Bridging the gap between damage and business recovery in developing countries: a quantitative model using satellite recovery indicators

PI, Flavia de Luca

11 October 2017

On average 169 million people are affected by natural disasters each year. The poorest countries suffer the highest casualties and biggest relative economic losses following such disasters. This project will investigate the use of satellite imagery analysis as a low-cost, highly informative indicator for developing countries to measure the speed and quality of recovery.

Investigators: Dr Flavia de Luca (Civil Engineering), Dr Dina Damen (Computer Science)

This project was funded by the Cabot Institute Innovation Fund to the value of £2200

Project descriptor:

Satellite post-disaster recovery indicators can be highly informative for developing countries. The limits identified so far for the usefulness of these indicators are (i) the lack of quantitative relationships between damage, recovery indicators and economic recovery, and (ii) the relative expensive costs for the acquisition of satellite imagery. On the other hand, the quantity and quality of information freely available from satellites and mapping tools (e.g., Google Earth, Openstreet map) is dramatically increasing.

The aim of this project is to investigate the quantitative relationship between damage, recovery indicators and business recovery investigating a modelling approach suitable for developing countries based on low-cost remote sensing data.

Two case-studies are considered: Nepal and Haiti. Number of vehicles in historical satellite imagery in Google Earth and number of tagged businesses in Google Earth and Openstreet map will be collected three times during the project. Data collection will be done using Amazon Mechanical Turk for the imagery annotation in a manual and semi-automatic fashion.

Data trends will be compared with field survey data (available for the Nepal), and damage and recovery data available online (available for both Haiti and Nepal). The correlation between damage, satellite recovery indicators and business recovery will be quantitatively investigated.

The Haiti case will allow the investigation of a multi-hazard scenario given the occurrence of the 2010 earthquake and the 2016 hurricane Matthew.

The project will also include liaising with satellite imagery owners to scope the opportunity to train an automatic annotation algorithm.

[1] Platt, Brown, Hughes (2016).