HomeNewsUtilizing social media to gather information on landslides in close to real-time...

Utilizing social media to gather information on landslides in close to real-time – The Landslide Weblog



Utilizing social media to gather information on landslides in close to real-time

Earlier than I begin my submit at present, as a British Citizen I need to categorical my unhappiness on the passing of Queen Elizabeth II.  She has reigned for all of my life and rather more.  She was one of the best of us.

As loyal readers will know, for a few years I collected information on landslides that trigger fatalities.  I began that work in 2003, and it has been extensively revealed and cited.  That work was made potential by the emergence of on-line information sources, and the power to scan them, plus social media.  Others have additionally collected this type of information on a world scale, most notably the workforce at NASA, and naturally there was a variety of efforts to compile related information at nationwide and regional degree.

These kind of research have modified our understanding of landslides in time and house.  However amassing the information is immensely time consuming (that means that I’ve ceased to take action, given my different commitments).  So, lately there have been makes an attempt to automate the method.  In a paper (Pennington et al. 2022) simply revealed within the Worldwide Journal of Threat Discount, Catherine Pennington from the BGS and a global workforce have taken this work a big step ahead utilizing Synthetic Intelligence (AI).  This analysis has additionally been reported in a wonderful article in EOS.

The work is intelligent.  An algorithm regularly scans social media for posts on landslides utilizing a really wide selection of phrases in a number of languages (usefully the article supplies the total checklist of those phrases).  It then harvests pictures from posts that present a optimistic indication.  AI is then used to analyse the picture to find out whether or not the submit includes a real landslide.

The algorithm has been educated with a big set of landslide pictures that have been evaluated and categorised as landslide or different by human observers.  While removed from excellent, the algorithm is performing nicely.

The workforce now have a demonstrator operating at, utilizing Twitter information: https://landslide-aidr.qcri.org/service.php#residence

Have a look, it’s petty cool.  That is the map of landslides that it has detected within the final seven days:-

Landslides detected from Twitter over a seven day interval utilizing the algorithm of Pennington et al. (2022). Picture from the International Landslide Detector.


Inevitably there are some challenges, notably together with methods to differentiate between, for instance, tweets that present latest landslides and those who present the outcomes of landslide mapping, and thus which can be older.  However science is iterative, and this is a crucial and fascinating step.



Pennington, C.V.L., Bossu, R., Ofli, F., Imran, M., Qazi, U., Roch, J. and Banks, V.J. 2022.  A near-real-time world landslide incident reporting instrument demonstrator utilizing social media and synthetic intelligence.  Worldwide Journal of Catastrophe Threat Discount, 77, 103089. https://doi.org/10.1016/j.ijdrr.2022.103089.



Please enter your comment!
Please enter your name here