Assessing and Improving Open Geospatial Data#

As established in phase 1 of this project, open source data respoitroies, primarily OpenStreetMap, are foundational to the daily work of geospatial practitioners in several international development organizations. Creating data in existing open repositories allows teams to build tools and workflows that leverage these data sources and continue to refine and implement them over time.

Based on this idea, we have identified three project locations where we want to update the foundational OpenStreetMap data, notably roads and buildings. These data have proven to be useful in many projects, and are also the most frequently unavailable, or out of date. To perform the data updates, we have hired two companies to update the data in OpenStreetMap directly, and we coordinated a map-a-thon internally at the World Bank for the other location.

See the embedded section here for details on the project locations.

Data Improvement#

Each project location was selected based on having available locations for improvement of foundational geospatial data, and based on project requirements. Each project presented unique requirements for data collection that determined how much data was improved, and how much area was covered. For each location, we collected a snapshot of OpenStreetMap as a baseline, and then compared to the final results

Location

Buildings baseline

Buildings added

Kasai Health Zones

8533

200000

Chad Refugee Camps

49043

168322

CAR mapathon

2200

2003

Conclusions#

OpenStreetMap is, and will continue to be, the best source for certain foundational datasets such as roads and buildings. When the data are incomplete in an area of interest, leveraging OSM as a home for updated data is a great option. However, not all areas are best served by hvaing areas digitized in OpenStreetMap, and the density and size of an area can quickly eat into budget. The table below describes considerations and when digitizing in OSM is a useful solution

Consideration

Useful

Problematic

Size of area

small, prescribed areas, such as single cities and refugee camps

Large administrative areas

Data required

Specific infrastructure such as buildings and roads

non-OSM data, such as population or landuse data

Finally, while we had great success with the two commercial companies we contracted, we were less successful with the Mapathon. Mapathons are great tools for educating people about OSM and how it can be used, but are inconsistent in output. Mapathons should be limited to educational opportunities and marketing campaigns.