News — 24 September, 2012
From Remote Tracing to Field Mapping in Padang
Padang, West Sumatra, has been identified as one of Indonesia's most vulnerable cities; in 2009 an earthquake claimed over 1,100 lives and destroyed or damaged more than 300,000 buildings. Additionally, more powerful earthquakes are predicted for the coming decades with over 300,000 people currently living in the tsunami inundation danger zone. In total over 800,000 people are at risk from earthquake and tsunami activity in Padang. This post details the successful combination of volunteer led remote mapping and training on the ground.
During the week of September 17th - 21st, the Humanitarian OpenStreetMap Team conducted training for Badan Nasional Penanggulangan Bencana (BNPB, the national disaster prevention agency), Badan Penanggulangan Bencana Daerah (BPBD, the provincial disaster management agency) and university students in Padang. Thanks to the efforts of remote mappers we had a fantastic base dataset to demonstrate the potentials of OpenStreetMap and to build upon with additional data sourced on the ground.
The Humanitarian OpenStreetMap Team has had a successful history working in Padang, and this post is largely intended to update and thank all of those who have contributed to disaster preparedness in the city. Here we coordinated the first imagery tracing job with the new (at the time) Tasking Manager and purchased DigitalGlobe satellite imagery. The task was a fantastic success: Padang city was split into 1,017 small areas and 81 contributors from around the world worked to digitally trace all of the visible buildings and roads. Tracing was undertaken by OpenStreetMap volunteers, as well as a collaboration with GIS Corps and an amazing contribution from Red Bank High School. Most of the area was digitized by the end of 2011 and the results are visible on the map.
Recording building details, such as structure, composition and function, is incredibly important for our work in Indonesia. These details allow the Indonesian government and others to predictively model (using InaSAFE) the impact of natural disasters by forecasting the impact on existing buildings and infrastructure; from this we can estimate the number of people that will be impacted and to what level. Most importantly, these results can be used to act in order to reduce the impact of future disasters. The required building details are difficult to determine from aerial imagery; however, as the Padang example illustrates, the potentials of combining remote tracing with on the ground recording is hugely effective.
In Padang, remote users provided a fantastic dataset of building locations and footprints. We were able to build on this information once we arrived on the ground and fill in building and structure details. This building, a house in an area of earthquake and tsunami risk, demonstrates the collaberative nature of this project well.
With the first round of training complete, and background tracing in place, Padang is in a fantastic position to have more information recorded and added to the map. This geospatial data will improve existing models and help in both future disaster recovery and planning for risk reduction.
During our week in the field we initially collected details of buildings surrounding our training venue, and then moved further afield to conduct targeted recording of public buildings and high risk areas. The work required to map Padang completely at such a level of detail is large, but achievable. The Padang example demonstrates that volunteers working remotely from aerial imagery play a hugely important role in the recording process and provide a great assistance to mappers on the ground. Tracing Padang proved the Tasking Server to be a powerful and important tool for managing collaborative efforts. The system has been used during multiple projects since and continues to support the efforts of volunteers working from home. A thank you to everyone who participated in our original tracing task, you have helped mappers on the ground create an incredibly important dataset.