Alternative Data for Disaster Management and Lessons for COVID-19 Response

All Blog Alternative Data for Disaster Management and Lessons for COVID-19 Response

Alternative Data for Disaster Management and Lessons for COVID-19 Response

September 05, 2020 | Blog | By Author 4

Effective data management and analysis should be a foundation of both policy and programmatic responses to the COVID-19 pandemic. On June 19, Saraswati in collaboration with SIAP SIAGA, a disaster risk management program funded by the Australian Department of Foreign Affairs and Trade and managed by Palladium convened an online discussion with over 60 participants, ranging from government to the private sector, academia, media and other development partners, to explore alternative data for disaster management with a focus on Indonesia’s COVID-19 response.

This discussion involved frank exchanges about not only Government of Indonesia data and data management but also the potential of alternative data sources from satellite and other open data sources, to data from e-commerce platforms, telecommunications firms and data from big tech. Representatives from Facebook Indonesia, UN Pulse Lab Jakarta and Iykra, an Indonesian company that has been providing data analysis to Indonesia’s National Disaster Management Agency (BNPB), provided a range of interesting examples. Here are some key takeaways.

Iykra established a Solver Society network of domain and data experts to use free data from Google and paid data from Flight Radar to provide analysis on mobility and early potential rates of infection. Google mobility data allowed Iykra to create visualizations that offered one source of evidence for how people in different communities were moving before and after government Large-Scale Social Movement Restrictions that began in April



Source: Ari Kuncoro, Iykra

Data purchased by Iykra from Flight Radar allowed Iykra to review all flights to Bali from Wuhan from January-February 2020. Combining these data with estimated arrivals per flight and rates of infection, Iykra produced models with estimates of infected arrivals from Wuhan to Bali in those early weeks of the pandemic.

Similar to Google mobility data, Facebook’s Data for Good initiative has provided a combination of free and limited-availability data to support analysis by partner institutions of where populations are and how they have been moving during the pandemic. Facebook-supported disaster mapping has helped to better understand challenges in connectivity and access to electricity, for example in the wake of the earthquakes in West Nusa Tenggara and Central Sulawesi in 2018.

Source: Noory Okthariza, CSIS 

Pulse Lab Jakarta (PLJ) has explored mobility patterns during and post-disasters, for example in Indonesia following the 2018 Central Sulawesi earthquake and tsunami to support insights on disaster response. They have done this using data based on agreements negotiated with telecommunications providers. PLJ has also supported the BNPB by visualizing COVID-19 data from each province in Indonesia for easier analysis and response.

Source: Pulse Lab Jakarta on Twitter

The promise of these alternative data sources is exciting, but we are all still in the early days of learning about how best to access and use these data. As a result, significant challenges exist.

Only a fraction of relevant data available is being used to inform policy and development responses, since effective use tends to require proofs of concept that, in turn, rely on access to key data. While data philanthropy has appeal, in reality data providers are more interested in shared value partnerships with those who request their data. Establishing these partnerships can take time.

While there has been hesitancy based on concerns about data validity, one participant cautioned against demanding 100 percent validity in applying alternative data sources for disaster response, saying: “let’s use available data as best we can and explore the insights rather than waiting for perfect validation.”

Data management and analysis also require data expertise. Across the board, from government to the development sector, there is limited human resource capacity to manage complex data. The West Java provincial government—one of the best examples of local government data response—hired in-house data professionals to design, develop, and manage an integrated provincial system. The government also deployed over one hundred civil servants to help with COVID-19 data gathering and reporting into the system.

Government data are critical, both as the primary source for policy analysis and response, and also as a source for data collaborations. By its own admission, the Indonesian government has faced data management challenges, especially early in the pandemic and in receiving data from remote regions with poor internet connectivity. Complicating matters, ministries, regional governments and local government departments can use different data management systems, apply different COVID-19 case categorizations, and report data in a variety of different formats, from scans to PDF documents and infographics. In the absence of an integrated national application programming interface, data need to be downloaded manually and individually for analysis. Participants hoped this experience would provide impetus for the government’s One Data initiative.

Alternative data—such as those from private institutions—has a potential to fill the gaps of information and complement government’s official data. However, alternative data come with a qualification: these data typically only represent a certain population (for example, one telecommunications firm’s subscribers). The value in informing policy and interventions is in overlaying various data sources.

Partnerships are and will be important, most significantly to combine issue and sector experts with more than one data provider. Facebook, for example, partners with leading academic institutions and the Indonesian Red Cross in a limited data access arrangement.

Privacy was an important part of the conversation—from government regulations protecting the safety and privacy of patient information to Facebook’s measures to provide only aggregated and not personal data based on user consent via opt-in. The protection of personally identifiable information is paramount, and all agree that data sharing should be accountable and transparent.

But privacy becomes more controversial when discussing actionable data. Contact tracing, for example, while important, opens the possibility of increased government surveillance of the population through compulsory application. There is concern that the pandemic should not be an excuse for the government to roll back data privacy.

There was no argument, however, about the need for extensive testing, timely and accurate data reporting, and analysis of those data to promote willing cooperation of a better-informed public. Discussion participants were particularly interested in exploring data sharing, data research collaborations and knowledge sharing on open data developments and policy going forward. A key lesson from using alternative data in support of disaster management is that such partnerships and collaborations will prove very useful.