Research Dive Review: Data Innovation for Disaster Management

Team photo.

Earlier this month Pulse Lab Jakarta hosted its second research dive, this time in partnership with UN OCHA, bringing together data innovators from across Indonesia to further disaster management through the sophisticated use of image-based datasets. Below we share our impressions from the event and the next steps for the teams.

The Spirit of a Hackathon

We were honoured to host 16 researchers from across Indonesia, along with two mentors, for the research dive. The participants self-organised into four research teams, focussing on:

  • Image classification related to haze events;
  • Inference of visibility levels from images of haze in Sumatra island;
  • Quantification of the impact of volcanic eruption from satellite imagery;
  • Modelling risks and assessing hazards using the landslides in Garut as a case study.

Over the four days the teams were competing against the clock as they prototyped analytical tools and generated research insights under the four topics.

Framing the Challenges

Agus Wibowo, the Head of Data and and Information at the National Disaster Mitigation Agency (BNPB), gave the teams a sense of direction with a presentation on current disaster management practice and the need to understand the risks and the near-real-time impact of haze, volcanic eruptions, floods and landslides. He highlighted the potential of real-time monitoring to evaluate and target programmes during implementation and improve decision-making processes.

Agus Wibowo also shared a specific case study with the teams, where the Data and Information team had used drones to collect data on the condition of the land, but had not effectively surveyed the area due to weather conditions. His team was, thus, looking for models of settlement damage and classification of damage zones from images, in order to better manage a disaster.

Diving into Image-Based Datasets

The researchers had access to 5400 images related to haze collected from social media, gigabytes of time-series satellite imagery capturing an active volcano pre- and post-eruption from the National Institute of Aeronautics and Space Indonesia (LAPAN) and Google Earth, as well as UAV images of the recent landslides in Garut.

Despite the time constraints, all the teams managed to develop new methods for mining and analysing the image-based datasets. One group successfully developed a method for automatically generating a narrative from photos taken during a haze event, while another group developed a prototype model to infer the visibility levels from the same dataset. These methods are important for retrieving relevant information in real-time and for measuring the impacts of a disaster.

The other teams found that combining images from social media and analysis from GIS offers potential for better analysis of damage from a disaster. In addition, they found that a vulnerability model could be developed to predict disaster impacts. All of the findings could be useful for policy-makers for spatial planning and risk reduction, as well as for civil servants during emergency response, recovery and rehabilitation.

Next Steps

The teams have been encouraged to submit a technical paper regarding their research findings. The papers will be collected into a Technical Report on Image Mining for Disaster Management and shared with networks of policy-makers and academics in Indonesia and beyond.

Pulse Lab Jakarta wishes to continue the collaboration with academia in Indonesia and to share the knowledge, methods and tools generated by these partnerships with decision-makers.

Pulse Lab Jakarta’s next research dive, in March 2017, will bring together academics and other stakeholders to explore datasets related to the Millennium Development Goals.

Pulse Lab Jakarta is grateful for the generous support of the Department of Foreign Affairs and Trade of the Government of Australia.



UN Global Pulse Asia Pacific
United Nations Global Pulse Asia Pacific

UN Global Pulse Asia Pacific is a regional hub that aims to drive data innovation and sustainable development to ensure that no one is left behind.