Identifying the vulnerable population using de-identified telecom data and satellite images for enhancing disaster management

Principal Investigator (Affiliation)
  • Prof. SHIBASAKI Ryosuke
    Center for Spatial Information Science, The University of Tokyo
Co-investigator
Project Associate Prof. MANANDHAR Dinesh
Center for Spatial Information Science, The University of Tokyo
Project Assistant Prof. MIYAZAKI Hiroyuki
Center for Spatial Information Science, The University of Tokyo
Project Assistant Prof. FAN Zipei
Center for Spatial Information Science, The University of Tokyo
Project Assistant Prof. WITAYANGKURN Apichon
Center for Spatial Information Science, The University of Tokyo
Project Researcher ARAI Ayumi
Earth Observation Date Integration & Fusion Research Initiative, The University of Tokyo
Research Participant
WU Guangming (August 2020 - March 2021)
CHENG Qianwei (August 2020 - March 2021)
GUO Zhiling (October 2020 - March 2021)
International Coordinator
tbc
Research Institutions in Japan The University of Tokyo
Cooperating Organization in Japan
Partner Country Republic of the Union of Mozambique
Research Institutions in Mozambique
Cooperating organization in Mozambique Eduardo Mondlane University / Instituto Nacional das Comunicações de Moçambique / Instituto Nacional de Estatistica
Support Organization United Nations / World Bank Group
General Description of the Research Project Gaps to be filled by this project
Timely and granular population statistics is crucial for effective disaster management. Particularly in a resource poor setting, the source of population data is limited, and data could be easily outdated if they have to rely on official statistics alone such as population censuses. Telecom data, which are passively collected by mobile network operators for billing and network management purposes, are updated with each network event—a call, an SMS, and data communication. These generate highly localized and frequent digital footprints as timestamps associated with the cell tower location of the event in near-real time. Given the wide coverage and ubiquitous usage, especially in resource-poor environments, telecom data are a critical data source to inform monitoring and policy design for disaster management.

Our approach
We develop a system to identify the time and seasonally varying distribution of the vulnerable population by using the de-identified telecom data and satellite images for enhancing disaster management. Proposed system can estimate the spatial distribution, mobility patters, and headcount of the vulnerable population. Output of the system enables responders, policy makers, and responders to produce the disaggregated statistics of the vulnerable population, which are not effectively covered by official statistics alone.

Impact of this project
Our approach particularly benefits resource poor countries to create population statistics because they are often most in need but do not have enough capacity and resources to produce timely and actionable statistics. This is because the population coverage of the telecom data and satellite images is almost ubiquitous. Algorithms used in the developed system will be published as an open-source package on GitHub that is an open platform for sharing and building software. It enables our approach to scale in other countries.
Links Shibasaki & Sekimoto Laboratory