28 May 2020
The July 2018 Xe Nammoy hydropower complex dam failure: a new paper
On 23 July 2018 a saddle dam at the Xe Nammoy hydropower complex in Laos failed and breached, releasing 350 million cubic metres of water. The resultant flood inundated an area of about 46 square kilometres along the Vang Ngao River, a tributary of the Mekong River basin, causing massive damage. I featured a detailed review of that event by Richard Meehan and Douglas Hamilton in 2019. They considered the cause of the failure:
An initial review of this failure by the first author was presented in late 2018, and was followed six months later by a review by an independent expert panel drawn by the Lao government from the International Committee on Large Dams (ICOLD). Both reviews concur in finding that the failure was caused by a foundation failure beneath one of the project saddle dams.
A paper has recently been published in the journal Geomorphology (Latrubesse et al. 2020) that also consider carefully the causes and impacts of this event. Whilst the paper is focused mainly on modelling and understanding the flood that resulted from the breach of the dam, it also considers the failure mechanism of the dam itself. Interestingly the authors have examined the materials from which the dam was constructed.
What is not in doubt is that heavy rainfall prior to the failure induced the breach event. However, the dam did not overtop – indeed analysis in the paper suggests that the water was at least 15 m below the crest of the saddle dam when failure occurred. This suggests that the problem was a structural problem within the dam or within its foundation. Latrubesse et al. (2020) provide this illustration of the aftermath of the failure at the saddle dam. This is the clearest picture of the failure site that I have seen:-
Note the weathered material that formed the saddle dam, sitting on top of basaltic bedrock. Images B and C show slumping in the aftermath of the breach.
The core of the dam used weathered materials quarried locally. The research team examined the characteristics of these materials. They concluded that the dam materials may have had a lower clay content than the designers had anticipated, which in turn provided a higher level of permeability than had been expected. Thus, Latrubesse et al. (2020) suggest that water penetrated into the core of the dam, driving piping and, ultimately, triggering a rotational failure in the dam itself, which then allowed the breach to occur.
This mechanism of failure is a hypothesis rather than a definitive analysis. But of course it is interesting at this point because of the similarity in mechanism to the failure of the Edenville Dam earlier this month.
On reflection 1: an official report on a collision between a train and a landslide in the UK in 2019
The UK Rail Accident Investigation Branch has published a report into a collision between a train and landslide debris at Corby in Northamptonshire on 13 June 2019. Key finding:
The investigation found that the cutting slope had failed because it was not designed to cope with a large volume of water that had accumulated at its crest. Flood water had accumulated at the crest because two adjacent flood storage ponds had overfilled with water from a nearby brook.
On reflection 2: Coastal rockfalls in Sidmouth, Devon
The coastal cliffs of Sidmouth in Devon, in the Southwest of the UK, have undergone three significant collapses in a 24 hour period. Many parts of the UK are undergoing an exceptionally dry Spring, so the failures are generating large plumes of dust.
Latrubesse, E.M, Park, E., Sieh, K. et al. 2020. Dam failure and a catastrophic flood in the Mekong basin (Bolaven Plateau), southern Laos 2018. Geomorphology, 352, 107221.
27 May 2020
The aftermath of the Sanford dam failure in Michigan
The catastrophic failure of the Edenville Dam in Michigan last week was not the only dam failure that day. About 16 km downstream from Edenville lay the smaller Sanford Dam, which also failed. In a sense this is an understandable collapse – the dam would not have been constructed to withstand the flows associated with the failure of the Edenville Dam. It would perhaps have been surprising if the structure had withstood such an event.
Planet Labs have collected images of the aftermath of the failure of the Sanford Dam. This is a Google Earth image, collected in November 2018, which shows the site of the dam:-
Note that the fuse plug, designed to allow an emergency increase in flow to prevent dam failure, is clearly visible. This is the Planet Labs high resolution SkySat image of the aftermath of the failure:-
The terrible flood damage downstream of the breach is all too evident. The dam itself has been almost completely removed.
Rebuilding these sites is going to be a long and expensive process.
An early failure similar to the Edenville Dam?
Meanwhile, in the comments on my earlier posts about the Edenville Dam, Bruce Feinberg has noted the similarity with the failure of the Kelly Barnes Dam on 6 November 1977 at Toccoa in Georgia, USA. This was another earthen dam that breached during heavy rainfall, killing 39 people. The USGS investigated this failure, and the report is online. The report makes shocking reading – the dam was poorly documented and in a very poor state of repair at the time of failure. Photographs from 1973 show that a slope failure in the face of the dam had already occurred. The USGS report hypothesises that the breach may have been caused by a further slope failure in the downstream face of the dam:
[Slope failure] appears to be a distinct possibility, particularly on the downstream slope when the previous slope failure is considered along with the possibility of the development of tension cracks upslope of the previous failure together with a computed factor of safety that is marginal. The long period of rain would have saturated tension cracks, if they existed, and the entire downstream slope would have become essentially saturated and even more susceptible to failure. A local downstream slope failure similar to that observed in 1973 could have caused limited breaching allowing localized overtopping. This concept would corroborate the hydraulic computations.
This proposed mechanism of failure is indeed similar to that of the Edenville Dam.
On reflection 1: no resting place
A landslide has disturbed Vicksburg National Cemetery in Mississippi, requiring archeologists to relocate the remains of the Civil War Union soldiers. Work will now be undertaken to stabilise the slope.
On reflection 2: A landslide video from Colombia
A nice landslide video was captured on the Florencia – Neiva highway in Colombia:-
La vía Florencia – Neiva permanece cerrada desde la madrugada de este martes. En video quedó registrado uno de los derrumbes en el kilómetro 40, entre Suaza-Florencia, que impiden el tránsito vehicular #VocesySonidos pic.twitter.com/JrEV5AR1H7
— BluRadio Colombia (@BluRadioCo) May 27, 2020
26 May 2020
Punatsangchhu I: identifying ancient landslides in high mountain areas
There is a new paper in the journal Scientific Reports (Dini et al. 2020, available online) about the use of InSAR to study the movement of the a large rock slope adjacent to the large Punatsangchhu-I hydroelectric plant, which has been under construction for over 11 years. Punatsangchhu I will be a 134 m concrete gravity dam, built to supply electricity to Bhutan and India.
The project is running late and over budget. In 2006 the estimated cost of the project was $554 million. By July 2015 this had reached $1.74 billion. The project was meant to be completed in November 2016, but in April 2017 this had slipped to December 2022, a delay of over six years. BBS reported in February 2019 that the completion date had slipped to 2024; I assume that the budget has escalated as a result.
The primary cause of this delay is a landslide. In July 2013 the east-facing bank of the project site failed and displaced by more than 5 metres. This landslide has required extensive mitigation. However, the project suffered another major, and on this occasion fatal, landslide January 2019.
Google Earth imagery shows the extensive works underway to mitigate the east-facing slope, allowing the abutment of the dam to be completed. The Google Earth image below shows the site in 2017. I have placed the marker at what appears to be the crown of a slope failure, although Dini et al. (2020) suggest that it extends further up the slope:-
The paper by Dini et al. (2020) uses InSAR to monitor the movement of the east-facing slope at the Punatsangchhu I dam site. One of the great things about InSAR is that it allows us to go back in time to look at movements that occurred 15 or more years ago. Their findings make uncomfortable reading. First, the slope was displacing in 2007, even though construction started in 2009. In other words, the dam appears to be located by a large, ancient creeping landslide. Second, portions of the slope have continued to move throughout the construction period, and indeed the area of active displacements continued to expand right through to 2018, the end of the study period. The paper notes that:
Stabilisation measures currently only focus on a small portion of the slope, however, the unstable area is larger than previously evaluated.
And third, that the site is probably a large, ancient landslide. Dini et al. (2020) have found evidence of movement over an area of 8 km2. They include this Google Earth image of the site:-
The InSAR data in Dini et al. (2020) shows different areas of movement across this large slope, covering about 60% of the total area. The authors include that this might be a deep seated gravitational displacement.
This tale is a classic illustration of the importance of site investigation and of understanding the behaviour of the landscape in these high mountain areas. Loyal readers will know that I have highlighted previously that many large hydroelectric schemes in high mountain areas are not managing slopes sufficiently, with tragic and expensive consequences. The Google Earth image below shows the upper portion of the large rock slope. The morphology here, of a steep upper section and then a more planar mid-slope, would always make me deeply suspicious about the presence of an ancient landslide, and indeed when I worked in the Himalayas back in the early 2000s that is how we mapped such sites:-
so, I’m deeply intrigued as to whether this large slope was identified as being problematic before the project started.
On reflection 1: landslides at the 2018 tsunami in Sulawesi
The debate about the role of landslides, versus tectonic deformation, in the initiation of the 2018 Sulawesi tsunami rumbles on. A new paper concludes that landslides probably did play a role.
On reflection 2: A lucky escape
In Sudbury, Canada a man was rescued after spending more than 12 hours trapped in a landslide. He has a broken arm.
Dini, B., Manconi, A., Loew, S. et al. 2020. The Punatsangchhu-I dam landslide illuminated by InSAR multitemporal analyses. Scientific Reports 10, 8304 (2020). https://doi.org/10.1038/s41598-020-65192-w.
22 May 2020
Edenville Dam breach: interpreting the failure
Many thanks to all of those who contributed to the discussion yesterday about the catastrophic Edenville Dam breach in Michigan. I thought it would be helpful to summarise views expressed, many of which have come from experts in the field.
First, it is clear that this was not an engineered failure – in other words, it was not planned. There was some discussion on Twitter and in the comments that this was the failure of a fuse plug – i.e. a designed failure point that would release water to prevent overtopping. I can find no evidence that Edenville Dam had a fuse plug, and I do not think that a fuse plug failure would behave in the way shown in the video.
Planet Labs have a wonderful high resolution image of the aftermath of the failure; I wpould be surprised if a fuse plug is intended to leave this type of catastrophic breach:-
The style of failure implies that the dam had become saturated in this area. A key question is going to be why this happened. One suggestion is that the water level exceeded the impermeable barrier, allowing water to flow into the structure. An alternative is that the dam was suffering from seepage prior to the floods. The Google Earth imagery is interesting – this image is from 2018:-
Is there an indication here that there was deformation in the dam? Or that works had been undertaken? I’m not sure. It will be interesting to see both the monitoring records for the dam and the maintenance that had been undertaken, as well as the design cross-sections.
The mechanism of failure is undoubtedly a rotational slip. It is possible that this started as a smaller failure at the crest of the dam, which then drove a larger failure in the main face. However, I favour the interpretation that high pore water pressures, and a loss of unsaturated conditions, through the dam volume drove the failure. There are some indications in the video that high pore water conditions were present in the lower part of the structure.
Readers have rightly pointed out that earthfill embankment dams are not unusual and, when well designed and maintained, they are not unsafe. This dam was completed in 1924. However, these structures do require maintenance – would you expect a train built in 1924 to still work without extensive restoration – and they were designed for a time when rainfall levels were different. Climate change – global heating – is driving increases in rainfall intensities and durations, meaning that the Probable Maximum Flood is increasing in very many places.
I always get howls of protest when I say that climate change is important, but it is the case. These structures, worldwide, are going to need a substantial upgrade to cope with that increase in rainfall, and that’s going to be very expensive. In the interim we will see more failures of this type.
There is also some interesting analysis of the performance of the dam prior to failure online, using INSAR data. It is astonishing that such an interpretation can be generated so quickly. At present I find it hard to interpret this data though – the results seem to indicate deformation across much of the structure, and the section that failed seems to show uplift not, as I would expect, subsidence. This needs further work, but INSAR remains an exceptionally exciting area of work for these types of investigations, and for pre-failure monitoring.
The failure of the dam is a catastrophe for people living in this area. Planet Labs have an online gallery of high resolution images of the impacts. The Planet Labs image below shows some of the downstream flooding for example:
But we must not forget that the effects lie upstream as well. There are numerous houses located around the lakes whose value will have been based upon the proximity to the water. The failure of the dam will have a profound impact, and of course the ecology of the lakes will also have been destroyed.
On reflection 1: my most successful blog day
The past 24 hours have been my most successful day in the 12 years of this blog, with almost 50,000 individual visits. Thank you.
On reflection 2: Controversy over the protection of the Great Western Railway in Devon from landslides
In Teignmouth in Devon in SW England there is controversy over plans by Network Rail to change the alignment of Isambard Kingdom-Brunel’s wonderful Great Western Railway to protect it against landslides. The article claims that the cost of these works is £500 million.
Planet Team (2020). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://www.planet.com/
21 May 2020
Edenville Dam failure: the astonishing video of the collapse sequence
Yesterday I posted about the remarkable and devastating failure of the Edenville Dam in Michigan on Tuesday 19 May 2020. There are now heartbreaking images online showing the damage that the resulting flood has caused downstream. Given the known weakness of the dam, this is an unacceptable situation.
From a technical perspective, the most remarkable aspect of this failure is that the sequence of events that induced the breach is caught on video. I tweeted about this yesterday, but it turns out that the landslide that initiated the collapse was also captured, by Lynn Coleman, and posted to Youtube by MLive. This video is astonishing:-
As always we should crowdsource an interpretation of this sequence of events, but allow me to give an initial (and not definitive) interpretation. At the start of the video a small amount of water is seen to have overtopped the dam:-
It would be tempting to surmise that this was a simple overtopping, but I don’t think that is correct. I think the video shows that the crest of the dam has deformed and dipped, creating a depression through which water has started to flow. In other words, the video starts with the dam wall undergoing the early stages of failure, which in turn has allowed a small amount of overtopping.
The Edenville Dam failure then develops apace. The slope fails rapidly, initially forming a large toe bulge and there is major deformation at the crest:-
The failure is rapid and mobile – the still below is only two or so seconds later. Note the blurring of the toe of the landslide due to the rapidity of motion. there is also a hint of some dust or vapour in this area, and above the main body of the slide, which is interesting too:-
The landslide has clearly not failed through the full width of the dam as there is no sign of water pouring through. However, it is likely to have left only a very slender thickness of dam in place. I would hypothesise that this rapidly collapsed under the pressure of the reservoir water as a few seconds later water appears on the landslide deposit:-
The full breach rapidly develops after this.
This video is going to be a classic in the teaching of geotechnical failures, but it also clarifies the events that led to the Edenville Dam failure. It would have been simple to ascribe this to a simple overtopping event that occurred when the capacity of the spillway was exceeded. But in reality the events are are more worrying than that – the dam appears to have undergone a slope failure; a failure of its integrity. This should never occur, and to me it suggests that the problems at the Edenville Dam went further than known issues with the spillway.
Comments and thoughts welcome please. Please also read my follow-up post to understand this event.
On reflection 1: Cyclone Amphan
Cyclone Amphan made landfall on Wednesday and is now bringing heavy rainfall to NE India and Bangladesh. There are some early indications of landslides; we await to see the full impact, probably tomorrow or on Saturday.
On reflection 2: landslide damage to a chairlift in Alberta
A landslide on Tuesday 19 May 2020 damaged a chairlift at Nitehawk Adventure Park, a small ski field located in Grande Prairie, Alberta. From the images the damage is serious.
20 May 2020
Edenville dam: a major dam collapse in Michigan
In Midland County, Michigan, USA a dam collapse is underway, driven by heavy rainfall. CNBC has a good article about the ongoing accident, which is causing extensive flooding. Unfortunately, as I write, reports are coming in that a second dam, at Sanford in the same area, has also breached.
The Edenville Dam is located at 43.813, -84.376. MLive has excellent aerial footage of the breach, and the extensive downstream flooding:
The town of Edenville is located about 1 km downstream of the dam:-
At the moment the cause / mechanism of the failure is not clear. ABC12 has a report from 2018 that the licence for the dam had been withdrawn by the Federal Energy Regulatory Commission because the structure had insufficient capacity to handle the Probable Maximum Flood. The FERC ruling is available online, and states the following:
Of particular concern is the project’s inability to pass the Probable Maximum Flood (PMF) due to inadequate spillway capacity … Currently, spillway capacity at the Edenville Project can only pass about 50 percent of the PMF.
NBC25 reports that the designs for a remediation of the hazard were being prepared, with construction anticipated in the period 2021 to 2023.
On reflection 1: Cyclone Amphan
Cyclone Amphan will make landfall in a few hours from now in NE India; this remains an extremely dangerous storm. As I have noted previously, although we categorise tropical cyclones on the basis of wind strength, most of the damage is typically caused by water. India and Bangladesh have well-established systems for evacuating people exposed to storm surge (although in the time of Corona Virus these will be tested to the maximum), but this storm is likely to cause substantial levels of inland damage from flooding and landslides.
On reflection 2: submarine landslides in the Gulf of Mexico
New research, published in Geophysical Research Letters, has detected 85 previously unknown submarine landslides in the Gulf of Mexico. Most of these landslides were triggered by the passage of seismic waves from distant earthquakes. The study suggests higher than anticipated levels of hazard to both underwater infrastructure and coastal communities.
19 May 2020
A long run out landslide from Yudi Peak in Alaska
With thanks from me to loyal reader Hig for pointing it out, the Facebook page of Alpine Air Alaska has some stunning photographs, and a very cool video, of a long run out landslide that has recently occurred on Yudi Peak, near to Girdwood in Alaska. This is one of their images:-
The crown of the Yudi Peak landslide is located at 61.03, -148.96 at an elevation of about 1430 m. It has a run out distance of about 2 km, stopping at an elevation of about 1009 m. This image, also posted to the Facebook Page of Alpine Air Alaska, shows the full length of the landslide:-
From the images, that landslide appears to have initiated as a wedge failure. The volume is not very large. but the terrain is steep and the initial sliding surface was snow and ice, providing a high level of mobility. There may have been some entrainment once the landslide reached the bedrock areas. The best view of the landslide is on a video that Alpine Air Alaska have flown of the full length of slide. It is very rare to have such good coverage of the full track of a slide.
The landslide is first visible on satellite imagery on 10 May 2020. It is not visible on an image on 6 May 2020, so it occurred sometime in that window.
It is well established that Alaska sees large landslides in the Spring, and that their increasing size and frequency is driven by global heating.
On reflection 1: landslide threat from Cyclone Amphan
Cyclone Amphan is now an extremely dangerous tropical cyclone, tracking northwards to make landfall, probably in West Bengal, tomorrow. Whilst it will weaken a little in the next 24 hours, it has the potential to cause very substantial levels of damage, compounded of course by the increased vulnerability of the population as a result of Covid-19. This storm will bring extreme rainfall to the hilly areas of Bangladesh, NE India and possibly E. Bhutan. The potential for severe landslides and floods is high.
The ever wonderful Save the Hills blog, which features community efforts to manage landslides in the Kalimpong area on NE India, is likely to feature some of the impacts. I await the outcome of this storm with some trepidation – expect news of landslides towards the end of the week.
On reflection 2: landslides from a Mw=5.2 earthquake in China
Xinhua reporters on Tuesday morning saw rescuers approaching Yakou Village, Xiaohe Township in the epicenter of Qiaojia. Excavators are removing debris of landslides along the road.
18 May 2020
The current landslide crisis in East Africa
In recent months I have highlighted the increasing occurrence of fatal landslides in Africa associated with bouts of heavy rainfall. I have noted on a number of occasions that I seem to record more landslides now in this region than was the case in the past. However, this was an observation rather than a measurement, so I decided this morning to take a look at the data in more detail.
I have good quality data on fatal landslides for this region from 1 January 2004 to the present. The graph below shows the cumulative number of fatal landslides in this region from 1 January 2004 to 16 May 2020 (I have yet to update the data for the weekend just gone):-
There is a very marked steepening of the curve in the data. In the early years of this work I recorded few fatal landslides in this region, perhaps anomalously so. From about 2008 this seems to have changed; perhaps at this point reporting from this part of the world improved, or perhaps there was an genuine increase in the number of landslides. There are a number of steps in the graph, marking periods of particularly heavy rainfall.
But from about 2018 the number of fatal landslides has markedly increased, and at the moment the curve is much steeper than it has been previously in the record. The key question is of course why?
Seasonal rainfall in East Africa is controlled by the Indian Ocean Dipole, which is the local equivalent of El Nino. The BBC Website has a good explanation of the Indian Ocean Dipole. The graph below shows an indicator of the state of the Indian Ocean Dipole, using a NOAA dataset called the Dipole Mode Index (DMI):-
In the very early years of my dataset the DMI was largely negative, which might explain the low number of fatal landslides. But what is marked here is that since 2018 the value has been positive, and indeed reached a level not seen anywhere else in the period snce 2004.
Others have suggested that the current prolonged period of heavy rainfall in East Africa might be associated with this unusual pattern in the Indian Ocean Dipole. The fatal landslide data would seem to support this view.
On reflection 1: More fatal landslides at Ramban
Recently I highlighted the high incidence of fatal landslides on the Srinagar-Jammu National Highway. Over the weekend two more people were killed in the infamous Ramban area.
On reflection 2: larger earthquakes trigger larger landslides
A new study by researchers from the USGS and NASA shows that larger earthquakes trigger larger landslides.
15 May 2020
A potential major rock slope failure above Barry Glacier in Alaska
Yesterday, officials at the State of Alaska released a statement highlighting the risks of a major rock slope failure above the terminus of Barry Glacier in Alaska. This was prompted by a letter released by a group of scientists that highlighted the problem. The scientists’ statement starts:
We, a group of scientists with expertise in climate change, landslides, and tsunami hazards, have identified an unstable mountain slope above the toe of Barry Glacier in Barry Arm, 60 miles east of Anchorage, that has the potential to fail and generate a tsunami. This tsunami could impact areas frequented by tourists, fishing vessels, and hunters (potentially hundreds of people at one time). We believe that it is possible that this landslide-generated tsunami will happen within the next year, and likely within 20 years.
The threat of a large and potentially dangerous tsunami is looming in Prince William Sound, where an increasingly likely landslide could generate a wave
with devastating effects on fishermen and recreationalists using the area, the state’s top geologist said today.
The problem is a large, creeping rock slope that shows in the order of 185 m of deformation in some places. Failure of the slope could generate a tsunami in Harriman Fiord. The landslide is quite clear on Google Earth, although the image is now three years out of date. The location is 61.152°, -148.165°:-
This is a very large landslide – the scientific team estimate a volume of about 500 million cubic metres. Such large rock slope failures have a well documented track record of generating truly enormous local tsunami, with the Lituya Bay event in 1958 being the archetype of course. But, as I highlighted earlier this week, the behaviour of large rock masses is hard to anticipate.
In this case, there is an urgent need for monitoring, backed up by a detailed investigation of the landslide and its constituent materials. The science team has outlined an appropriate approach:-
To inform and refine hazard mitigation efforts, we would like to pursue several lines of investigation: Detect changes in the slope that might forewarn of a landslide, better understand what could trigger a landslide,and refine tsunami model projections. By mapping the landslide and nearby terrain, both above and below sea level, we can more accurately determine the basic physical dimensions of the landslide. This can be paired with GPS and seismic measurements made over time to see how the slope responds to changes in the glacier and to events like rainstorms and earthquakes. Field and satellite data can support near-real time hazard monitoring, while computer models of landslide and tsunami scenarios can help identify specific places that are most at risk.
The State of Alaska is taking this problem seriously. Its statement indicates that they will take a proactive approach:
The Division of Geological and Geophysical Surveys looks forward to cooperating with interested agencies and institutions to monitor the situation. One important step would be to monitor movement of land in the area by placing solar-powered GPS monitors on the potential landslide area to detect increased rates of movement that frequently precede catastrophic landslide failures.
The problem is being reported in the media in Alaska. Responding to these types of problems is challenging at the moment because of the Corona Virus crisis, but this does require attention. Failure is not inevitable, but the effects could be catastrophic.
On reflection 1: the 2019 fatal landslides database
Many thanks for those who have contacted me with corrections to the 2019 fatal landslide database. I’ll try to get a revised version online in the next few days and will highlight it here.
On reflection 2: Landslide problems in Kenya deepen
Kenya has been struggling for weeks with massive landslides caused by heavy rainfall. The problems have now deepened as a landslide has destroyed the water supply to the Kangemi slum on the outskirts of Nairobi. This is home to about 20,000 people.
Many thanks to Hig – Dr Bretwood Higman of Ground Truth Alaska – for keeping me updated about the Barry Glacier landslide.
14 May 2020
The 2019 fatal landslide dataset
Yesterday I posted a map of the events in my fatal landslide dataset. In the post I provided the map as a diagram and as a dynamic Google Map that you can use to examine any given area. I posed the question as to whether it would be helpful if I made this data available online. The resounding reply was yes.
So, I have put the 2019 fatal landslide dataset into a Google Sheet (the equivalent of Excel) and have made this available. I summarised this data in a post a few weeks ago, although note that there have been some minor changes since then as I have refined the dataset. Such data is not static as I discover more events, or am able to provide better estimates. I should point out that the 2004-2017 dataset is available online as a GIS tool.
I have two motivations for posting this data:
- In the spirit of an age of open data I’m keen to share the information, and am eager for it to be used. I’m happy for it to be incorporated into research, to be exploited for teaching (especially now that we’re all having to teach largely online) and to be used by practitioners. I would ask only that you cite the two papers that I have listed at the foot of this post.
- I’m also keen to have corrections and/or additions. There are uncertainties in this sort of data, of course, and you may well understand better than do I as to whether I have recorded events correctly. I’m particularly concerned that I miss events in Latin America (most particularly in Central America and in countries such as Venezuela – see below), and I worry that I capture far fewer events in China than was the case in the past.
If people are able to provide corrections then I will make the improved data available. The Google Sheet itself is locked (i.e. you cannot change the online data).
The spreadsheet should be straightforward. I should explain a few columns:
Column d: this is the pentad data that I sometimes use to plot graphs. It divides the year into 73 blocks of five days, so that pentad 1 is 1 to 5 January, pentad 2 is 6 to 10 January, etc. Plotting data like this makes it much easier to compare between years
Column j: this is an estimate of the accuracy of the location data, suing the code given at the top of the spreadsheet. You will see that for a small number of locations I do not have sufficient information to plot a meaning location.
Columns o and p: I separate fatalities and injuries from landslides triggered by earthquakes from those from other causes, simply to ease the analysis of the data.
Column q: This is the cumulative number of landslide fatalities for the year, excluding events triggered by earthquakes
Column r: This is the cumulative number of landslide events for the year, excluding events triggered by earthquakes
Column t: Sometimes reports identify a primary trigger for the event (recorded in column s) but also a secondary factor (recorded in column t).
The data does allow you to do some quite cool stuff, such as demonstrate that it is the western (mountainous) side of Colombia that is affected by landslides, but it also asks the question as to whether I am missing similar events in in Venezuela and Ecuador?
As always I welcome your thoughts.
A rather spectacular bridge failure has been induced by a landslide at Cinyiru in Banten Province in Indonesia. The report suggests that “We have dispatched a backhoe and construction workers to repair the bridge, so that cars can again ply on it.” I think that is quite optimistic:-
Froude, M. J. and Petley, D. N. 2018. Global fatal landslide occurrence from 2004 to 2016. Natural Hazards and Earth System Science, 18, 2161-2181, https://doi.org/10.5194/nhess-18-2161-2018.
Petley, D.N. 2012. Global patterns of loss of life from landslides. Geology 40 (10), 927-930.