Afghanistan landslide of May 2, 2014

The Badakhshan landslide, showing its extreme mobility
Photo by Blal Sarwary as posted on Dave Petley's Landslide Blog
Links to sources are in adjacent text.

Our thoughts and condolences are with the survivors of the devastating landslide in Afghanistan.

On his Landslide Blog, David Petley has drawn from the Twitter feed of Bilal Sarwary, a BBC reporter on the spot where the landslide occurred in the Argo District of Badakhshan Province in Northeast Afghanistan was a failure in deposits of windblown sand (loess) after a period of heavy rainfall and flooding. With no hope of recovering the bodies of victims, Afghanistan has declared the site a mass grave and dedicated Sunday as a day of national morning.

While Petley conservatively says that "is now believed to have killed at least 350 people," the news media are reporting as many as 2000 deaths. Even the conservative value makes it the worst landslide to date in 2014, and the death toll is bound to increase as reports of missing people come in. There are about 4,000 survivors in need of care. The landslide is believed to have occurred in two phases, separated by enough time that people from adjacent villages had arrived to help with rescue efforts, only to be caught by the second landslide. Note that the Oso landslide in the U.S. two months ago also had two phases, but these were separated only by about 4 minutes.

Location of the landslide
From Cnn.com here

Petley based his conclusion that the landslide had occurred in loess on the high mobility of the landslide shown in the photo above, and the observation that there are no large boulders visible in the photo, an indication that the sediments were probably wind-laid. Landslides in loess are not unknown in this part of the world, though much more reported from China than Afghanistan. Loess covers huge areas of the world (see map), including some that are prone to large earthquakes (up to M8.5), such as this part of Asia. Loess covers 6.6% of the total area of China. In the middle part of the Yellow River in China, it can be 1000' thick. Chinese scientists have pioneered may of the studies of loess failure.

Map of location of loess deposits from here.

Loess was deposited during the Pleistocene by winds blowing across deposits of the receding glaciers. The climate was dry and, in many areas of the world, such as China and Afghanistan, has remained relatively arid since the loess was accumulated.  It is highly porous and has a weak cohesion, giving it rather unusual geological engineering properties.* It can contain up to 20% water in so-called "macropores", see photo. The mosaic structure of loess gives it strength, but the existence of macropores makes the strength vulnerable under seismic shaking or heavy rains. When loess fails due to heavy rains (or over-irrigation), it is because the pores fill with water, destroying the strength of the soils. However, even relatively dry loess has been known to fail.

Macropores in loess, Tianshui, Gansu Province
from the Zhang reference cited at *

There are a number of ways in which landslides can be triggered in loess areas. Zhou et al. (2002)** have classified loess landslides according to three "indicators": (1) the landslide material; (2) the position of the surface along which the slide moves; and (3) the mechanism of movement. Landslide materials can be either loess alone, or loess and underlying soil/rock, such as laterite, which is common in this part of the world. The development surface can be solely within the loess, or within the loess and underlying soil/rocks. The mechanism of movement can be slow gliding, or "collapsing-gliding." They point out that (in China, but I'm assuming that this applies to Afghanistan), landslides occur in (a) the rainy seasons of July, August and September, and (b) the ice melting period of March, April and May. This Afghanistan slide, if triggered by heavy rains, has occurred early compared to to the (Chinese) rainy season.

Water affects/causes landslides in three different ways: rainfall, surface water, and ground water. Both rainfall and surface runoff can increase the water content of a hillslope, and even saturate it. When saturated, e.g., below the water table, the cohesive force of the soil and the frictional force of layers within it can greatly decrease. Surface water also erodes the topography, particularly in tectonically active areas of the world. This leads to unstable, steep slopes prone to landsliding. Changes in groundwater affect the mechanical characteristics of soil. For example, when water content of soil increase up to 35%, shear-resistance decreases by 60% (Zhou et al., page 160). Ground water can also dissolve components of the soil, changing its composition and strength. 

Zhou et al** point out the potentially large influence of humans on inducing landslides. Remembering that this article was written in 2002, so that the statistics are not current, they report that in the 20 years preceding their report, population had reached 81.49 million, accounting for 7.8% of the nation's population. With this increase of population, "irrational human activities" such as excavating foothills, filling hillsides, and over-cultivating accelerated natural disasters. Because industry is relatively sparse in these loess areas, the output of industry and agriculture was below the national average, accounting for 6% of that of the nation. These areas tend to be poor, and frequent natural disasters "are major factors hindering the social and economic development of [the] loess area." On the borders of the loess plateau, flat ground is rare and slopes are universally leveled to build houses, and many people live in cave-houses cut into the slopes. All of these activities reduce the support at the bottom of the slope and decrease its stability. Vegetation, that normally stabilizes a slope, has been heavily impacted since the 1950's by deforestation. 

ASIDE: Although there appears not to have been an earthquake that triggered the Afghanistan landslide, it is worth keeping in mind that this huge region in Asia is subject to major earthquakes, which commonly trigger landslides if they are greater than M6, and wet weather.

COMMENT ADDED APRIL 6: John SanFilipo has commented that "There was a small (?) quake just west of the slide area near Rostaq reported on 12 April. Thanks, John! It will be interesting to see if anyone looks for a cause and effect.

On December 16, 1920, the M7.8 Haiyuan earthquake (also called the 1920 Gansu earthquake and estimated by some to have been as high as M8.7) killed at least 100,000 people, probably double this number according to the USGS. A Chinese paper in existence at the time Ningxia Daily, reported 240,000 killed. Strong trembling occurred for 10 minutes. Reportedly, 500,000 houses and cave dwellings collapsed.* The shaking occurred over an area of about 50,000 square kilometers, most of which is covered by loess. The landslides triggered by the earthquake blocked rivers and buried villages and farmlands.

There are several prominent consequences of earthquakes. (1) Fissures are produced both near the fault and far away from it. Near the fault, the fissures are primarily related to tectonic deformation and are spread on both sides of the fault. However, in zones of weak shaking far away from the fault (all the way through much of the loess region in China) fissures can be formed by subsidence in the loess caused by the loss of strength of the loess when it is shaken (liquefaction). (2) Landslides occurred in three types of areas in the Haiyuan quake: (1) near or in the high intensity shaking zone. However, because the soil cover was thin in this region, the landslides were quite small. (2) In an area of moderate shaking, more than 1000 sq. km. in area, there were 650 landslides reported (see the Zhang ref below), and 41 of them dammed rivers to form so-called "barrier" or "quake" lakes. These slides were big because the area was hilly and covered with loess ranging from 20-50 m (66-165') in thickness. (3) More distal regions where the damage caused by the landslides was more severe than that caused by the shaking of the quake itself. This area covered another 1000 sq. km.



*Zhang, Zhenzhong, Geological disasters in loess areas during the 1920 Haiyuan earthquake, China, GeoJournal, 36 (2/3) 269-274, 1995.

Zhou, Jin-xing, Zhu Chun-yun, Zheng Jing-ming, Wang Xiao-hui, Liu Zhou-hong, Landslide disaster in the loess area of China, Journal of Forestry Research 13(2), 157-161, 2002.