1 SDI and VGI Parallel Universes? Max Craglia Institute for Environment and Sustainability Spatial Data Infrastructures Unit
2 Background There is a growing awareness that we are living at a time when environmental changes have significant impacts on our economy and social well being, and urgent action is required. Understanding the complex interactions between natural and human systems requires reliable and timely spatial information. Sea Level trends in mm/y Forest Fire Risk Floods Risk Source: Marcos & Tsimplis, as quoted in JRC/IES Source: JRC/IES Source: JRC/IES
3 What is a Spatial Data Infrastructure A framework of data, technology, policies, standards, and human resources, necessary to facilitate the sharing and using of spatial information. The term infrastructure is used to emphasise not just hardware and data (equivalent in the rail system to carriages, power lines, rail tracks, stations) but also the need for coordinating structures and international standards and agreements (on gauges, timetables, safety rules, signalling, etc.) without which the system cannot operate consistently and safely. http://europa.tiscali.it/futuro/speciali/quiz_giovani/374123859quiz.html
4 Example of why we need an SDI: Impacts of Flooding In the period 1998-2002 floods comprised 43% of all disaster events in Europe 100 major floods 700 dead Half a million displaced people 25 billion Euros uninsured economic loss Along the Rhine, 10 m people live in areas liable to extreme flooding, potential damage estimated at 165 bn. Euros 101,000 kms of coastline, population doubled in last 50 years. Assets within 500 mt of coast = 500-1000 bn euros.
Environmental phenomena do not stop at national borders! 20% of the EU citizens (115 million) live within 50 Kms from a border. 5 60 million EU citizens live less than half an hour (25 kms) from a border Near - boundary population importance 115 M 82 M 70 M 64 M 60 M 60 M 59 M 45 M 39 M 23 M 22 M 16 M 12 M POP at 50 Km Germany POP at 30 Km France United Kingdom POP at 25 Km Italy Spain Poland POP at 10 Km Romania Netherlands POP at 5 Km
Water Framework Directive 6 Moving from a national perspective
.the environmental perspective! 7 River Basins
Good scientific principle, but 8 70% of all fresh water bodies in Europe are part of a trans- boundary river basin 801.463 km², 81 million inhabitants, 19 countries
9 INSPIRE Directive General Provisions INSPIRE lays down general rules to establish an infrastructure for spatial information in Europe for the purposes of Community environmental policies and policies or activities which may have an impact on the environment. INSPIRE to be based on the infrastructures for spatial information established and operated by the Member States. INSPIRE does not require collection of new spatial data INSPIRE does not affect existing Intellectual Property Rights
INSPIRE Components Metadata 10 Interoperability of spatial data sets and services Network services (discovery, view, download, invoke) Data and Service sharing (policy ) Coordination and measures for Monitoring & Reporting INSPIRE is a Framework Directive Detailed technical provisions for the issues above will be laid down in Implementing Rules (IR) JRC is responsible for overall technical coordination of INSPIRE
INSPIRE Spatial Data Scope 11 Annex I Coordinate reference systems Geographical grid systems Geographical names Administrative units Addresses Cadastral parcels Transport networks Hydrography Protected sites Annex II Elevation Land cover Ortho-imagery Geology Harmonised spatial data specifications more stringent for Annex I and II than for Annex III
12 Annex III Statistical units Buildings Soil Land use Human health and safety Utility and governmental services Environmental monitoring facilities Production and industrial facilities Agricultural and aquaculture facilities Population distribution demography Area management/restriction /regulation zones & reporting units Natural risk zones Atmospheric conditions Meteorological geographical features Oceanographic geographical features Sea regions Bio-geographical regions Habitats and biotopes Species distribution Energy Resources Mineral resources
13 SDI + VGI Research Issues and Challenges Metadata: generate + user feedback to assess quality starts overcoming current limitations of data producer view of the world Interoperability: additional challenges of multilingual/multicultural current solutions of top down harmonisation not appropriate? Alternatives? Service chaining: The answer to turn data into information? Challenge of operating on even more heterogeneous data? What type of services would be useful? How to choose your plume model?
Data in common model Example of Service Chaining for Forest Fires Transformation into common model European common model cd Data Model 14 ForestFireRegistration_Common DB representation (Portuguese data) + fireid: char + dateal: char + timeal: char + timein: char + datein: char + dateex: char + timeex: char + codecom: char + nuts3: char + namecom: char + tba: char + fba: char + nfba: char + cause: char Feature Access Service DB representation (Spain data ) Schema Transformation Service Feature Access Service
Data in common model 15 Aggregation Service Counts number of fires within each NUTS unit Classification system Classification Service Nuts data Map Service Symbology Rendered map with legend
16 Example 2: fully automated preliminary classification of multi-spectral satellite imagery It belongs to the public domain: A. Baraldi, V. Puzzolo, P. Blonda, L. Bruzzone, and C. Tarantino, "Automatic spectral rule-based preliminary mapping of calibrated Landsat TM and ETM+ images," IEEE Trans. Geosci. Remote Sensing, vol. 44, no. 9, pp. 2563-2586, Sept. 2006 Fully automated, i.e, it requires: a) no free parameter and b) no reference (supervised) data set of examples (i.e., no ground truth data) Push-and-go (press-and-run) button-like implementation. Almost Real time (2 minutes per Landsat scene) Spectral prior knowledge-based, exclusively. No inductive data learning mechanism, neither unsupervised (e.g., data clustering) nor supervised (e.g., data classification). Non-iterative (1-pass classification). It is based on prior spectral knowledge driven from a dictionary of real-world spectral signatures in planetary reflectance which account for atmospheric effects. Suitable for mapping multiple sensors Landsat 5 TM and 7 ETM+-like imagery, SPOT-4/5-like imagery, AVHRR-like imagery IKONOS-like (e.g., QuickBird-1, OrbView-3, etc.)
Output map product types 1 to 3: Standard preliminary classification maps featuring different degrees of informational granularity: 72, 38, or 15 spectral categories in the Landsat system version (49, 32, or 13 in the SPOT system version and 70, 37, 15 in the AVHRR system version, respectively). Label indexes are monotonically decreasing with category-specific biomass estimates (label 1: SVVHNIR, label 2: SVHNIR, etc.). High 17 Input image 72 SCs 38 SCs 15 SCs Output value-added product types 4 to 7: Value-added products suitable for stratified second-stage supervised classification, image segmentation, data clustering. a b c d a) (Novel) Greenness b) Canopy Water Content c) Canopy Chlorophyll Content (NDVI) d) Water/Shadow spectral index Low 17
Application domain example A: Papua. Spectral classification map pair difference. Landsat-7 ETM+ imagery. 18 CLASSIFICATION MAP PAIR DIFFERENCE in {0, 23} CLMap1999 CLMap1993 Discretized label pair similarity map Input image 1: Landsat 5 TM image of Papua (acquisition date: 1993). Observation: Preliminary classification map pair difference CLMap1999 CLMap1993 It is possible because the classification map indexes decrease monotonically with biomass estimates. Input image 2: Landsat 7 ETM+ image of Papua (acquisition date: 1999). Fuzzy quantized similarity index, Legenda 18
19 Service-centred data infrastructure is crucial But we are still at the early stages Few services available beyond WMS, WFS, etc. Classification of services still to be developed Documentation of services still too poor, not allowing choice between competing services, nor understand quality or fitness for purpose. Yet without the move towards services, no way of opening SDI to wider audiences who are not interested in data but in information. Last but not least
20 Policy and impacts: pressure of VGI on existing data pricing policies + opportunity provided by VGI to measure impact (identification of community of users) How can we foster synergy SDI-VGI? i.e. make the parallel lines meet sooner rather than later?
21 http://ijsdir.jrc.it Published only On-Line, free access Creative Commons Immediately published in the Under Review Section
22 Thank you for your attention! Massimo.craglia@jrc.it For more information on INSPIRE: www.ec-gis.org/inspire