Interoperability and architectures Simon Cox CSIRO Exploration and Mining 23 May 2006

2 of 71 NR Interoperability Outline Precedent – the SEE Grid demonstrator/use-case WFS refresher The Natural Resources puzzle SOA model - publish-find-bind?  Registries are out-of-scope for this workshop  focus only on bind leg  “informational service interface profile”

4 of 71 NR Interoperability

5 of 71 NR Interoperability

6 of 71 NR Interoperability

7 of 71 NR Interoperability

8 of 71 NR Interoperability

9 of 71 NR Interoperability

10 of 71 NR Interoperability

11 of 71 NR Interoperability

12 of 71 NR Interoperability

13 of 71 NR Interoperability

14 of 71 NR Interoperability

15 of 71 NR Interoperability

16 of 71 NR Interoperability

17 of 71 NR Interoperability

18 of 71 NR Interoperability

19 of 71 NR Interoperability Status quo: Different formats and standards Data Structures Proprietary Software Versions of Software User

20 of 71 NR Interoperability The vision: Same format and standards User GML for Natural Resource Information NRML

21 of 71 NR Interoperability Standard informational service profiles Business case revolves around:  Existence of multiple sources of same information types  Users want to replace or merge data sources, re-compose service chains, on demand at run time  Providers are willing to cooperate  regulatory push  market pull Technical requirements  non-intrusive  discoverable  IPR protection, access and authentication control, accounting  orderly governance framework

22 of 71 NR Interoperability The mechanism: OGC Web Feature Service GML WFS Server Data-source organised for custodian’s requirements Community-specific GML application language  TigerGML, LandGML, GeoSciML, CSML, MarineXML, NRML etc private  public boundary WFS Client HTML

23 of 71 NR Interoperability Standard transfer format allows multiple data sources WFS Client WFS Server WFS Server B WFS Server C

24 of 71 NR Interoperability PIRSA Web Feature Service (WFS) Common Interface Binding – GML/XMML GA Geochemistry Feature Data Source DOIR Geochemistry Feature Data Source DOIR Web Feature Service (WFS) GA Web Feature Service (WFS) Geoserver (Open Source) PostGIS (Open Source) Oracle PostGIS (Open Source) CLIENT APPLICATIONS DATA ACCESS SERVICES DATA SOURCES WebMap Composer GA Reports Application PIRSA Geochemistry Feature Data Source Summary architecture

25 of 71 NR Interoperability Data re-use WFS Client WFS Server multiple views of same data (features) SOS Client SOS Server (observations) WCS Server (coverages) WCS Client

26 of 71 NR Interoperability Combining with observation service WFS Client WFS Server WFS/SOS Client/ Server (orchestration) WFS Client (simulation) (mapping) WFS Client (analysis & reporting) SOS Server Sensor BPEL?

27 of 71 NR Interoperability Service implemented as a set of operations  “REST” - request-response message pairs  carried over http Fine-grained  ~ features GetFeature request  feature type  properties of interest – projection clause  filter condition – selection clause Web Feature Service WFS details

28 of 71 NR Interoperability Report client Many servers, one report

29 of 71 NR Interoperability PIRSA DOIR GA Many servers, one map Mapping Client

30 of 71 NR Interoperability

31 of 71 NR Interoperability Combine their data with your data

32 of 71 NR Interoperability GML-based data can be …. Rendered into a queryable map … … formatted into a report or …. … read and used by any WFS/GML enabled application Pre-requisite: community-standard data model/encoding One service – many uses

33 of 71 NR Interoperability The Natural Resource Information Infrastructure puzzle (some pieces) Data Bases / Systems Information Products Information Products Information Products XML GML document Style Layer descriptor “ARO” application Web services WMS WFS Schema.XSD Processes Feature Catalogue UML Abstract model Sensor

34 of 71 NR Interoperability Information Products Information Products Information Products Support Business needs NRM planning & Investment decisions Are created by Processes That use Data Bases / Systems Are modelled by UML Abstract model Are described in Information Product Template (metadata) Feature Catalogue That specifies Is implemented by Schema.XSD “ARO” application Sends request / query Returns response XML GML document Is validated by Accesses Web services WMS WFS Used to describe Style Layer descriptor Is displayed with Are accessed through Sensor Are fed by Workshop Scope

35 of 71 NR Interoperability Use-cases define requirements Who are the users?  give them names! What do they want?  tell a detailed story, sketch wireframes  “first X, then Y, visualized this way, allowing selection of Z” Analyze the requests and data flow to determine the objects and features of interest

Domain modelling Simon Cox CSIRO Exploration and Mining 23 May 2006

37 of 71 NR Interoperability Geographic Information Models Viewpoints: Features, coverages and observations

38 of 71 NR Interoperability How the world is seen: Fields vs. objects mapped-objects “Feature” viewpoint earth-observations “Coverage” viewpoint

39 of 71 NR Interoperability 1.Classic geometry-centric GIS approach One shape per feature …  well suited to drawing maps  strictly an implementation strategy for portrayal  hides the (semantic) business object Tenement Points, lines and polygons are primary objects  Annotated with a set of attributes, often a row of scalar values  layer name => semantics

40 of 71 NR Interoperability 2. Conceptual object model: features Digital objects correspond with identifiable, typed, objects in the real world  mountain, road, specimen, event, tract, catchment, wetland, farm, IBRA, bore, reach, property, lease-area Feature-type is characterised by a specific set of properties Specimen  ID (name)  description  mass  processing details  sampling location  sampling time  related observation  material ……

41 of 71 NR Interoperability Variation of a property across the domain of interest  For each element/position in a spatio-temporal domain, a value from the range can be determined Discrete or continuous domain  Domain is often a grid  N.B. Arc/Info “coverage” == multi-geometry domain, attribute-table for each element 3. Spatial function: coverage (x 1,y 1 ) (x 2,y 2 )

42 of 71 NR Interoperability Feature properties For “normal” features, the value of each property is constant on the feature spatial values are just another property  multiple geometries possible, e.g. scale dependence, dimensionality Some properties may vary across the feature  property value is described by a coverage Specimen  ID (name)  description  mass  processing details  sampling location  sampling time  related observation  material ……

43 of 71 NR Interoperability Cross-sections through information SpecimenAu (ppm) Cu-a (%)Cu-b (%)As (ppm)Sb (ppm) ABC A Row gives properties of one feature A Column = variation of a single property across a domain (i.e. set of locations) A Cell describes the value of a single property, often obtained by observation or measurement

44 of 71 NR Interoperability 4. Property-value estimate: Observation An Observation is  an Event  time, location, responsible-party  whose result is an estimate of the value of some Property  = Determinand, Measurand, Analyte, Phenomenon  of its Feature-of-interest  e.g. tract, catchment, sampling-station, specimen  obtained using a specified Procedure  e.g. sensor, algorithm, model, process-chain, simulation

45 of 71 NR Interoperability Observation model – Value-capture-centric view An Observation is an Event whose result is an estimate of the value of some Property of the Feature-of-interest, obtained using a specified Procedure

46 of 71 NR Interoperability Feature-of-interest centric view Specimen  ID (name)  description  material  mass  processing details  sampling location  sampling time  related observation  … Properties  attributes  association rolenames

47 of 71 NR Interoperability Observation model Defines the terminology for linking items of interest in observational science

48 of 71 NR Interoperability Observations and coverages If the property value is not invariant across the feature-of-interest  varies by location, in time the corresponding observation result should be a coverage individual values must be tied to the location within the domain, so the result will be an array of e.g.  time-value  position-value  stationID-value …

49 of 71 NR Interoperability Domain modelling Primarily involves domain-specific specializations of the second-layer (wrt Observation model):  Features-of-interest  feature-type catalogue for the domain  Determinands/properties  property dictionary or ontology  Procedures  standard procedures A common vocabulary for these domain types must be adopted by the community

50 of 71 NR Interoperability Example: AWDIP model

51 of 71 NR Interoperability Information service requests Information services will be based around the domain feature- types e.g. “please tell me about”:  all properties of that tract/catchment/well  i.e. the values of all associated observations  the history of salinity at that station  i.e. the salinity time-series/temporal coverage with the station as the FoI  the reliability of that salinity value  i.e. the procedure and sampling parameters for the observation

52 of 71 NR Interoperability Several views of the same information SpecimenAu (ppm) Cu-a (%)Cu-b (%)As (ppm)Sb (ppm) ABC Result/Observation view Used for:  Quality/confidence assessment  Result calibration  Database insertion and update Feature view Assembled by:  Aggregation of multiple Observation/result having same featureOfInterest Used for:  Object description Coverage view Assembled by:  Using suitable sampling regime on feature-of-interest  Aggregation of multiple Observation/result having same observedProperty Used for:  Property variation  Pattern/anomaly/feature detection

53 of 71 NR Interoperability Processing/value-adding chains make homogeneous Observations according to a sampling regime  pixels, stations, clock-ticks assemble results into a discrete Coverage  image, log, time-series detect a Feature  anomaly, intelligence Observations initiate chain leading to feature detection make heterogeneous Observations on a pre-existing feature  patient, artefact, catchment assemble results into description of set of Feature properties Observations provide metadata on property values

54 of 71 NR Interoperability Section view Sometimes they appear together Properties Map view  Survey  Continuous logs  Intervals  Horizons  Point observations Position is 1-D  arc-length from collar  Collar (Point Property)  Shape (Curve Property)  Samples

55 of 71 NR Interoperability Which viewpoint? What are you interested in? … describing a discrete object ? - Feature  property set characterizes feature type  associated observations provide property metadata … variation of property within domain ? - Coverage  domain = feature of interest  e.g. tract, swath, time-period  associated observations provide results to populate coverage … data acquisition event/process/quality ? - Observation  explicit procedure, capture time

56 of 71 NR Interoperability OGC Information Service interfaces Web Map Service Catalog Service Web Feature Service Web Coverage Service Sensor Observation Service Sensor Planning Service (tasking) Sensor Alert Service (subscription) Web Processing Service

57 of 71 NR Interoperability Value-adding chain Observation  estimate of value of a property for a single specimen/station/location  data-capture, with metadata concerning procedure, operator, etc Coverage  compilation of values of a single property across the domain of interest  data prepared for analysis/pattern detection Feature  object having geometry & values of several different properties  1. classified object, snapshot for transport  geological map elements  2. object created by human activity, artefact of investigation  borehole, mine, specimen

Formalizing the model using UML as the conceptual schema language Simon Cox CSIRO Exploration and Mining 24 May 2006

59 of 71 NR Interoperability Model of a specific feature-type Specimen  ID (name)  description  material  mass  processing details  sampling location  sampling time  related observation  … Static properties  attributes  association rolenames

60 of 71 NR Interoperability GML serialization Specimen collected in Dales Gorge, Karijini Dales Gorge specimen #1 rock T17:22:25.00

61 of 71 NR Interoperability UML – Class Diagrams Information Structure c.f E-R, plus  several distinct varieties of relationships  associations  association roles  association direction  inheritance  realization  aka interface  class attributes may have complex type

62 of 71 NR Interoperability Sampling features hierarchy

63 of 71 NR Interoperability UML extension mechanism - stereotypes Specialized meta-elements (Class types) > (ISO 19103/19136 profile)

64 of 71 NR Interoperability UML extension mechanism – constraints OCL may be used to express additional constraints

65 of 71 NR Interoperability UML extension mechanism – Tagged values Formal annotations – typically implementation details  Also see package tagged values

66 of 71 NR Interoperability UML/GML ISO & ISO provide 2 things: modelling pattern  based on General Feature Model  implemented as a profile of UML and usage patterns  which leads directly to an XML Schema components  base classes  implementation of utility classes & data-types from ISO 19107, 19108, 19111, 19123

67 of 71 NR Interoperability “GML Application Schema” follows patterns  rolenames on associations  navigability & cardinality  etc utilizes base classes and utility components  > → specializes gml:AbstractFeature  > → specializes gml:AbstractGML  incl. gml:AbstractGeometry  GenericName → gml:CodeType  GM_Object → gml:AbstractGeometry  etc

68 of 71 NR Interoperability Standard packaging – Hollow World UML Template Pre-loaded with the ISO series models and others for which canonical GML or GML-conformant implementations are available  ISO – basic data types  ISO – geometry & topology  ISO – temporal objects & reference systems  ISO – spatial position & coordinate reference systems  ISO – metadata (mostly dataset-oriented)  ISO 19136/GML  SweCommon, O&M, Sampling  Geo – solid geometry elements Configuration file to map UML elements to their canonical XML Schema representation  various namespaces

69 of 71 NR Interoperability Direct GML serialization <om:Observation gml:id="obsTest1" xmlns:om=" xmlns:xsi=" xmlns:xlink=" xmlns:gml=" xsi:schemaLocation=" Observation test instance Observation test T16:22:

70 of 71 NR Interoperability Challenges Non-invariant properties  encoding “coverages” in the result  array of  time-value  position-value  stationID-value Establishing standard catalogues/dictionaries:  Properties/Phenomena  Procedures/Instruments  Record schemata Patterns for specialization  Assay-measurement  WQ-observation  Met-observation

Thank You CSIRO Exploration and Mining NameSimon Cox TitleResearch Scientist Phone Webwww.seegrid.csiro.au Contact CSIRO Phone Webwww.csiro.au