the exchange of spatial data. These standards apply specifically to the publication of metadata within catalogues, but are also widely used to disseminate spatial data through standards-based Web Map Services (WMS), Web Feature Services (WFS) and Web Coverage Services (WCS). The WMS standard provides access to maps of spatially referenced data that are rendered using a pictorial format, such as PNG, GIF or JPEG, through simple browser requests submitted using URLs. WMS are widely employed for easily accessible cartographic visualizations of data stored on a server. WFS and WCS allow a client to retrieve and update not just images, but geospatial data itself. WFS is used for vector spatial data and acts primarily as a feature access service, while the WCS specification allows the exchange of gridded or coverage data that may also have a temporal dimension. In other words, in contrast to the simple image provided by the WMS, the user receives a collection of coordinate pairs or numeric pixel values which can be used in analytical tasks. Thus WFS and WCS can be used as downloading data services, and are frequently consumed by Web Processing Services (WPS) and within Web-based modeling activities. Web Processing Services, which make analytical functions and algorithms available through Web interfaces, are key in the use and exploitation of the data delivered in these ways. Their role is more fully discussed, with examples, later in this article.
Web clients for viewing and editing forest data
Within the context of the EuroGEOSS project, a dedicated Web map viewer was designed to access distributed forest data services that were published as OGC services, with the facility to remotely add data from WMS Servers (Figure 1). The viewer provided the standard functionality, but more importantly, it allowed the user interface to control the Web Processing Services described below. Another feature of the viewer was to provide the facility for a user to add geo-tagged Web2.0 resources, such as geo-referenced Flickr images or YouTube videos.
Web services and open standards can also be used to facilitate the updating of geospatial data for forestry. The TREES-3 project requires input from numerous national experts, who are called upon to validate the results of automated land cover classifications at the sample sites, and to refine and improve the classification of cloud-covered or burnt areas. This validation is vital in ensuring the quality of forest change estimates (Eva et al, 2012). In past exercises, this validation process has been carried out using a desktop-based IDL tool (Simonetti et al., 2011). Recently, a browser-based version of the tool was generated, allowing direct database updates by authorized experts who select one or more land cover polygons, and update or verify their classification. This requires no local data, all imagery and vector data is hosted at JRC, and no local software installation apart from a Web browser. Watch a video of the tool in use.
Web processing services for analysis and geo-visualization of forest data
From this growing culture of standardized, interoperable Web Services and data formats has arisen the concept of the Û÷Model Web’ ÛÒ an approach to the encoding and wrapping of processing algorithms which allows scientific models to be exposed as Web Services in a flexible, distributed architecture (Geller & Turner, 2007, Nativi et al., 2012). Though there are still many technical challenges in the delivery of the Model Web (Service, 2011), a workflow of interoperable model components has the potential to answer more questions than the individual models alone, allowing users to address complex questions in a variety of different contexts, drawing from multi-disciplinary and distributed repositories of data and information models.
To illustrate the value of such Web-based modeling, we will discuss three examples of WPS produced within the EuroGEOSS project, which are federated within the EuroGEOSS Discovery Broker. Two of these have been integrated directly into the above-mentioned Map Viewer to allow European forestry experts to perform specialized geo-analysis on central mapping resources, while the third example is a generic ecological modeling tool, which can accept a variety of user-defined environmental variables.
The European WPSs were specifically developed to allow scientists to address specific questions on forest change, using a standardized approach, which permits the results to be comparable and reproducible. The specific functions are: 1) To monitor the impact of forest fires in protected areas; 2) To assess the extent of forest change between two time periods.
WPS 1: Forest Fires in Protected Areas
In recent years, forest fires have affected about 300,000 hectares of forest every year across Mediterranean countries (EFFIS, 2010), posing a significant threat to the vegetation structure and biodiversity within protected areas. The service provides users with direct access to the EFFIS Burnt Areas for the Iberian Peninsula and allows users to select an area of interest or a protected area, in order to calculate the impact of the forest fire within it. The results are highlighted within the map viewer and can be viewed with other contextual Web-based information.
WPS 2: Monitoring Forest Change
The monitoring of forest change is becoming increasingly important as governments and forest authorities quantify changes in forest cover and carbon stocks to estimate anthropogenic forest-related greenhouse emissions. Due to the differences in forest definitions and forest inventory methods used by countries, it is quite challenging to obtain harmonized results that can be directly compared between countries. Within Europe, this can be overcome by the use of forest cover maps produced using consistent methodologies at a pan-European scale (Pekkarinen et al. 2009; Kempeneers et al. 2010).
However, this service provides scientists and citizens with a standardized and easy means of processing and interpreting the data through a Web browser, allowing them to analyze the extent of gain, loss and stability of forest areas within a specified AOI. In this example, the forest data are WCS layers of the JRC Forest Map produced for the years 2000 and 2006 at a resolution of 25 meters. The service allows the end-user to digitize an area of interest, which is used to analyze the changes in forest area between 2000 and 2006. The results are presented as a thematic map in the viewer along with statistical graphs.
A third service which has great potential for use in forest modeling under scenarios of climate change is eHabitat, a WPS which computes the likelihood of finding ecosystems with equal properties to those specified by a user. The eHabitat is just one component of the Digital Observatory for Protected Areas (DOPA) that is currently being developed at the Joint Research Centre of the European Commission in collaboration with other international organizations, including the Global Biodiversity Information Facility (GBIF), the UNEP-World Conservation Monitoring Centre (WCMC), Birdlife International and the Royal Society for the Protection of Birds (RSPB). DOPA is conceived as a set of distributed databases combined with open, interoperable Web services to provide end-users, from park managers to scientists and decision-makers, with the means to assess the state of protected areas at the global scale (Dubois et al., 2010).
The main idea behind eHabitat is to provide a service that allows end-users to find areas with similar ecological properties to a reference location. The Mahalanobis distance (Mahalanobis, 1936) is currently used as the similarity measure, though other algorithms may be employed in future. Mahalanobis distance is independent of the different scales and units of the measurements, and its potential applications are therefore practically unlimited, provided that appropriate input data are available. There is a broad range of interdisciplinary possibilities, ranging from socio-economic modeling and ecological forecasting to the optimization of environmental monitoring networks. When chained with services that predict the effects of climate change, eHabitat can be used for ecological forecasting by decision-makers assessing different strategies when selecting new areas to protect. It can also be used for species distribution modeling (Dubois et al., in review). eHabitat can use virtually any thematic data to define the characteristics of ecosystems, and when combined with forecasts as to how those variables will change, it becomes a powerful tool for predicting the ecosystems’ future persistence under different climatic or development scenarios.
Multi-disciplinary information integration is recognized by the scientific community as essential for the understanding of complex issues such as the response of biodiversity to global changes. This calls for flexible, interoperable and scalable systems that allow integration with existing, and heterogeneous, services and data systems. We hope that we have illustrated some of the support for environmental management that can be achieved using such systems.
One danger, however, of an analysis environment based on numerous interacting model services is the potential use of a broad range of data types from uncontrolled sources, not to mention the potential for error introduced by the models themselves. Forest managers operating in this distributed environment would encounter many different types and levels of uncertainties, which may well propagate through complex analysis tasks. For future developments, we are building on the lessons learned from the UncertWeb project, which promotes and develops tools and standards for quantifying and communicating uncertainty in a distributed, interoperable Model Web (Cornford et al., 2010, Bastin et al., 2011)
Achard F., Stibig H.-J., Eva H.D., Lindquist E., Bouvet A., Arino O., Mayaux P. (2010) Estimating tropical deforestation. Carbon Management, 1(2), 271ÛÒ287.
Bastin, L., Cornford, D., Jones, R., Heuvelink, G.B., Pebesma, E., Stasch, C., Nativi, S., Mazetti, P., Williams, M. (2012). Managing Uncertainty in Integrated Environmental Modelling Frameworks. Environmental Modelling and Software, 2012 available online at http://dx.doi.org/10.1016/j.envsoft.2012.02.008.
Cornford, D., Jones, R., Bastin, L., Williams, M., Pebesma, E., Nativi, S. (2010) UncertWeb: chaining Web services accounting for uncertainty. Geophysical Research Abstracts, Vol. 12, EGU2010-9052.
Dubois, G., M. Clerici, S. Peedell, P. Mayaux, J.-M. GrÌ©goire, E. BartholomÌ© (2010), A Digital Observatory for Protected Areas – DOPA, a GEO-BON contribution to the monitoring of African biodiversity, In: ÛÏProceedings of Map Africa 2010Û, 23-25 November 2010, Cape Town, South Africa.
Dubois, G., Schulz, M., Jon, S., Bastin, L., Peedell, S. (in review) eHabitat, a multi-purpose Web Processing Service for ecological modeling. Environmental Modelling and Software, updated draft submitted May 2012.
Eva H.D., Achard F., Beuchle R., De Miranda E., Carboni S., Seliger R., Vollmar M., Holler W., Oshiro O., Barrena V., Gallego J. (2012) Forest cover changes in tropical South and Central America from 1990 to 2005 and related carbon emissions and removals Remote Sensing. 2012, 4, 1369-1391; available online at: http://www.mdpi.com/2072-4292/4/5/1369
FAO (Food and Agriculture Organization of the United Nations) (2010). Global Forest Resources Assessment 2010 Retrieved July 11, 2011 from http://www.fao.org/docrep/013/i1757e/i1757e.pdf
Geller, G. N., Turner, W. (2007). The model Web: a concept for ecological forecasting, Geoscience and Remote Sensing Symposium, 2007. IGARSS 2007. IEEE International, 2469 ÛÒ 2472, 23-28 July 2007.
GEO (2009). Group on Earth Observations, GEO 2009-2011 Work Plan. December 2009. Available at: http://www.geosec.org/documents/work%20plan/geo_wp0911_rev2_091210.pdf (accessed 22 July 2011).
INSPIRE (2011) Infrastructure on Spatial Information in Europe http://inspire.jrc.ec.europa.eu/ Accessed 08.2011
Kempeneers, P. Sedano,F., Seebach, L., Strobl, P., San-Miguel-Ayanz, J. (2011) Data fusion of different spatial resolution remote sensing images applied to forest type mapping, IEEE Transactions on Geoscience and Remote Sensing, 49 (12), 4977-4986.
Lindquist, E. et al. (2012) Improving Access, Utility and Analyses of FAO Forestry Statistics via Geographic Web-Services Earthzine Available online: https://earthzine.org/2012/06/01/improving-access-utility-and-analyses-of-fao-forestry-statistics-via-geographic-web-based-services/
MEA (Millenium Ecosystem Assessment) (2005). Û÷Forests and Woodland Systems’ Chapter 21 in Ecosystems and human well-being, Volume 1: current state and trends : Ândings of the Condition and Trends Working Group. R. Hassan, and R. Scholes, Ed. Neville Ash, 200, pp. 587-614.
Nativi, S., P. Mazzetti, Geller G. N. (2012). Environmental Model Access and Interoperability: the GEO Model Web Initiative. Environmental Modelling & Software, available online at http://dx.doi.org/10.1016/j.bbr.2011.03.031
Pekkarinen, A., Reithmaier, L., Strobl, P. (2009): Pan-European forest/non-forest mapping with Landsat ETM+ and CORINE Land Cover 2000 data. ISPRS Journal of Photogrammetry and Remote Sensing 64: 171-183.
Service, R.F. (2011). Coming soon to a lab near you: drag-and-drop virtual Worlds. Science, 331(6018):669-671.
Simonetti, D., Beuchle, R., Eva, H.D. (2011). User Manual for the JRC Land Cover/Use Change Validation Tool. EUR ÛÒ Scientific and Technical Research series ÛÒ ISSN 1018-5593 . Luxembourg: Publications Office of the European Union. ISBN 978-92-79-18986-9
Tomppo, E., Schdauer, K., McRoberts, R., Gschwantner, T., Gabler, K., StÌ´hl, G. (2010) Introduction. In National Forest Inventories ÛÒ Pathways for Common Reporting. Eds. Tomppo, E., Gschwanter, T., Lawrence, M., McRoberts, R. Springer Heidelberg, 607pp.
Vidal, C., Lanz, A., Tomppo, E., Schadauer, K., Gschwanter, T., di Cosmo, L., Robert, N. (2008) Establishing forest inventory reference definitions for forest and growing stock: a study towards common reporting, Silva Fennica, 42 ( 2), 247ÛÒ266.
Lucy Bastin was a visiting senior scientist at EC-JRC from 2010-2011, working on the generation of Web-based validation tools for use by an international community of experts in biodiversity and forestry monitoring. She holds a senior lectureship in the School of Engineering and Applied Science at Aston University, U.K., where she applies spatio-temporal analysis techniques to challenges in conservation planning, infection monitoring, and other environmental and socio-demographic contexts.