Nessun titolo diapositiva

1
Spatio-temporal gap analysis for the
UNHCR-InterSOS WebGIS using open source software
Giorgio Guzzetta guzzetta_at_fbk.eu Cesare
Furlanello furlan_at_fbk.eu
http//mpa.fbk.eu
2
Overview

• Opportunity
• Large availability of data for analysis and
  modeling
• from InterSOS profiling through the
  geodatabase-backed WebGIS
• Goal
• build instruments for spacetime assessment
  indicators on
• risk
• needs
• fact-based decision making
• Tools
• Rich geodatabase technologies
• Geostatistical modeling
• GFOSS interoperability

3
Motivation

• ReliefWeb - UN Office for the Coordination of
  Humanitarian Affairs
• Global Symposium on Information for Humanitarian
  Action
• Geneva, Switzerland, 22 26 October 2007
• Standards and recommendations
• Trigger/indicators for slow-onset crisis
• High standards of analysis
• Strategic humanitarian information for decision
  making
• Agreed standardsindicators
• (Final document of the WG on Humanitarian
  Financing Supported by Information and Analysis)

4
Summary

• The challenge
• Sophisticated models for automated event
  detection and gap analysis
• as a process (WebGIS integration)
• Our model epidemiological surveillance and
  preparedness
• The framework
• Open source software combining in an
  interoperable way
• Geodatabase management (PostGis)
• Geostatistical analysis (R, GRASS)
• WebGIS technologies (Mapserver)
• Applications to humanitarian crises (West
  Darfur)
• Population patterns
• Movement tracking
• Gap analysis (example drinkable water according
  to UNICEF standards)

5
The challenge
Extending the WebGIS capabilities
Gap analysis
6
The challenge
Extending the WebGIS capabilities
Gap analysis
7
The framework
8
The framework
9
Application 1 population patterns

• Maps developed
• overall population density before and after the
  crisis
• population densities by ethnic group (arabs vs.
  africans)
• population percentage prevalences by ethnic
  group (not shown)

• Retrieval of data
• population distribution before after the
  crisis, by ethnic group.
• PostGreSQL query RdbiPgSQL
• 558 records (villages) population covered gt500k.

• Cleansing of errors in data entries (NAs, wrong
  geocoding)
• data loss lt1
• Gaussian smoothing of point data
• conversion of data to object of class ppp
  spatstat
• density.ppp function spgrass6
• Normalization of gaussian smoothing result
• peak of gaussian so that its integral equals
  population village
• Isolines on population density maps
• function contour.im spatstat
• function contourLines grDevices

• To be implemented

10
Application 1 population pattterns
Overall population before the crisis
11
Application 1 population patterns
Overall population after the crisis
12
Application 1 population patterns
Arab population before the crisis
13
Application 1 population patterns
Arab population after the crisis
14
Application 1 population patterns
African population before the crisis
15
Application 1 population patterns
African population after the crisis
16
Application 2 movement tracking

• Maps developed
• quarterly movements of people fled
• quarterly movements of refugee returnees (not
  shown)
• quarterly movements of IDPs returnees (not shown)

• Retrieval of data
• standardization of movement information from the
  tables (PHP)
• calculation of nearest border points for movt. to
  Chad border (SQL)
• PostGreSQL query RdbiPgSQL
• gt2500 movt. in 2002-2007 total population
  involved gt1 M.

• Collapsing similar movt. to enhance readability
  of maps
• movt. with same destination and date similar
  length and direction
• intensity of collapsed movt. (hhs involved) sum
  of simple movt.
• origin of collapsed movt. average of origins
  weighted by intensity
• Visualization of zones with intense destruction
  or abandoning
• before-crisis population density isolines on
  abandoned or destroyed villages
• Visualization of movements
• larrows from origin to destination of collapsed
  movements
• color related to movt. intensity
• double color scale ordinary movt. vs. very
  intense extraordinary movt.

• To be implemented

17
Application 2 movement tracking
18
Application 3 gap analysis
Map developed access to drinking water from
improved sources
19
Application 3 gap analysis
Next map
20
Application 3 gap analysis
Next graph
21
Application 3 gap analysis UNICEF standard
indicators on drinkable water
22
Application 3 gap analysis
Previous map
UNICEF standards on drinkable water
23
Application 3 gap analysis
UNICEF standards on drinkable water
24
Application 3 gap analysis
Previous graph
UNICEF standards on drinkable water
25
Summary of results

• Opportunities from automated procedures
• Population patterns
• continuous monitoring of data to detect changes
  in population distribution and triggers for onset
  of crisis events (OCHA recommendations)
• information on squatted zones for the planning
  of returns of refugees and IDPs to their villages
  of origin
• identify densely populated areas to be related to
  gravity of gaps
• Movement tracking
• modelling patterns of migration during crises
• identifying temporal patterns in the squatting of
  the villages
• extraordinarily intense movements may be useful
  learning data for automated event detection
• Gap analysis
• continuously updated spatial information on gaps
  and emerging needs
• improved coordination of humanitarian crisis
  management through ticketing
• gaps may be linked to crisis event prediction

26
Work in progress
Integration in PL/R and WebGIS Ticketing / GeoRSS

• Console interface for the management of events
  and gaps
• Automated information on tasks through GeoRSS
  feeds

Next future
Bayesian techniques Statistical machine
learning Event detection for predictive
models Demographic models from epidemiology
Thanks to Sergio Odorizzi (InterSOS), Roger
Bivand (Norges Handelshøyskole)