Flooding through the lens of mobile phone activity

TL;DR

This paper explores how mobile phone activity data can be used to characterize and detect floods, providing insights for emergency response and early warning systems.

Contribution

It demonstrates the potential of combining CDR data with satellite imagery and rainfall data to identify flood impacts and abnormal human activity patterns.

Findings

Abnormal activity patterns correlate with flood-affected areas.

Cell tower activity signals can serve as signatures of flooding.

Preliminary validation shows promise for emergency management applications.

Abstract

Natural disasters affect hundreds of millions of people worldwide every year. Emergency response efforts depend upon the availability of timely information, such as information concerning the movements of affected populations. The analysis of aggregated and anonymized Call Detail Records (CDR) captured from the mobile phone infrastructure provides new possibilities to characterize human behavior during critical events. In this work, we investigate the viability of using CDR data combined with other sources of information to characterize the floods that occurred in Tabasco, Mexico in 2009. An impact map has been reconstructed using Landsat-7 images to identify the floods. Within this frame, the underlying communication activity signals in the CDR data have been analyzed and compared against rainfall levels extracted from data of the NASA-TRMM project. The variations in the number of active phones connected to each cell tower reveal abnormal activity patterns in the most affected locations during and after the floods that could be used as signatures of the floods - both in terms of infrastructure impact assessment and population information awareness. The representativeness of the analysis has been assessed using census data and civil protection records. While a more extensive validation is required, these early results suggest high potential in using cell tower activity information to improve early warning and emergency management mechanisms.