Geo Data Integration on UNHCR GeoPortal

Geo Data Collection on MobileDevices

Goma, Eastern Democratic Republic of Congo (DRC)

9-18 March 2010

A data management mission took placefrom 9th to 18thMarch to Goma, DRC. The objectives were: first to explore new solutions to improve field data collection mechanism and second to help integrate currently available information tothe UNHCR GeoPortal.

1. Team composition

FICSS/DPSM was represented by Edouard Legoupil, Data Management Officer, who joined a delegation of Google made of:

  • Divon Lan, Product Manager, Sub Saharan Africa, Tel Aviv.
  • Evans Arabu, Geo/ODK specialist from Nairobi.
  • Sean Askay, Google Earth Outreach, MTV
  • France Lamy, Program Manager, Partnership.

The mission coordinated the work with the UNHCR Operation Data Manager for Eastern Congo, (Colince Guedje Ngatsi);,the UNHCR Regional Officer, bBased in Kinshasa(Alicia Ortega-Gonzales) and the UNOPS Associate GIS Officer (Jean-Laurent Martin) from the IDP Data Center project. The mission reported to the UNHCRCoordinator for Eastern Congo(Karl Steinacker) and the Head of GomaSub-Office(Matsi Notz).

2. Advantage ofmobile data collectionand comparison with existing systems

Mobile data collection systemsare more and more used by various organizations in the field.

Strengths

-Powerful questionnaire language (skip logic, value constraint…);

-Integrated photo/navigation system (GPS)/barcode/fingerprint collection (depending on the device);

-Easier submission of data – data collection and data entry at the same time;

-Enforces data standards through XML format.

Weaknesses

-Cost, durability, training needs;

-Creating electronic forms is more complicated than paper forms;

-Battery usage should be monitored (possibility to use solar charger or car battery);

-Updating existing data in an offline environment is complex;

-Susceptible to physical damage of smart phones, computers crash too common.

During the mission we had to opportunity to assess the mobile data collection system used currently by the WFP “Vulnerability Assessment Mapping”Section (SuperWaba). The table below describes the differences between the two systems. The Android/ODK system is judged more flexible than the Windows/SuperWabaas it offers more ways to collect, format and transmit data.

Application / Open Data Kit:
/ SuperWaba:
Operating System / Google Android / Windows mobile
Data Standard / Xforms standard + Java Rosa / XML outputted is not attached to a schema, ca not be used as an interchange format
Device / Replaces several devices: a PDA+ GPS + Video / Few PDA fully integrate: PDA+ GPS + Video
Connectivity / Wifi, GSM, Sd Card / Sd Card only
Robustness of the application / Some minor bugs have been observed – cf annex 1 / No information. WFP has already the system deployed in 6 countries.
Forms deployment / Easy deployment of new forms through Wifi / Needs install on Sd card

3. Activities of the mission

  1. Integration of the three currentdatabases (“éevaluation rapide”, IDP registration and GIS)and data consistency analysis
  1. Training for UNOPS and UNHCR Data management unit technical staff on advanceddatamanagement tools
  1. Installation of local replication of the UNHCR GeopPortal in Goma Sub Office
  1. Installation of an analysis system on the IDP registration system currently managed by UNOPS
  1. Transcription of different forms used in the field in the standard format used by Open Data Kit
  1. Preparation of a short video presenting the test organized by the mission
  1. Preparation of action items to start piloting the system – cf. annex 3

4. Findings

Separation of data collection and project implementation

Data collection is better managed as a specific project in an operation than as a subcomponent of a programme. It allows then to separate the information collected neutrally in the field, the information provided by partners implementing a project and the information from UNHCR officers evaluating a project.TheUNOPSIDP data center prevents conflict of interest in data management and is a good example of this good practice.

Positive reaction from field staff to mobile data collection system

As shown in the video produced by the PI Unit of UNHCR Goma Office, the staff who tried the system wasenthusiastic. Field staffs are all from a generation that is keen on technology and feel gratified to use such technology. It was impressing to see how quickly they did understandhow to use the phones and the forms.

Google currentsoft commitment

Google foundation has totally changed its scope in 2009. It skipped from grant funding to technology support[1]. Google strategy is to serve the public with the biggest amount of information. The major goal of Google for this mission is not to support UNHCR operation to establish contracts for the supplyof licensed content (cf in annex 2, the example of contract sent by the Google mission leader). On our three main requests, Google’s response, at this stage, remainsed limited:

  • ODK support: ODK is originally a project from the University of Washington. The Google delegationinsisted on the fact that they will not be able to support formally an application that is not part of the Google.org pProducts (
  • Contribution of phones: Google gave (OR LENT??) to UNHCR three G1 Android phones for a total value of around1000 USD (350 USD per phone);
  • High resolution imagery: Although the release of high resolution imagery has been clearly expressed during all the preparatory meeting(Google requested to delimit specific areas for the procurement of HR imagery from UNOPS in North Kivu), the Google mission did not made any announcement for such release.

Need of a common strategy for geodata productionand public data release

Many big NGO’s, who are not necessarily UNHCR IP, do have dedicatedGIS staff based in the Kivus and do collect information on their own.UNHCR, either directly or throughtheIDPDataCenter, is collecting different types of information, some of it being public domain data (Infrastructure and sectoral assessments). It would be profitable for all actors to establish a true inventory of the data available within the humanitarian community.

Although UNHCR has already its own web platform for geographic data dissemination, the UNHCR GeoPortal, the collected public domain data need then to be published in a publicly available platform. Google Map can be considered as one of the options to do so,Open Street map is another one (they have furthermore a particular focus on humanitarian use[2]).This discussion must be initiated within the humanitarian community while doing this inventory.

Expectation from local data management staff for moretechnical training

Despite the fact that providing training was not part of the initial ToR of the mission, it appears that there was a strong need for more training on data management for (local) staff in French language.

5. Recommendations

  1. The Google Teaminformed UNHCRthat some “googlers” were ready to volunteer on the ODK project. (Any Google employee is allowed to spend 20% of his time on personnal project).UNHCR should insist on a more consistent commitment (i.e. ODK becoming a Google product). Publishing the video on the UNHCR YouTube channel is a way to express this position.
  1. Follow up with Google aboutthe supply of high resolution imagery for the Kivus on Google Map
  1. Contact mobile phone manufacturers to obtain alternative devicesto furtherthe testingphase. For instance, Motorola has just launched one of the most advanced models of Android (called Milestone in Europe or Droid in the US).
  1. Clarify the vision of the local actors concerning the creation and the dissemination of public data in Eastern DRC. This vision would be better established through a diagnostic from an external actor(i.e not directly a UN agency).
  1. Organize a follow-up training session in a upcoming pilot which should include the identification and testing of alternative systems that could be available, favoring open source solutions.

Annex1: List of improvement to ODK

Open Data Kit is an Open Source project. As such, there is a public list of feature request (called “Enhancement”) or bug report (called “Defect”) in the Google code forge website:

During the mission, the following points were noticed:

  1. Debugging. Some forms crash.

It is possibly related to this ticket:

  1. Issues have been raised concerning the default langage of the form.

Cf also this ticket:

In general, ODK is not yet fully localized (i.e. the translation of the product into different languages).The word “form” for example has not been translated.

  1. Cascading selection is an intersting features for Admin level question.

Cf ticket:

  1. ODK has not been created for data update. Some improvement should be made for data reviewing:
  2. Send the data back to the device and incomplete data;
  3. Change the way XML files are stored in the device in order to make them searchable;
  4. Build a search engine to find the xml files to be reviewed.

This may be linked to this ticket:

  1. Improvement in the user interface for long questionnaires: Allow more than one question per page.

No ticket in the forge yet.

  1. Integration of ODK with a navigation application. The user should be able to visualize mapping tiles offline

No ticket in the forge yet.

Using the 5 phones, officers in the field may define new tickets using the following template:

  1. What steps will reproduce the problem?
  2. What is the expected output?
  3. What do you see instead?

Annex 2: GOOGLE SAMPLE LICENSED CONTENT

  1. Licensed Content:
  1. Licensor will provide to Google the following content, some of which is customized, according to the following technical requirements and specifications (the “First Deliverables” and “Second Deliverables” collectively referred to as “Deliverables”):
  1. First Deliverables:
  2. Geocoded datasets on health/education/water/etc service and product providers for integration into Google Maps. Licensor annually collects information on X thousands of features (e.g, schools, water points, health facilities) as part of its routine needs assessment. Licensor will make the location and basic attributes such as the address and types of services of schools, water points, health facilities publicly available and accessible through Google’s web-based mapping tools, particularly for humanitarian organizations but also for the general public. Licensor already has a number of datasets that are of interest to Google, including several thousands of schools, water points, health facilities (others?). The inventory of schools, water points, health facilities for which Licensor has the minimum data requirements for integration into Google Maps is listed under Appendix 1. In total, up to about XX,000 schools, water points, health facilities in the Project Countries will be provided under this phase.
  1. Second Deliverables:
  2. Geocoded datasets on schools, water points, health facilities, as described above, for approximately X,000 schools, water points, health facilities from DRC, and/or other countries.
  1. Technical Requirements and Specifications:
  2. All points of interest will be uploaded into the Google Map database and will be publicly available. Data transfer to Google will be done in “bulk” (e.g. Excel file of providers), or this can be done directly by UNHCR. The ideal scenario would be where existing UNHCR datasets from the initial countries and from other countries are provided in bulk to Google, whereas newly collected data are uploaded directly by UNHCR (this can be done either centrally or by platforms), thus requiring minimal input from Google. Google will provide guidance, at least initially, on the use of its mapping tools.
  3. The contents of the datasets that will be provided to Google include, at the minimum, the name and type of health/education/water service/product provider, address, and geographic coordinates. Specific attributes such as type of service provided may also be added. The detailed data scheme is provided under Appendix 2.
  1. The Licensed Content will be delivered to Google according to the following schedule:
  1. First Deliverables: Licensor will deliver to Google the First Deliverables listed above by no later than 30 calendar days from the Effective Date; and
  1. Second Deliverables: Licensor will deliver to Google the Second Deliverables listed above according to the following schedule:

Country / Number of Points of Interest (POIs) / Time period (BY THE FOLLOWING TIMELINES, OR AS MUTUALLY AGREED UPON BY THE PARTIES)
Congo
Uganda
X country / XX
XX
XX / (END OF )Q2 / 2010
Q2 / 2010
Q3 / 2010
Q3 / 2010
  1. License Fees. In consideration of the Licensed Content licensed to Google hereunder, and subject to delivery and acceptance of the Licensed Content in accordance with Section 3.4 of this Agreement, Google will pay to Licensor a total of US$ X 000 in fees (“License Fees”), upon Google’s acceptance of the Deliverables.

The License Fees are for perpetual and irrevocable use of the Licensed Content by Google, as set forth in this agreement.

Annex 3: Open Data Kit Pilot project action Items.

Action items 1: Publication of the mission video

Lead:UNHCR Division for External Relations

Activity: Publication of the mission video on the UNHCR you tube channel- need first clearance from Google and Premiere Urgence

Action items 2: Improvement of Open Data Kit

Lead: Google

Activity: Software development

Action items 3: Supply of high Resolution imagery for Kivu

Lead: Google.org

Activity:Publish high resolution imagery on Google Map (eventually premier accoubt only).

Action items 4: Finalisation of the forms and of the database connection

Lead: UNHCR Data Management Officer FICSS/BSC

Activity: XML configuration and jobs scripting

Action items 5: Enhancement request / bug report

Lead: UNHCR Data Management Officer Goma + UNOPS GIS Officer

Activity: Create specific ticket – cf annex1.

Action items 6: Test of different Android devices.

Lead: UNHCR Emergency ICT Unit

Activity:Choice of the most appropriated Aandroid device

Different devices are available. Points to be considered are the following:

-Rugability vs cost of acquisition and operation

-Physical keyboard / virtual

-Adapted screen for high luminosity

Action items 7: Procurement of the devices and/or request for sample

Lead: UNHCR FICSS / DPSM

Activity: Procurement and shipment of the devices

Action items 8: Preparation of a deployment methodology.

Lead: UNHCR Data Management Officer Goma + UNOPS GIS Officer

Activity: Create a sharable document so that the system can be used in other operations.

Action items 9: Preparation of ToR for an IP to execute the pilot project

Lead: UNHCR Snr GIS Officer in coordination with Data Centre

Annex 4: Terms of Reference - GeoData Inventory and public Domain Data Publication in Eastern DRC

Context

Numerous humanitarian actors do collect some information in Eastern DRC.

Currently there is no inventory of the types and the extents of the data collected by those organisations

Objectives

The objectives (See action item 9 if Annex 3)of the study would be

  • Meet with all partners that have GIS or DB capacities
  • Prepare an inventory of the data available
  • Initiate discussion about the exchange and publication of public domain data

Expected output

  • Populate a metadata data base, such as GeoNetwork Kivu and train partners on how to maintain this GeoNetwork node.
  • Prepare a policy for public domain data publication
  • Elaborate a procedure for data publication (on Google MapMaker / Open Street Map)

1

[1]“ When we reviewed our progress in early 2009, it became clear that while our partners were doing excellent work with our grant funding, we could do more to effectively use Google's engineering talent by focusing on the technical contributions we could make. We shifted our focus to engage in engineering projects at that time.”

[2]