British Geological Survey, Durham University, ICIMOD, NASA, and University of Arizona

Please note: A higher-resolution version of the map, and black-and-white versions, are available from: ftp://topdog.dur.ac.uk/njr/nepal_eq/UPDATE_20150508/

Key messages (all observations based on available cloud free satellite imagery)

• Major or disruptive landsliding is limited to a zone that runs east-west, approx. parallel to the transition between the Lesser and High Himalaya. This zone includes parts of the districts of Gorkha, Dhading, Nuwakot, Rasuwa, Sindhupalchok, Dolakha, Ramechhap, and Khavre. A VDC level list of potentially affected areas is provided (see Appendix 2).
• We observe several large valley-blocking landslides (Appendix 1). Most observed landslides comprise comparatively small rockfalls and debris avalanches that have a significant local impact and loss of life.
• The need to plan measures to mitigate landslide deaths and disruption in the forthcoming monsoon season must remain a priority. Many more damaging landslides may occur in the 2015 monsoon, likely to start in late June, than would be expected if a major earthquake had not occurred.

Background and methodology

The UK mapping team has worked to three objectives that were defined by the UK Department for International Development: identification of large landslides (1) blocking valleys and potentially causing significant secondary hazards; (2) affecting villages, and (3) disrupting other infrastructure, such as roads. The team has undertaken a thorough investigation of available satellite imagery from multiple sources.

Landslide observations have been interpreted from satellite imagery by highly-experienced geoscientists. If additional data have been included (e.g. Tomnod crowd-sourced information on damage) they have been clearly identified on the database and on the maps.

The satellite imagery has been sourced from the International Charter Space and Major Disasters as well as directly from data providers. Pre- and post-earthquake imagery was used to ascertain that landslides were active post-earthquake; i.e. either there was no landslide present on the landscape pre-earthquake or the pre-existing landslide had moved visibly on the imagery since the earthquake.

The coverage, detail and quality of the landslide inventory are primarily determined by the available imagery. The main challenge has been to obtain cloud-free optical imagery and activities are still ongoing to source these data. The imagery has been investigated at 1:5,000 to 1:10,000 scales (with smallest features that can be distinguished ranging from about 5 to 15 m). The image spatial resolution ranged from 2 m to more than 22 m – some images were greyscale (panchromatic) while others were colour (multispectral). At 22 m resolution it is only possible to distinguish major events (larger than ~100 m on the ground) and the detail is not available to ascertain if they are solely associated with the earthquake or are part of the natural landscape processes in the area.

The team has obtained access to satellite imagery covering c.55,000 square kilometres with approximately 25% cloud-free (figure 1). Analysis of imagery from the high Himalayas by the Charter Project Manager’s institution (NRSC-ISRO) has also been evaluated by the UK team.

The imagery was loaded into a Geographic Information System (GIS) and points / lines /polygons digitised on screen, directly onto the imagery. In some cases it was not possible to obtain the satellite imagery but they could be viewed on the supplier’s web portal, in which case the landslide information was traced out on the portal and transferred by eye to the GIS. Maximum uncertainties on landslide locations are ~100 m, but vary across the area depending on the type of imagery used.

Figure 1. Post-earthquake optical satellite imagery of Nepal (current to: 7 May 2015)

We summarise some preliminary conclusions:

Siwalik and Lesser Himalayan regions

• We have high confidence that there are few very large landslides. Most landslides are small; many are re-activations.
• Analysis of villages/infrastructure known to have been damaged by landslides shows that the latter are almost all small and extremely difficult to identify using the available imagery.

High Himalayan region

• We have identified several large valley-blocking landslides (landslide dams) on the basis of satellite imagery (see Appendix 1). We stress that further valley-blocking landslides may still be found as additional imagery becomes available; other areas in the High Himalayas should be assumed to be at risk until the existence of landslide dams can be confirmed or ruled out.
• A zone of relatively intense landsliding and disruption can be mapped in every valley for which cloud-free imagery is available, north and northeast of Kathmandu (Appendix 2). This zone contains numerous rockfalls and debris avalanches, with many events following pre-existing pathways or channels and having long runout distances (in most cases, all the way to the valley floor). These events are individually localised but together are extremely widespread, and have had catastrophic impacts on roads and villages. It is these small, but widespread, landslides that are likely to have caused the largest loss of life due to landsliding.
• Cloud-free images are not yet available for all valleys within this zone; until they are, we must expect that similar levels of disruption are present in ALL valleys within the zone, not just those indicated in the map and in Appendix 2.
• Where visible, high-elevation areas (> 3,500 m) show evidence of some large rockslides, along with extensive avalanching (the latter especially above 4,500 m). Most of these areas are far from permanent settlements or infrastructure, and pose no direct threat to the population.
• Large areas of the High Himalaya, including valley floors and also most areas above 2,500 m, have not yet been examined because of cloud cover.

Entire earthquake affected region

• The zone of intense landsliding corresponds to areas with high rates of seasonal rainfall-triggered landslides. We must thus expect that the risk due to continued landsliding in this area will escalate significantly as the rains begin. This continued landslide occurrence will threaten already-affected areas with continued failure, and will wash landslide sediment downstream onto valley floors and floodplains.

Landslides and major roads

The Arniko Highway (north to Tibet) is reported to be blocked from near Chaku to the Tibetan border by numerous rockfalls. This area has had chronic landslide problems prior to the earthquake, and the road is regularly impassable during the monsoon. The pre-existing Sun Koshi landslide dam that crossed the Arniko Highway at Jure appears intact.

The Prithvi Highway (the main route from Kathmandu south to India) has also experienced ongoing landslide problems since the earthquake, including large rockfalls onto the road. This road has suffered extensive landsliding in the past. In August 2000 a single landslide blocked the road for 11 days, leading to severe shortage of fuel and other commodities in the metropolitan area.  

Recommendations

1. Before the onset of the monsoon, the potential valley-blocking landslides and heavily-impacted valleys discussed in Appendix 1 & 2 should be inspected and monitored to assess the risk of landslide dam formation and further risk of failure during the monsoon. At present, there does not appear to be a high risk of imminent failure in any of the dams that may severely impact valleys downstream, but this may change as the monsoon develops.
2. Road access to, and through, the zone of intense landsliding is going to be very challenging at least through the 2015 monsoon season. It would be sensible to undertake a survey of the major transportation corridors to identify where the problems are most likely to arise, and to prepare for the possibility of BOTH immediate action to clear rockfall debris AND continued clearance and maintenance of major roads after future landsliding. Note that all major roads to China in the affected area appear heavily impacted.
3. Collation of observations of the distribution, character and impact of landslides from people in the field is required to refine landslide hazard and risk assessments and validate the findings of the satellite-based survey we have undertaken. We have already instigated a social-media call to initiate this, but additional measures may be required, and a method of collation agreed.
4. Users of post-event map products should be aware that several factors, unrelated to the actual distribution of landslides, may have influenced their apparent distribution: i) large areas of Nepal (remain unsurveyed, largely because of cloud-cover and image availability (Fig. 1), so these areas must be assumed to have been affected; ii) some maps are predictions based on pre-earthquake landslide susceptibility evaluations rather, than observations post-event. Our maps only contain observations; and, iii) different methods have been utilised by respective agencies to map landslides (for instance, some may not have taken full account of the occurrence of pre-existing landslides in their assessment of landsliding associated with the event).

Vanessa Banks, Alex Densmore, Tom Dijkstra, Colm Jordan, David Milledge, John Rees, Nick Rosser, Jack Williams.

Appendix 1: List of confirmed or potential valley-blocking landslides

1. Therapati, in the south of the Langtang National Park, in the Melamchi catchment. A landslide deposit has been identified on the edge of a high-resolution optical DG image (03/05/2015), and shows surface features characteristic of a rock avalanche (> 1M m³), which appears to toe-out across the valley base, and likely causes a blockage. No evidence of a lake forming is apparent, but the valley at this time of year does not appear to have an active channel. The valley drains into the Trisuli. If this interpretation is correct, this blockage may be a concern during melt / rainfall.
2. Satellite images suggest multiple large valley blocking landslides occur in the main stem of the Trisuli, between Resuo (Nepal) and north to Jilong Community (Tibet). Some of these have been identified by ICIMOD. This area presents concern.
3. Video footage from the Upper Bhote Koshi suggests several reaches of rivers with abnormal standing water, notably immediately south of Chaku – most probably from extensive rockslides which appear to have failed, removing extensive sections of the Arniko Highway and making it impassable. Flow downstream from these sites continues, so these blockages are unlikely to be at risk of catastrophic failure.
4. Satellite images and news reports describe large post-seismic landslides in the 20 km between Chaku and the Tibet – Nepal border at Kodari. Prior to the earthquake this road section had numerous deep-seated landslides in colluvium which tend to be highly active during the wet season. These sites pose significant risk of valley blocking should these landslides move significantly. In addition, satellite imagery indicates extensive rockfalls and channelized failures on this stretch of road. Indications are that this area is still cut off from areas to the south. The Chinese army has cleared the road to Zhangmu (sometimes referred to as Zham) on the northern side of the border, but the communities of Larcha, Liping, Tatopani and Kodari appear still to remain isolated.
5. Satellite images north of the Tibet – Nepal border at Nyalam show extensive areas of landsliding and avalanching. Rockslides have in places crossed the road and have reached the river, but none have generated lakes that are visible in the presently available imagery in the river network.
6. Near Poki: a 1 km long rockslide (identified also by ICIMOD / BGS), appears to have blocked a valley, but has now been breached by the river. This area presents concern, but reports and imagery suggest that water is now flowing across this dam.
7. Langtang: extensive landslides, rock avalanches and possible glacier snout collapses appear to have traversed the valley floor, and are likely to inhibit flow during melt. The situation in Langtang is well-documented elsewhere.

Note that the 2014 Sun Koshi landslide which cuts the Arniko Highway just upstream of the Bhote Kosi Hydropower plant, and the rock dam that it created, appear to be intact. The river and lake levels do not appear to have changed during the earthquake.

Appendix 2: List of valleys with relatively intense landsliding (including affected Level 4 Admin / VDCs)

Tama Koshi, north of the Tama Kosi Hydro Power construction site:

• Lamabagar, Alambu, Worang, Chilangkha, Bigu, Khopachagu, Laduk, Gaurishankar, Khare, Lamidanda, Suri, Jhyanhu

Upper Bhote Kosi, north of the townof Barhabise:

• Phulpingkatti, Tatopani, Listikot, Dhuyang, Marming, Gati, Barhabise, Maneswar

Bhote Kosi, north of Jalbire:

• Pangtang, Gumba, Golche

Valleys north of Melanchi, feeding into the Indrawati:

• Thapalkot, Ghurisakot, Ghuyang, Bhotang, Baruwa, Helamby, Syaphru

Dhunche north to the China border in Rasuwa District:

• Ramche, Dhunche, Haku, Danada Gaun, Thuman, Timure, Thulo Gaun, Phikuri, Laharepauwa, Kaule, Bhalche, Karki Manakanama, Jibjibe, Briddim, Chililme, Gatland, Goljung, Syaphru, Bhorle, Yarsa, Langtang Valley