Leveraging OpenStreetMap to improve disaster risk management in the Seychelles

By Michael Wagner, Grace Doherty, and Boris van Zanten

Open Cities Africa Seychelles is targeting the coastal areas of the archipelago’s three main inner islands (Mahe, Praslin and La Digue) to gather information on the risk of urban and coastal flooding. While these risks are real and relevant to anyone situated in the coastal areas, we are looking at the impact on tourism establishments in particular (with tourism being the main sector of economy in the Seychelles).

Our first stakeholder meeting was held in early 2018 at the Seychelles Geospatial Working Group, attended by the Department of Risk and Disaster Management, the Tourism Department, and several first responders, among others. Collaborators worked to define the details of the data model, prioritizing severe impact sites such as tourism establishments and buildings with more vulnerable “occupants,” such as schools, daycares, nursing homes, and clinics.

“We estimate that OSM now covers around 95% of the building outlines on the three main islands to date.”

In August, these organisations returned for a four-day hands-on workshop on GIS and OpenStreetMap (OSM), joined by secondary school teachers. The training covered the basics of OSM’s iD editor and JOSM and highlighted key GIS processes, such as using OSM data in QGIS. While one reason for providing the training was to support the collection of risk relevant data, these sessions have equipped attendees with data collection and analysis skills that are valuable far beyond the disaster risk management context.

We held an intensive one-week mapathon a week later at three parallel events on each of the targeted islands. Around 70 teachers and students from six secondary schools volunteered their time to the events. Mappers used the paper-based Field Papers approach for field data collection, alternating days between the field and data processing indoors. 

After performing some quality control and validation, our core team of four revisited some of the sites on Mahe. Overall, we captured about 60% of the tourism establishments in the coastal areas of the three main islands.

We would like to mention that the Centre for GIS and the National Bureau of Statistics made a great open data contribution to the OpenStreetMap Foundation. As a result, we were able to import about 32,000 buildings into OSM by August. We estimate that OSM now covers around 95% of the building outlines on the three main islands to date.

Parallel efforts: Drone workshop and mapping

April 2018 saw the launch of a drone workshop organized by World Bank, our team and drone trainers from Zanzibar. The results were orthorectified images (i.e. distortions removed, correct orientation and position, constant scale) with 7cm ground resolution; one image pixel represents 7cm in nature. These images were later used during the community mapping exercise to identify and trace risk relevant data.

Sustainability in product development

While we were planning for some (mobile) app originally to visualize and understand risk and impact under different scenarios using the data we collected we eventually decided against it. There is already great open source software available providing precisely such functionality, such as InaSAFE. Instead, stakeholders will benefit from an automated workflow to extract and convert risk relevant OSM data into a proper GIS data format on a regular basis that can be used easily for all kind of spatial analysis. Many of these stakeholders also use the main geodatabase of the Centre for GIS, either via direct access or via standard web services such as WFS and WMS.

The tool in development will import OSM data into that main geodatabase on a daily basis and thus, provide the stakeholders with up-to-date and easy-to-analyze risk data. To ensure sustainability, stakeholder organizations are advised to update existing data which may serve needs beyond their own organization (such as buildings, roads, rivers, etc.) on OSM instead of performing updates in siloed, closed environments.