Case Study #2: A DSS for Flash Flood

Transcription

1 S. Nativi, P. Mazzetti, L. Bigagli and M. Mancini Case Study #2: A DSS for Flash Flood Bringing together Geo-science systems and Decision Makers Ontology mapping issue Expanding Horizons 2003 UNIDATA/UCAR Boulder (CO) June 2003

2 Rationale - Flash-Flood Early Warning estimate is a complex task requiring a wide set of geospatial data providing information about current and future values of heterogeneus parameters (hydrological and not) - Most common sources Raingauge networks Satellites Weather radars Local Area Models GIS - Generally speaking, Decision Makers are NOT computer scientists or Hydraulics Engineers. - A DSS to be effective needs to express results according to Decision-Makers ontology.

3 The Service- Oriented Approach USER COMMUNITIES Device Adaptation services Current Situation Presentation Future Situation Presentation Single Situation Analysis DEVICE RENDERING Services HUMAN INTERACTION Services Information Resources Synchronising Services Heterogeneous Content Consolidation Services CO-OPERATION Services Hydrological Parameter Generation Services Decision Making Parameter Generation Services PROCESSING/TASK Services Protocol Adapting Services (to Geo-RI Interfaces) Binding Adapting Services (e.g. Marshalling services) MODEL/INFORMATION MANAGEMENT Services DATA/INFORMATION Weather Datasets GIS Datasets Geo DATA PROVIDERS Forecasted Datasets

4 The Starting Scenario: main issues Organization: - Data sources were managed by many different data providers; Technology: - Data were accessed through different communications protocols and they were encoded in proprietary formats; Content: - No specific pre-processing for flash flood early warning was available; Semantic: - Information was expressed according to data providers (scientists, researchers, ) or hydrologists conceptual models; Presentation: - Data were visualized through legacy applications that were not specifically designed for flash flood early warning problem;

5 The Starting Scenario: overview Decision-makers can access a lot of information, but they can utilize only a small part of them, effectively.

6 Our approach for a DSS The DSS is made up of two main components. Raingauge Network (Measured precipitation) MARTE network The first component act as an Expert System: Presentation and Graphical Enrichment Info Presentation System Situation Report Expert System LAM (Forecasted precipitation) Ancillary Data (GIS layers) RAMS Model Tuscany Region GIS 1. It extracts useful information from data sources; 2. It integrates and processes the information (according to a specific model); 3. It provides a standard and open situation report.! Enriched and Contextualised Situation Rendering A second component acts as a Presentation System: Decision Maker 1. It carries out proper presentation of the situation report. 2. It enriches the presentation with useful graphical and presentation features. 3. It contextualises the rendering according to the client device configuration

7 An operational implementation: the MIMÌ System MARTE network MARTE network Raingauge Network (Measured precipitation) Situation Report RAMS Model??? RAMS model Presentation and Graphical Enrichment Info Portal System Expert System LAM (Forecasted precipitation) Ancillary Data (GIS layers) Tuscany Region GIS Ancillary Data (GIS layers)! Enriched and Contextualised Situation Rendering Decision Maker Decision Maker -MARTE system: a ground-based sensors network providing data every 15/30 minutes, in a legacy format. It is managed by the National Hydrographic Service. -RAMS system: a Local Area Model providing forecasted precipitation data for the Tuscany Region. It works out one map per hour within 48 hours forecasting time. Maps are provided in legacy binary format along with a metadata text file. It is managed by the Regional Laboratory for Applied Meteorology.

8 The three Ontology Views Domain specifc Ontologies Hydrology Ontology Decision-makers Ontology RAMS forecasted precipitation data MARTE measured precipitation data Hydrogeologic parameter extraction Decision-making parameter extraction DSS Client Application Expert System other data

9 Domain Ontology To be effective the system needs to express results according to the Decision-Makers ontology. Generally speaking, Decision Makers are NOT computer scientists or Hydraulics Engineers Portal Portal Antecedent Moisture Content Sub-basin section status Rainfall Event shape?? th Qtr 3rd Qtr 2nd Qtr 1st Qtr Alarm Threshold! Situation Report 0 East West North Decision Maker Decision Maker Not effective Effective

10 The Ontology Integration issue Hydrology Conceptual Model (Simplified View) DSS Conceptual Model (Simplified View)

11 Hydrogeological Model An Hydrogeological Model developed by the Polytechnic of Milan is the core component of the expert system. It defines the procedures to extract information from the input data and generate the hydrological outputs which are mapped to the DM s concepts (Conceptual Model). - Briefly, the expert system tasks are: - To update the situation (sub-basins critical sections status) each time a measured map (from MARTE) or a forecasted map (from RAMS LAM) is available; - To aggregate the rainfall on each sub-basin; - For each sub-basin, to check the existence of a rainfall event according to the value of the aggregated rainfall; - For sections with active events, to compute and compare the cumulated rainfall with a threshold, based on hydrological parameters; - To manage sectionstatus, each section is in one of three possible status (no event, event without warning, event with warning); - To achieve the model output: a summary of the situation (sections status, rainfall series, )

12 Computational Architecture - Component-based approach; - Multiplatform approach; - Extensible solution; - Distributed solution Presentation System Expert System Content Collector Presentation Adapter Client Application Middleware Info/Data Resources

13 MIMÌ Technology baseline Adopted technologies: - Java Platform (J2SE, J2EE, J2ME, Java Web Start); - Web Service technology XML/XSD; XSLT; SOAP/HTTP(S); HTTP/HTTPS; WSDL; - SVG;

14 Raingauge Network Data (Real Time Data) Local Area Model (Forecasting Data) GIS Data (Time-invariant Data) Legacy Data MARTE System RAMS System National Authority GIS Hydrographic National Centre Web Server (Situation Information Publisher) Intranet Legacy Data Legacy Data Hydrological Expert System (Situation Information Generator) Web Servers (Presentation Information Publishers: e.g. River Network, DEM, etc.) Tuscany Regional Meteo-Laboratory National Arno Basin Authority Hydrogeological Model Polytechnic of Milan Expert System Presentation System Contestualised Presentation Server HTML SVG DSS Implementation - Expert System Hydrological Model Implementation Ontology Mapper - Presentation System Content Collector Contestualised Presentation Adapter - Client Application Thick Client Thin Client Mobile Client Thick Client Application (Expert Decision-makers) Thin Client Application (Decision-makers)

15 Raingauge Network Data (Real Time Data) Local Area Model (Forecasting Data) GIS Data (Time-invariant Data) Legacy Data MARTE System RAMS System National Authority GIS Hydrographic National Centre Web Server (Situation Information Publisher) Intranet Legacy Data Hydrological Expert System (Situation Information Generator) Web Servers (Presentation Information Publishers: e.g. River Network, DEM, etc.) Tuscany Regional Meteo-Laboratory Legacy Data National Arno Basin Authority Hydrogeological Model Polytechnic of Milan Expert System Presentation System Contestualised Presentation Server HTML SVG DSS Implementation - Expert System Hydrological Model Implementation Ontology Mapper - Presentation System Content Collector Contestualised Presentation Adapter - Client Application Thick Client Thin Client Mobile Client Thick Client Application (Expert Decision-makers) Thin Client Application (Decision-makers)

16 MIMÌ Human-System Interaction Model The Java-based thick client is specifically designed for decisionmakers showing DMs ontology information in a user friendly manner. The GUI is based on typical use-cases defined in collaboration with the final users; In particular: -Only the sections status for a given time is normally visible; -User can select the reference time choosing to view the measured situation or one of the forecasted situations; -Uuser can inspect a specific section obtaining charts of cumulated and aggregated rainfall

17 Thick Client 1) Login

18 Thick Client 2) View the real-time measured situation Critical Sections Its satus is reported by the shape and the colour

19 Thick Client 3) Select and view a real-time forecasted situation Available Forecasted Situations Just select one and the forecasetd scenario is diplayed

20 Thick Client 4) Inspect a basin section for a potentially dangerous forecasted situation Aggregated Rainfall Cumulated Rainfall Green = forecasted Blue = measured Threshold

21 Smart Phone Client J2ME enabled Device

22 Cell-Phone Client XHTML enabled Device The present situation At the Subbasin level The next forecasted situation At the Subbasin level

23 S. Nativi, P. Mazzetti, L. Bigagli and G. Villoresi University of Florence and IMAA-CNR The Presentation System To leverage interoperability and extend the DSS Decoupling Content from Presentation Info Towards the Cooperative work Expanding Horizons 2003 UNIDATA/UCAR Boulder (CO) June 2003

24 To decouple Content and Presentation information Content Information Raingauge Network Data (Real Time Data) Legacy Data Local Area Model (Forecasting Data) Legacy Data Web Server (Situation Information Publisher) GIS Data (Time-invariant Data) MARTE System RAMS System National Authority GIS Hydrographic National Centre Intranet Tuscany Regional Meteo-Laboratory Legacy Data National Arno Basin Authority Hydrogeological Model Polytechnic of Milan Hydrological Expert System (Situation Information Generator) Expert System Presentation Information Web Servers (Presentation Information Providers: e.g. River Network, DEM, etc.) Presentation System Contestualised Presentation Server Thick Client Application (Expert Decision-makers) Thin ClientApplication Mobile Client Application (Decision-makers)

25 Sub-basin Situations Raingauges Situations Expert System Information Resources Device-depended Rendering Basin critical sections Metadata Portal System Basin DEM Basin River Network Collector Adapter Contestualised Presentation Server Web Servers SVG HTML Tag-language Java Object Thick Client Application (Expert Decision-makers) Thin ClientApplication Mobile Client Application (Decision-makers) Client Applications

26 Expert System Information Resources Device-depended Rendering Web Servers Collector Adapter SVG HTML Portal System Content Info Assessment Contestualised Presentation Server Content + Presentation Info Rendering Tag-language Java Object Thick Client Application (Expert Decision-makers) Thin ClientApplication Mobile Client Application (Decision-makers) Client Applications

27 XSLT XSLT Sub-basin Situations Collector Adapter Raingauges Situations Content Info Basin critical sections Metadata ConsolidatedReport SVG SVG Basin DEM Portal System Presentation Info Basin River Network Computational Components HTML SVG J <TAG> Thick Client Application (Expert Decision-makers) DeviceConfiguration Thin ClientApplication Mobile Client Application (Decision-makers) Client Applications Web Service HTTP Service

28 Implemented Presentation Services - Global Situation Service The presentation of the present and future situations for all the basin sections - asituation Service The presentation of a given situation for all the basin sections - asection Evolution Service Only Thick and Thin Clients All Clients All Clients The presentation of the present and future situations for a given basin section

29 XSLT GlobalSituation Collector SVG SVG Portal System Global Situation Service XSLT Thick Client Adapter GlobalReport XSLT Thin Client Adapter Presentation Proxy Thin Clients need an HTTP-based Presentation Proxy HTML SVG Thick Client Application (Expert Decision-makers) DeviceConfiguration Client Applications Thin Client Application (Decision-makers) Web Service HTTP Service

30 Client Device SOAP message Demo Components Deployment Web Browser Any Device Web Browser HTTP/GET/POST Message Presentation Proxy (TOMCAT) port: 8080 Adapter Web Service Adapter Service Proxy Situation Report Servlet Servlet Server Servlet Server Adapter Application Collector Web Service Expert Proxy (TOMCAT) port: 8080 Situation Report Service Manager Presentation Server port: 8004 Situation Report Collector Collector Application Expert System Proxy Expert System Application Web Services Server

31 - Web Service SOAP Message XML/XSD - HTTP Service HTTP GET/POST Message Low-level Access Services SOAP Engine Web Service Operative System Adapter Application Web Browser Servlet Engine Web Service Proxy SOAP message HTTP/GET/POST Message

32 Mobile Clients Demo - Demo: asituationservice - Client application is any device Web Browser - Presentations are personalised for: PDA Smart Phone Cellular Phone

33 Exercises 1. Use a general purpose SOAP Client to access the Collector and Adapter web services Add a Presentation Resource (SVG fragment) The Tuscany Region boundary