Combination of high resolution satellite images population census statistics to estimating population exposed to risks of disaster Daniel Clarke (UNESCAP) Jean-Louis Weber (Consultant-UNESCAP)
Disaster Related Statistics Disaster Related Statistical Framework (DRSF): response to the Sendai Conference of 2014 by UNESCAP Part of / connected to the overall statistical system: population, society, economy, environment Support to the production of indicators and to policies Expert Group composed of statisticians and officials from disaster management agencies: Statistics: broad vision, for comparisons, solidarity mechanisms Operational Databases: accuracy of real time information for action Tests with pilot countries: collection of statistics, test of estimation of population leaving in disaster prone areas
Why population exposure to disaster is important variable? SDG Disaster risk reduction indicators 1.5 based on Sendai Target 2: Substantially reduce the number of affected people globally by 2030, aiming to lower the average global figure per 100,000 between 2020-2030 compared to 2005-2015 : sprawling of human settlements in zones prone to disaster has an effect adverse to Target 2 Operational disaster risk management: planning of emergency actions requires knowing exposed population to be able to act swiftly and efficiently
Why population exposure to disaster is important variable? SDG Disaster risk reduction indicators 1.5 based on Sendai Target 2: Substantially reduce the number of affected people globally by 2030, aiming to lower the average global figure per 100,000 between 2020-2030 compared to 2005-2015 : sprawling of human settlements in zones prone to disaster has an effect adverse to Target 2 Statistics Operational disaster risk management: planning of emergency actions requires knowing exposed population to be able to act swiftly and efficiently
What do we want to estimate? Risk zones (here: floods in South Korea - from UNEP GRID data base, historical risk) and population exposed
What do we want to estimate? Risk zones (here: floods in South Korea) and population exposed (in yellow/orange/red, below the grid )
Population census In several countries (and more and more): data by PSU (Primary Sampling Units) = municipalities, villages, wards At the international scale (for mostly all countries), statistics by Districts (Admin 2) need to estimate population density in a grid
A new data source: Global Urban Footprint GUF is a map of build-up areas produced by the German Aerospace Agency / DLR from radar satellite images of 2012, using the European Space Agency TEP cloud computing system. Data are sensed by TerraSAR-X and TanDEM-X radar satellites and images acquired at 3m ground resolution. Built-up areas pixels are derived at 12 m resolution and generalized at ~80m and now 30 m).
Google map and the GUF picture (here, 80 m pixels)
Google map and the GUF picture (here, 80 m pixels)
What do we map: the Google satellite view and the GUF pixels
Combining GUF smoothed and ADM2 POP2015 for Java A validation check with population density per villages
Bridging scales: Population-to-GUF with Population census statistics at the District (top) and Villages (bottom) levels
Bridging scales: Population-to-GUF with Population census statistics at the District and Villages POPtoGUF estimations at the District /Region level: National policies (orientations, budgets ), land planning, international cooperation (Sendai, Climate Change, SDGs ) and solidarity, statistics and indicators POPtoGUF estimations at the local level: planning for action, preparedness to face disasters, emergency responses, recovery (can be carried out with the same methodology) Novel statistics have to be fed by operational information = extract as much as possible from disaster management databases Big data are great opportunities to access information previously inaccessible
Thank you for your attention. Emergency management room at BNPB, the Indonesian Disaster Management Agency