1

Global Roads Data: A Strategy for Development

Results of the Global Roads Workshop, 1-3 October 2007 Lamont Campus of Columbia University, Palisades, NY

Alex de SherbininCenter for International Earth Science Information Network (CIESIN)

The Earth Institute at Columbia University

2

Outline

1. The goal

2. The need

3. Approaches to data development

4. A new global product

5. CODATA working group

6. Questions for APAN participants

3

1. The goal

A global roads data base that has: improved geographic and temporal coverage, consistent coding of road types, and good documentation of sources

Available free-of-charge on an “attribution only” basis (i.e., public commons)

Nominal scale of 1:250,000

4

Example of a 1:250k product

Source: ViaMichelin at http://www.viamichelin.com/viamichelin/int/dyn/controller/Maps

5

The “ideal” product

1:250k scale

Major travel routes only. Not urban streets

Attributes for road type, surface, and road use

Metadata on sources, purpose, timeliness and restrictions

Verification of accuracy

Consistent classes between countries

Connectivity between borders/tiles

Update frequency at least every 5 years, and preferably on rolling basis

6

2. The need

There is currently no globally consistent, reasonably complete, roads data product available to the development, disaster response, health, conservation, and research communities

Best available is VMAP, produced by the U.S. National Geospatial Intelligence Agency (NGA) Covers only 25-30% of the global roads network Little documentation of sources or verification of spatial

accuracy More data exist – but no release is anticipated

7

VMAP0 vs. Int’l Road StatisticsVMAP 0/ road network (IRF) = % coverage, km per country

20% coverage globally

Source: Andy Nelson, JRC

8

VMAP0 vs. Int’l Road StatisticsVMAP 0/ road network (IRF) = % coverage, km per country

< 20%

20-40

40-60

60-80

>80%

20% coverage globally

Source: Andy Nelson, JRC

9

VMAP1 publicly released dataVMAP 1- complete or almost complete countries

22 / 4%Source: Andy Nelson, JRC

10

International Steering Committee for Global Mapping (ISCGM) has completed the following

PrimarySecondaryTracks/UnpavedSource: Andy Nelson, JRC

11

User needs

Pre- and post-disaster planningEconomic developmentEnvironment and land useResearch communityPrivate sector

12

Disaster response

The map at left of travel time costs owing to a major flood in 2006 in the Horn of Africa region shows the value of combining road network data with digital elevation models (DEMs), flood remote sensing or meteorological data in order to plan for flood response, or to allocate additional travel time in the event of floods coupled with some other emergency. Bad road data will affect the validity of the results.

Produced by Paul Bartel, HIU, US State Dept.

13

Development policy

Source: Glenn Hyman, CIAT

The map at left represents an accessibility map based on low resolution/poor quality roads data, and the map at right represents an accessibility map for the same region based on high resolution/high quality roads data. Allocation of development resources based on the roads data at left would not yield optimal results, since some of the apparently most inaccessible regions actually have dense road networks.

14

Biodiversity conservation

Source: Vera-Diaz et al. (forthcoming).

Road expansion and improvement increases the farm gate price of commodities such as beef, soybeans and palm oil, and is a powerful economic incentive for the expansion of plantations on the forest frontier. These products are also under increasing global demand as food products and biofuel feed stocks. Conservation planning with better knowledge of road networks can diminish the cost of trade-offs between biodiversity conservation and the expansion of livelihood opportunities in agriculture and forestry.

15

3. Approaches to developing the data

“Mix-and-match” approachBuy from private sectorMilitary sources (e.g., NIMA)“Crowd sourcing”, e.g. OpenStreetMap Create a new data product

16

Mix-and-match approach

Pros: Can be accomplished quickly at relatively little

costCons:

Lack of consistent coverage among countries Problems with matching networks at borders

CIESIN’s SEDAC plans to develop a catalog and to carry out preliminary evaluation of available data

17

Mix and match approachCombining available national-level data*Total – 105 countries, 67% by area, 59% by population

105/59%Source: Andy Nelson, JRC

* Multiple sources including VMAP1, Global Map, CGIAR, World Bank DEC-RC, FAO Geonetwork, and others found by Andy Nelson

18

Buy from private sector

Held conversations with Teleatlas Willing to engage in data development

partnerships with a limited number of users Unwilling to sell a “skeletal” map or earlier

version without restrictions on further use Data for developing countries are still relatively

sparse Economist*: “risk of digital map monopoly…” by

Navteq & Teleatlas in the navigation map arena

* Economist, “Location, Location, Location”, October 4, 2007

19

Military sources

Main obstacle is military/intelligence community is not committed to public commons approach

Portions of their data which are available (e.g. VMAP1 tiles) may be useful for validation

20

Crowd sourcing

Pros Many hands make light

work Openstreetmap a

successful model of this approach

Cons Poor quality control Fewer inputs in low

income countries

www.openstreetmap.org

21

Create a new product

Pros Develop a consistent, well documented product Methods for integrating multiple source data

have been developed by Georigin for data-poor Africa

Build on top of this for future updates

Cons The cost will likely be > US$1million Bringing the approach to scale

22

Georigin approach

Example form Nigeria: A Russian 1:200 000 topographic map at left (georeferenced, cropped, datum shifted to WGS84) can be integrated with data from Landsat 7 (geometrically enhanced with GPS ground control points) and GPS tracks at right to produce a road map.

Source: John Dann, Georigin, Ltd.

23

4. Specifications for a new global product

1. Terminology and Classification Data model building on on UN Spatial Data Infrastructure specifications. Each road segment will include information on its provider, its collection

date as well as an indication of data quality and reliability. Initially, only information on primary, secondary and tertiary roads will

be collected.2. Database structure and functionality

The database would be structured so as to allow basic network analysis and routing functions in addition to cartographic representation.*

This implies ensuring topological consistency in the data, as well as the ability to establish connectivity with external data layers such as settlements and other transportation networks.

3. The database would be structured in order to allow versioning and maintenance of a historical archive of the evolution of global road networks.

* These would include deriving macro and meso-level transport costs, optimal routes between population centers, contingency plans in case of shocks to the network and optimized road rehabilitation investment decisions.

24

Data model

25

Approach proposed at first Global Roads Workshop – October 2007 A combination of the following data sets will be used to

manually digitize roads and attributes according to the data model described above. Scanned 1:200,000 paper maps developed by the Russian

military (ranging in dates from the late 1960s to the early 1980s) and the US Joint Operations Graphic (JOG) navigation maps.

Geocover Landsat pansharpened 15m imagery baselined to the year 2000, which are orthorectified and are available free of charge.

GPS tracks wherever available to add the most recent routes. Roads will be manually digitized and attributes assigned

according to a data model. Digitization could occur anywhere assuming a suitable

tool and management structure is developed

26

5. CODATA working group

A working group has been proposed under ICSU’s Committee on Data (CODATA)

This group will oversee quality control and move the process forward

Representatives of CIESIN (myself) and UN Joint Logistics Centre (Olivier Cottray) serve as co-chairs

27

CODATA working group members

28

6. Questions for the APAN participants

Let us know how you might be able to contribute: Improved methodologies National roads data sets for inclusion in the

catalog Funding opportunities Suggestions regarding low cost but reliable/well

managed “click worker” shops in Asia

29

For more information on the Global Roads Data workshop and

the overall strategy, visit:

http://www.ciesin.columbia.edu/confluence/display/roads