Land use mapping from remote sensing is frequently used to support emergency planning. This research seeks to augment such data with ancillary information from text mining of Flickr tags and other social media data.
The background to this research relates to the some of the main justifications and selling points of investments in Earth Observation (EO) / Remote Sensing (RS): its ability to rapidly provide synoptic data of events (fire, drought, flood, earthquake) to support emergency planning. EO and RS data are easily classified to identify core, affected areas. However marginally affected areas, are frequently missed because the RS signal is weaker, and the full impacts underestimated.
Social media provides reliable and immediate information about the scale and spatial impact of those events. Thus the aim of this research is to augment traditional analyses of EO / RS data in support of emergency planning with social media data (Flickr, Twitter). Here the EO / RS inference is more uncertain, but the use of temporally and spatially specific social media (tagged images, tweets) as ancillary data will more robustly identify ‘affected areas’.
To discuss the proposed research, please contact a.comber(at)leeds.ac.uk