Forest fires affect many parts of Southeast Asia, resulting in extensive environmental destruction, health problems, school closures and transport cancellations. Haze Gazer is a web‐based decision support system for disaster management authorities which harnesses multiple sources of data to provide insights on haze disaster dynamics.
Innovation Summary
Innovation Overview
Haze Gazer provides real-time insights on the: locations of fire and haze hot-spots, the strength of haze in population centres, the locations of vulnerable cohorts of the population, and the response strategies of affected populations, including movement patterns and institutional behavioural changes.
Haze Gazer is unique in that it uses multiple sources of data, generated by citizens online, concerning the situation on the ground, and harmonises the insights from these datasets with other sources of information, for example from satellites, to offer information on disaster event dynamics. Specifically, it uses advanced data analytics and data science to mine open data, such as fire hotspot information from satellites and baseline information on population density and distribution, as well as citizengenerated data, including the national complaint system in Indonesia called LAPOR!, citizen journalism video uploads to an online news channel, local radio feeds, and real-time big data such as text, image and video-oriented social media.
Replicability: First, because haze affects many countries in SEA, the platform has strong potential to scale as a regional tool to inform hazard-elated humanitarian efforts and to improve resilience. Second, it can scale in terms of insights, based on the integration of richer data sources.Third, if disaster management authorities agree to publish their operational practices, the platform will capture insights on both operational potential and real-world haze crisis dynamics. Finally the underlying mechanism of Haze Gazer can be applied to other types of disasters or sustainable development themes.
Innovation Description
What Makes Your Project Innovative?
According to CIFOR, fires begin and spread for many reasons, so it is misleading to think of ‘fires’ as the problem, or even as a single problem. Complex socioeconomic, ecological and governance factors are involved, meaning that the drivers, and the solutions, go beyond who actually lights the match. At present Indonesian disaster management authorities manage peatland fire and haze events based on hot-spot data from satellites as well as static data on population density and distribution. But to support affected populations better, the Government of Indonesia is starting to use more timely data and more information on the dynamics of the disaster, especially the situation on the ground.
Innovation Development
Collaborations & Partnerships
Haze Gazer was a collaborative, multi-partner effort. The Office of the President (KSP) provided access to the LAPOR! dataset and led integration of the system within public administration. The National Disaster Management Office (BNPB) was intimately involved in the design, by contributing insights on existing government systems, and testing of the tool. Crimson Hexagon was used during the initial proofofconcept research and DataSift provided access to the social media datasets.
Users, Stakeholders & Beneficiaries
Government bodies, Government staff, High-risk populations
Innovation Reflections
Results, Outcomes & Impacts
Haze Gazer represents an evolution of the National Disaster Management Office’s existing information system to include near realtime insights on the evolution of haze events and their social impacts on the ground. It enables Indonesia’s disaster management authorities to target better their interventions and to align their efforts with those of affected populations. This more targeted and agile approach by national and local disaster management authorities is expected to increase community resilience in the face of haze events.
Challenges and Failures
Text analysis and issue classification posed a challenge for the development team when dealing with unstructured datasets from social media (text, image and video). The Partners relied on UNORCID’s technical advice in order to understand dimensions of interest and to develop the taxonomies for analysing the unstructured social media data. Another challenge came in the form of identifying other sources of data that could contribute insights on local haze event dynamics, and in understanding how these many varying sources of data could complement one another. Converting local radio feeds into useful information was also a challenge, which was overcome by using Google’s speech to text tool to develop a database which could be searched for key terms by applying the taxonomy.
Conditions for Success
Access to an array of datasets was central to the depth of insights available on Haze Gazer and to the success of the initiative. Political leadership was necessary for the integration and uptake of the tool within the public administration. Web development and data science skills were necessary throughout implementation.
Lessons Learned
Most efforts to use online media, especially social media, to inform public policy have focussed on text mining. This project looked at multimedia sources, including image, videoand audiobased inputs. The partners have found that these other information sources are a useful complement to textbased inputs. In addition, the importance of understanding existing government systems, and ensuring that the new tools and systems provide the same functions as the legacy systems, as well as offer additional functions, should be emphasised. Hazer Gazer, continues to provide the functions of the legacy system, while at the same time offering new insights based on public discourse on multimedia platforms.
Status:
- Implementation - making the innovation happen
Innovation provided by:
Date Published:
7 August 2016