13 May 2011

The third anniversary of the Wenchuan Earthquake – loss of life in seismic landslides

Posted by Dave Petley

Yesterday, (12th May) was the third anniversary of the Wenchuan Earthquake.  To commemorate this event, the Chengdu University of Technology has been organising a conference examining earthquake induced landslides.  My presentation is in the morning session today and will focus primarily on the loss of life in landslides triggered by earthquakes.  In the talk I will also mention our Nature Geoscience paper, which will be published on Sunday,which looks at the mass balance of the mountain chain.  I will blog about this (very exciting) paper on Sunday when it has been published.

The loss of life component of the talk is included in the PDF that is available here: 11_05 China keynote Petley blog

I will try to describe the key aspects of this talk below.

Regular readers will be aware that one aspect of my research is to collect data on the occurrence of landslides that kill people worldwide. I have been doing this since September 2002; occasionally I publish this data on this blog. In this presentation I examined this data to look at the occurrence of landslides triggered by earthquakes. Overall in the period from 1st September 2002 to 31st December 2010 I recorded a total of almost 82,000 fatalities from landslides. This figure shows the cumulative total of fatalities during this period:

You will see that the data are dominated by two earthquake events – the Kashmir earthquake of 2005 and the Wenchuan earthquake of 2008.  Unfortunately, there is some uncertainty in terms of the actual losses from landslides in the former event; the best estimate according to official figures is in the order of 26,400 lives.  The published figure for the Wenchuan Earthquake is about 20,000 fatalities.

Interestingly the trend of losses with time from rainfall induced landslides is approximately constant with time:

The cumulative trend for earthquake-induced losses shows the impact of the two large events, but in fact over this period a total of 32 earthquakes that triggered landslides responsible for fatalities have occurred:

Interestingly, this means that more than 50% of the people killed by landslides during this period died in events triggered by earthquakes.  This of course emphasises the very urgent need to improve our understanding of seismically-triggered events.

The data (in closed – i.e. black – symbols below) can be combined with information from the literature (the open symbols) to provide an initial examination of the relationship between the number of landslide-induced fatalities and the earthquake magnitude.  There is huge scatter in the data, but perhaps two thresholds can be drawn.  The first, the orange line, represents the limit for 90% of events; the second, the red line, represents the maximum boundary for all events:

This analysis is somewhat tentative at the moment, and more work is needed.

Finally, much is (rightly) made of the increase of landslide losses in the aftermath of earthquakes.  I have looked at my data for Sichuan Province before and after the earthquake.  The data do seem to suggest a marked increase in recorded fatalities:

This implies that there has been an additional c.300 fatalities from landslides in the aftermath of the earthquake, although more work is needed on this.  Interestingly, there is no obvious increase in the number of fatal landslides occurring, but rather an increase in the average number of fatalities per event.  This may imply that in the aftermath of the event I am failing to record the smaller landslide events:

All of this serves to emphasise that the data presented here probably consistently under-estimates the true costs of landslides.  Actual losses are undoubtedly rather higher, although I must emphasise the uncertainty associated with the loss of life from the large earthquake events.

Comments and thoughts are welcome!