Large catastrophes often trigger international humanitarian response. This is a particular context in which many independent actors, including governmental agencies (e.g. search and rescue teams), non-governmental organizations (NGO’s such as Doctors Without Borders), corporations (e.g. Google or Microsoft) and international organizations (including the United Nations Office for Coordination of Humanitarian Affairs) work together to provide first response and subsequent relief and reconstruction assistance. In the absence of a clear command and control structure, situational awareness needs to be acquired by each actor independently. Needless to say that this community is eager to develop and use technology and systems to acquire and share information, and that collaboration and information sharing is generally considered as mutually benefitting.
In the early onset of disasters, information is sparse. Traditionally, there are three main sources of information: scientific monitoring systems (e.g. seismological or meteorological networks), official information (briefings by the local emergency management agency) and media reports. Information management for each source requires different technological solutions, respectively focused on modelling, web portals for information sharing, and linguistic processing. However, more recently a fourth source of information is becoming available through Web 2.0: information from citizens, sometimes labelled crowd-sourcing. In case of a disaster, local (and remote) citizens can and do provide information (e.g. eyewitness reports) or analysis (e.g. compiling reports in an information feed). However, this fourth source is not widely used yet by emergency managers because the reliability of the information is not well understood and hard to assess in a time-critical environment.
Iscram Summer School 2009 From Mashups To Modelling (Tom De Groeve)
1. From mash-ups to modelling: technology for international crisis situation awareness ISCRAM Summer School 2009 Tom De Groeve – tom.de-groeve@jrc.ec.europa.eu Joint Research Centre of the European Commission 27 August 2009 ISCRAM Summer School 2009, Tilburg
2. ISCRAM Summer School 2009, Tilburg European Commission European Commission Directorates General (per policy area) Humanitarian Aid (DG ECHO) External Relations (DG RELEX) Environment and Civil Protection (DG ENV) Research (DG JRC) Research policy and funding (DG RTD) Development aid (DG DEV, DG AIDCO) … Directorate General Joint Research Center 7 institutes (per research area) 2500 staff Institute for the Protection and the Security of the Citizen IT, Engineering, Statistics, Geomatics Global Security and Crisis Management
3. ISCRAM Summer School 2009, Tilburg European Commission European Commission is a large player in international disaster management Policy and action for disasters Humanitarian aid office (ECHO): largest humanitarian aid donor in the world Civil Protection Mechanism Coordination of European first response Monitoring and Information Center (MIC) External Relations: Prevention, mitigation, reconstruction programmes Development: Reduction of vulnerability and improvement of resilience Humanitarian aid Civil Protection Mechanism Reconstruction
4. European Commission Role of the European Commission Principle of subsidiarity Need for specific mandate at European level Intermediary between multilateral organisations and EU Member States European institutions European Council Heads of state European Parliament European Court of justice European agencies European Centre for Disease Control ISCRAM Summer School 2009, Tilburg Multilateral org’s European Commission EU Member states
5. European Commission Natural disasters ISCRAM Summer School 2009, Tilburg Affected government Assessment UNDAC Response OCHA Response org (e.g. NGO, IFRC) Funding Coordination ECHO (funding) EC Monitoring and Information Centre EU Member state Search & Rescue
6. European Commission Health ISCRAM Summer School 2009, Tilburg WHO Data aggregation ECDC Coordination European Commission DG SANCO Surveillance Policy making EU Member state
7. Lecture overview International situation rooms Information management roles, capabilities and needs Detection, analysis, briefing, action Information management tools Mash-ups versus Spatial Data Infrastructure Access, analyze and share information Information modelling Make information work for your business processes ISCRAM Summer School 2009, Tilburg
8. Part 1 International Situation Rooms Roles, capabilities and needs ISCRAM Summer School 2009, Tilburg
9. International situation rooms European Institutions ECHO: humanitarian aid Humanitarian aid funding Civil Protection Mechanism (MIC) International search and rescue Cross-border European disasters Multi-lateral assistance External Relations (RELEX) Community level foreign affairs Reconstruction after war or disaster European Council Political foreign affairs response Health (SANCO) Epidemic control measures Border control (FRONTEX) Border monitoring NATO African Union World Bank United Nations Office for Coordination of Humanitarian Affairs Sudden onset response coordination UNDAC: disaster assessment World Food Program Early warning for food crises Response to food crises Peacekeeping Operations Pre or post conflict World Health Organisation Pandemics International Non Governmental Org’s International Federation of Red Cross National (non) governmental org’s Urban Search and Rescue: USAR.nl Italian Civil Protection ISCRAM Summer School 2009, Tilburg
10. International situation rooms Roles Primarily an advisory role, rather than an operational role Information processing for situational awareness and political response. Wide Scope Geographically: continental or global Thematically (any kind of crisis) Tasks (Support) rapid decisions for sudden onset disasters or crises Regular briefs on slow onset or continuous crises Staff Generalists or specialists in response Experience in decision making, less in information management Small number of staff, rarely working in 24h shifts Information No access to intelligence information Sometimes access to private network of experts (delegations, regional offices, roster of experts) Operation Not much routine because of the variety of crisis situations No or little standard operating procedures ISCRAM Summer School 2009, Tilburg
11. ExampleGlobal Disaster Alert and Coordination Systemwww. .org an example of a disaster alert and impact system for international humanitarian assistance ISCRAM Summer School 2009, Tilburg
12. International humanitarian aid A complex system with many stakeholders No “Command and Control Centre” Help is based on scarce information on the disaster What, when, how, who? Decisions must be made very quickly (within 72h) ISCRAM Summer School 2009, Tilburg Coordination: UN OCHA Humanitarian Aid Flow Donors ECHO, etc. Charity UN WFP,HCR… Int. NGOs IFRC, MsF Local Government Local NGOs Victims
13. Inefficiencies in humanitarian response Monitoring disasters 24/7 monitoring capacity is expensive Many heterogeneous sources of natural hazard monitoring hard to keep up to date Response can be delayed because Not alerted / monitored Affected government does not appeal Not sure if others respond Size and type of response must be needs driven (Madrid Declaration 1995) Size of disaster can be under/overestimated Information on needs can be incomplete, vague, lacking ISCRAM Summer School 2009, Tilburg Is it a disaster?? How many people?? What are the needs?? Who will respond?? What is offered?? What is needed now?? What is the damage?? time
14. Visit to NERSS, 29-30 July 2009 12 May 2008, 6:41 UTC 14 minutes after earthquake 4500 emails, 2700 SMSs and 100 faxes sent to first responders globally Department of Earthquake Disaster Emergency Management Red earthquake alert: “high likelihood of a disaster, with need for international assistance” Global Disaster Alert and Coordination System
15. Visit to NERSS, 29-30 July 2009 M 6.7 M 6.0 “Is an event of humanitarian concern?” The objective is to distinguish between large earthquake in unpopulated or resilient regions smaller earthquake in highly populated and vulnerable regions
18. Visit to NERSS, 29-30 July 2009 Cyclone category IV 7 million people with high winds 1.8 million in storm surge zone Cyclone category I No people affected
20. Visit to NERSS, 29-30 July 2009 Prior to GDACS: monitoring through bookmarking
21. Visit to NERSS, 29-30 July 2009 With GDACS: One-stop-shop Information standards; added-value systems; System of systems
22. Source of situational information Example: international humanitarian sudden-onset disaster OCHA system Where can we find information? ISCRAM Summer School 2009, Tilburg
23. Sources of situational information Early warning and alert systems Timely knowledge about the occurrence of a natural hazard Geophysical, meteorological measurement systems Automated consequence analysis Modelling the likely impact International and social media Rich source, very timely but not always true and complete ISCRAM Summer School 2009, Tilburg
24. Sources of situational information Office for Coordination of Humanitarian Affairs (UN-OCHA): Mandate to coordinate humanitarian response Sends disaster assessment and coordination (UNDAC) teams, search and rescue teams (through the INSARAG network) Sets up an On Site Operations Coordination Centre (OSOCC), humanitarian information centres (HIC) Disseminates all information through ReliefWeb Local government, with its local emergency management authority (LEMA): Main source for official information on the scale of the disaster ISCRAM Summer School 2009, Tilburg
25. Sources of situational information Alert systems Consequence Analysis Tools Media UN-OCHA Local Emergency Management Agency ISCRAM Summer School 2009, Tilburg Reliability Timeliness
26. Information needs versus sources ISCRAM Summer School 2009, Tilburg Early warning or alert Automated consequence analysis Media OCHA LEMA Situation Source contains information for need X
27. Information needs versus sources ISCRAM Summer School 2009, Tilburg Need clusters Source contains information for need X
28. Scope and mandate ISCRAM Summer School 2009, Tilburg Health Response Political IFRC Civil Protection UNDPKO Humanitarian NATO WFP Action WHO SANCO RELEX Policy Funding ECHO European Council Sudden onset Slow onset Time scale
29. Scope and mandate ISCRAM Summer School 2009, Tilburg Health Political NATO WFP Humanitarian 20 IFRC UNDPKO Staff 10 European Council RELEX Civil Protection 5 ECHO SANCO National Global Regional Time scale
30. Responsibilities ISCRAM Summer School 2009, Tilburg Monitoring Declare crisis Briefing on state of the world Briefing on crisis (strategic) Gather info for action (tactical) Action Decision
37. Gather info for action (tactical) ISCRAM Summer School 2009, Tilburg
38. Roles and capabilities in local response ISCRAM Summer School 2009, Tilburg Offensive Pre-emptive Exploitative 9 8 3 2 7 1 Before After 4 6 10 5 8 11 9 Protective Corrective Defensive From Bharosa, Janssen (2009)
39. International situation rooms Responsibilities Capabilities ISCRAM Summer School 2009, Tilburg Monitoring Detect Evaluate relevance Declare crisis Briefing on state of the world Map Get data Analyze Brief Briefing on crisis (strategic) Gather info for action (tactical) Communicate Share Collaborate Action Decision
40. Part 2 Geospatial information management tools Mash-up versus Spatial Data Infrastructure ISCRAM Summer School 2009, Tilburg
41. Mash-up Mash-up Combination of different web services Combination of data and functionality API: application programming interface Mostly map based GeoRSS, KML: geotagged data JavaScript Examples of APIs Google maps Google geocoding Bing Panoramio: photos … Create your own http://www.programmableweb.com/howto http://www.wayfaring.com ISCRAM Summer School 2009, Tilburg
49. Spatial Data Infrastructure Professional GIS system Database technology to store and process geospatial data Handles topology, projections, attributes, metadata, long transactions, conflict resolution Visualization powerful on desktop Functionality can exposed as web services Good for Data creation Data editing and maintenance Data analysis Map creation Not so good for Web-based interactive maps Handling non-GIS data formats (e.g. KML, GeoRSS) ISCRAM Summer School 2009, Tilburg
50. Spatial Data Infrastructure GIS = Geographic information system (or science) Mapping ISCRAM Summer School 2009, Tilburg
51. GIS Handling, storing geospatial data Coordinate in 2D or 3D space special database techniques Spatial Reference System projection Imagery large volumes of data Most (>80%) data has geospatial component Manipulating, querying geospatial data Nearby point, line, polygon “In” area, “intersecting” with line Raster statistics sum of population in pixels ISCRAM Summer School 2009, Tilburg
52. Visit to NERSS, 29-30 July 2009 Rubble(pink) Standing buildings(red) Built up structures (red) Automatic damage assessment using textural analysis
53. GIS systems: network enabled Web mapping Web querying Web processing Routing Nearest objects GIS Model ISCRAM Summer School 2009, Tilburg My system
66. Mash-up or SDI? Without mash-up? Dynamic information streams cannot be used Web 2.0 data not useful Missing out on lots of new developments, APIs and tools Rigid desktop visualization Without SDI? No advanced modelling Own data can not be stored or displayed Dependent on commercial base maps Can’t create or digitize data Dependent on network availability and information service providers ISCRAM Summer School 2009, Tilburg
67. Part 3 Information modelling Mash-up versus modelling ISCRAM Summer School 2009, Tilburg
68. Modelling Case studies Impact modelling: GDACS Media mining: EMM ISCRAM Summer School 2009, Tilburg
69. Modelling Combination of information to obtain more useful information Make use of available information to estimate/calculate useful information Mathematical, physical, statistical Quantitative or qualitative Examples Likelihood for need for international humanitarian intervention after a natural disaster Tsunami wave height at coast given an earthquake Breaking news Filter and cluster information ISCRAM Summer School 2009, Tilburg
70. Visit to NERSS, 29-30 July 2009 GDACS automatic and manual event analysis Alert Coordination Volcano Monitoring Networks Disaster Level II Alert Disaster Level I Alert Earthquake Observation Networks Automatic Evaluation of scale of disaster Manual Evaluation of scale of disaster Event Alerts Start of coordi-nation Flood Watch Networks Trop. Cyclone Observation Networks Geographical, Socio-economic, population data Eye witness and information from Local Government, IFRC, ECHO, NGO Models
71. Visit to NERSS, 29-30 July 2009 Disaster alert: systematic impact analysis Risk analysis Risk = Hazard x Population x Vulnerability Hazard = 0 then Risk = 0 Population = 0 then Risk = 0 Vulnerability = 0 then Risk = 0 Impact analysis is similar Impact = Event magnitude x Population in affected area x Vulnerability Volcano Monitoring Networks Earthquake Observation Networks Flood Watch Networks Trop. Cyclone Observation Networks tsunami earthquake Socio-economic Indicators (e.g. UNDP, World Bank) Landscan or GPW
72. Earthquake mechanism Plate tectonics Relative motion of plates ISCRAM Summer School 2009 Terminology Hypocentre and epicentre (on surface) Magnitude: logarithmic measure of energy Intensity: energy on surface at given distance from epicentre
73. Earthquake mechanism Energy propagates P and S waves Attenuation functions Depends on local geology ISCRAM Summer School 2009 Energy shakes buildings Earthquake engineering Vulnerability curves
75. Earthquake effects Shaking and ground rupture damage to buildings or other rigid structures. Site or local amplification (Mexico City effect): transfer of the seismic motion from hard deep soils to soft superficial soils Landslides and avalanches ISCRAM Summer School 2009 Soil liquefaction water-saturated granular material temporally loses their strength and transforms from a solid to a liquid buildings or bridges tilt or sink into the liquefied deposits Tsunamis Fires break of the electrical power or gas lines
76. Earthquake data Occurrence Near real time (<15 min) Location and magnitude, with uncertainty USGS NEIC (US) EMSC (Europe) GEOFON (Germany) JMA (Japan) … ISCRAM Summer School 2009 Propagation Shakemaps (USGS) ESRC (Russia) Missing datasets Building stock Location, number, type of buildings Localized attenuation functions
77. GDACS earthquake alert system Scraping of earthquake parameters Agreements with seismological institutes US National Earthquake Information Center (NEIC) European Mediterranean Seismological Centre (EMSC) Japanese Meteorological Agency (JMA) Indonesia, Germany, France, Italy… Calculation of impact Population in 100km Landscan dataset Cities Critical infrastructure nearby Nuclear plants Hydrodams database not complete (e.g. in China) Airports Secondary effects Landslides Tsunamis Logistics Airport capacity nearby Visit to NERSS, 29-30 July 2009
78. Earthquake alert system Reporting Multi-lingual reports: English, French, Spanish, Italian, Turkish, Chinese Full, dynamic web report Short static email report Fax report (PDF) SMS message (Voice messages) Creation fully automatic, based on templates Alerting Professional SMS provider 200 SMS/second Global coverage Professional Fax, Voice providers Over 10000 users Alert logic Green, Orange, Red Regional filter No cancellation; only new alert if alert level increases Visit to NERSS, 29-30 July 2009
79. Visit to NERSS, 29-30 July 2009 The JRC Tsunami Early Warning System After the 2004 Tsunami in Banda Aceh, Indonesia, JRC decided to include Tsunami models for a quick evaluation of the possible impact of a Tsunami as a consequence of an earthquake
80. Visit to NERSS, 29-30 July 2009 The JRC Tsunami Early Warning System Timeline: December 2004: no Tsunami model, only earthquake alerts March 2005: rough estimation of Tsunami probability, Travel time model October 2006: SWAN-JRC Model for height distribution and locations identification International Tsunami Workshop, Ispra 5-6 October March 2007: full integration of the Tsunami model in GDACS August 2007: start of the grid calculations for over 30000 tsunami scenarios March 2008: integration of matrix calculations in GDACS JRC Models: Travel Time model – March 2005 SWAN-JRC Model – October 2006 Tsunami Grid system – March 2008
81. Visit to NERSS, 29-30 July 2009 Model characteristics fast running reliable unbreakable automatically activated on request (web service) integrated in the GDACS Based on the integration of the shallow water propagation speed Similar to the x-ray technique Provides the time in each point, starting from a source Run-time: 20 seconds -60 20 8 JRC Travel Time model
82. Visit to NERSS, 29-30 July 2009 SWAN-JRC Tsunami Model Automatic calculation of wave generation and propagation to the coast Triggered by GDACS Automatic fault generation Direction, length, height Based on initial earthquake parameters (lat/lon/magnitude) Wave propagation SWAN code (C. Mader) rewritten in C language for faster processing Post processing: List of affected cities (with wave height) The code does not calculate the run-up to the coast much finer bathymetry is necessary not relevant for early warning Run-time: 20 minutes Indonesia Magnitude 8.4 12 September 2007 11:10 UTC Major locations identified: Kandan, 3.3 m Belowa, 2.7 m Pandangaget, 1.9 m
83. Visit to NERSS, 29-30 July 2009 Tsunami Grid Historical epicenter Bounding box (Ring n. 0) Ring n. 1 Ring n. 2 For early warning, 20 minutes is too long. Pre-calculation of scenarios 10143 tsunami sources grid around historical Tsunami events (from the NOAA Tsunami sources database) 10x10 grid of 0.5 degrees around each data point 13 calculations per point magnitude from 6.5 to 9.5 every 0.25 131856 calculations 2 TB space
84. Visit to NERSS, 29-30 July 2009 Tsunami Grid 10184 data points in a grid of 0.5 degrees
85. Integration in GDACS GDACS tsunami alert Logic: earthquake in sea, magnitude > 6.5 Look up tsunami scenario, and associated maximum wave height If wave height > 3: Red alert If wave height > 1.5: Orange Otherwise: Green Take maximum of earthquake and tsunami alert Reduction of false tsunami alerts by 90% Tsunami Analysis Tool Combine Real-time sea surface buoy data Tsunami scenarios Earthquake monitoring New tsunami calculations Allows to confirm tsunamis, based on buoy data Visit to NERSS, 29-30 July 2009
86. GDACS Impact calculation allows to filter important events from unimportant ones Uses well-chosen information feeds, from sensor networks Provides information feed of high (or known) quality Visit to NERSS, 29-30 July 2009
87. But don’t underestimate crowd-sourcing USGS uses information collected from local people through the web to adjust their shaking models Visit to NERSS, 29-30 July 2009
88. But don’t underestimate crowd-sourcing EMSC does the same And uses hit peaks as a confirmation of a strong earthquake Visit to NERSS, 29-30 July 2009
89. But don’t underestimate crowd-sourcing Network of hard disk sensors Network of iPhone or Andoid phones Visit to NERSS, 29-30 July 2009
91. Europe Media Monitor What is the News Brief? Summary of news stories from around the world, automatically classified according to thousands of criteria. It is updated every 10 minutes, 24 hours a day. "Top Stories" automatically detects the stories that are the most reported in each language at the moment. Search: previous news stories (over 20 million articles are indexed). How does it work? Generated automatically by software algorithms without any human intervention. alert definition consists of a list of multilingual keywords What info is extracted? Clustering Geotagging, place names Named entities: people, organizations, titles, functions Quotes Related clusters in other languages; in time Visit to NERSS, 29-30 July 2009
93. What is MedISys? MedISys (Medical Information System) is a real-time news alert system for medical and health-related topics. MedISys: processes over 20000 articles per day from over 4000 sites of approximately 1600 news sources (news and medical sites) in 45 languages, dynamically updates statistics on all news topics every 10 minutes, categorises articles in pre-defined medical topics in 25 different languages, How does it work? MedISys turns news into a signal. Over time stories come into the news then gradually disappear. Some topics appear regularly whereas other topics occur infrequently. MedISys keeps track of these topics comparing current news events with those in the past in order to quickly identify breaking news. What does MedISys offer over other news providers? The system uses dynamic statistical modelling techniques to: suppress news noise, i.e. stories that are regularly in the news, enhance stories with a poor signal, and identify individual events on a temporal and locational basis. Visit to NERSS, 29-30 July 2009
96. Mash-up versus modelling Visit to NERSS, 29-30 July 2009 Modelling Mash-up Quality Trust Customized Quantity Flexibility General International situation room