Data Collection
The data for this study comprises 17 years, monthly based data, and the data was collected through various online publishing sources. We created an updated catalog of the data for all the landslides in Pakistan from 2003–2019. The main sources for the data include NASA global landslide catalog, ArcGIS landslide database, Pamir Times, Dawn News, The News International, Reliefweb (NASA GLC, Pamir Times, Reliefweb, Dawn News, The News, ArcGIS Data). There were a total of 1089 landslides recorded in the 17 years period with 180 being fatal landslides which claimed lives of 1072 people (Note: Our study excludes the event of earthquakes in Oct 2005 due to unreliability of the regional casualties occurred exclusively due to landslides event in Northern Pakistan and Azad Jammu and Kashmir (AJK)).
This study focuses on the number of landslides (land movement, landmass fall, rockfall, rockslide, debris flow etc.), and their main triggers and their impacts (loss of life, injury, loss of property, vehicles, crops, business etc.) in the districts of all over Pakistan including the Gilgit-Baltistan region and AJK. There is also enough evidence (according to residents of the Northern region) that a high number of landslides and rockfalls are not even reported because of their small scale or remoteness of the regions and people are so much used to such events that the small-scale landslides or rockfalls are cleared from the roads and it does not get recorded. However, our understanding is that a full database can help scientists and researchers to study the landslide phenomena in a more effective way.
Methodology
We have analyzed the impacts of landslides in Pakistan by using the percentage analysis and we have used ArcGIS 10.5 for the maps to analyze the concentration/hot spots of the landslides in the districts included in our study (Fig. 5a-c). R-Studio is used to make the graphical assessment of the landslides data based on 204 weeks (Fig. 1).
Analysis and results
The data based on 17 years was segregated in two portions, 2003–2010 and 2011–2019, to examine the behaviors of the landslide triggers in spatio-temporal terms. Table 1 shows the frequency of the landslide events recorded yearly along with the casualties (dead and injured) and the frequencies of landslide triggers (rainfall, heavy/extreme rainfall, earthquakes, construction/mining and spontaneous).
Table 1
Number of landslides along with recorded casualties and their triggers 2003–2019
| | | | | | Triggers | | |
Year | Number of Landslides | Fatalities | Injuries | Rainfall | Heavy Rainfall | Earthquakes | Construction | Random |
2003 | 194 | 16 | 9 | 2 | 111 | 46 | 1 | 34 |
2004 | 5 | 25 | 6 | 1 | 0 | 3 | 0 | 1 |
2005 | 19 | 12 | 0 | 2 | 5 | 9 | 0 | 3 |
2006 | 54 | 85 | 40 | 23 | 22 | 0 | 1 | 8 |
2007 | 101 | 127 | 45 | 3 | 94 | 0 | 2 | 2 |
2008 | 23 | 44 | 14 | 7 | 10 | 3 | 0 | 3 |
2009 | 7 | 4 | 3 | 1 | 0 | 0 | 1 | 5 |
2010 | 50 | 135 | 126 | 9 | 14 | 0 | 1 | 26 |
2011 | 28 | 15 | 9 | 4 | 0 | 0 | 4 | 20 |
2012 | 22 | 68 | 25 | 3 | 13 | 0 | 2 | 4 |
2013 | 23 | 8 | 3 | 0 | 11 | 0 | 1 | 11 |
2014 | 29 | 35 | 37 | 6 | 6 | 0 | 1 | 16 |
2015 | 85 | 49 | 14 | 11 | 14 | 50 | 2 | 8 |
2016 | 270 | 229 | 41 | 30 | 208 | 0 | 1 | 31 |
2017 | 59 | 80 | 45 | 15 | 7 | 0 | 10 | 27 |
2018 | 42 | 52 | 40 | 7 | 2 | 0 | 11 | 22 |
2019 | 78 | 88 | 48 | 12 | 36 | 0 | 8 | 22 |
Most landslides were recorded in 2016 with the highest number of fatalities that account for 21.4% of all the fatalities in our study period. Rainfall related events were the major cause of deaths and it accounted for 41.67% of the total deaths caused by the landslides in our 17 years period. Interestingly, 29.44% of the fatal landslides were triggered spontaneously (which could be attributed toward unstable rocks, deforestation etc.). The rainfall was responsible only for 32.23% of fatal landslides from 2011–2019 as compared to 61% from 2003–2010. The majority of fatal landslides in the last decade were caused by construction events (road construction, digging, mining along steep hills and mountains) and spontaneous events that account for 66.1% of the fatalities from 2011–2019.
Karakoram Highway (KKH) is the main link that joins all the regions of Gilgit-Baltistan with the rest of Pakistan and it is especially important because most of the trade between China and Pakistan is happening from this land route. Billions of dollars of goods and services pass through KKH every year and it is highly vulnerable to landslides. The highway is a hot spot for the landslides and rockfall due to several reasons including continuous expansion of KKH. Most of the landslides are recorded in the rainfall seasons (monsoon) which causes movement of the lose rocks and debris on the slopes. There is only a small portion where the highway is being protected by a cemented wall of rock, barely few feet high, and in practice this is only effective in cases of small rockfalls but falling of heavy boulders are a common occurrence.
Pakistan has 154 districts and landslides in the last 17 years are reported from 48 districts, among which 28 districts only recorded one or two landslides. The remaining 20 districts recorded larger numbers of landslide events. Baltistan and Diamer districts were responsible for 39.58% of all the landslides in our study period. The months of April through June is the time when such events are occurring most often and it has triggered 50.5% of all the landslides in our study duration. However, it was not the deadliest since it accounted for 25.54% of the deaths, whereas the first quarter of the year is responsible for 30.54% of all the deaths in 17 years. In our study period October through December accounts for the least number of landslides events with just 11.11% of the total landslide occurrences.
The Wilcoxon-Mann-Whitney test was performed on the fatal landslides of the two data sets and it found that there is significant increase in the occurrences of fatal landslides in the period 2011–2019 compared to 2003–2010. The magnitude of the recorded number of occurrences of the landslides in all the districts of Pakistan shows that there is large clustering in the Northern regions that include Gilgit-Baltistan and regions of AJK most notably in Muzaffarabad and Neelam valley (Fig. 2b).
The gender proportion that had sustained casualties was challenging because of the unavailability of credible information. In our study period from 2003–2010, only 32% of the gender of fatalities could be verified and based on these statistics 49% of the fatalities were men and 51% of fatalities were from women while 19.44% were children. During the period from 2011–2019 the fatalities among men were more than 78% while 22% of deceased were women, these statistics are based on 50% of the known identity of the gender in the fatal landslides during this period. In the second part of our study period 39.87% of fatalities occurred due to construction (labor, mining, expansion etc.) whereas 19.62% were children among the fatalities from 2011–2019 (Fig. 3).
There is a large difference between the two periods of our study while studying the gender proportions mainly because in the later part of study where construction and spontaneous landslide events were involved, mostly men are doing the jobs of labor on the roads sides and driving the vehicles to transport goods, and who were hit with spontaneous boulders and landslides.
The landslides occurrences according to the monthly basis was analyzed and it showed that April is the month with the greatest number of landslides occurrences and it also was responsible for the largest number of fatal landslides, 32 fatal landslides, which claimed lives of over 250 people (Fig. 4). One similarity that can be found is that large numbers of landslides are occurring during the second quarter of the year or at the start of summer season when snow melts and hence the vulnerable slopes give weigh and lose rocks also get help from precipitation or heavy precipitation events, for example, cloud bursts or typhoons (Fig. 4).
The damages were also accessed that can be attributed to landslides (both directly and indirectly). In the time period from 2003–2010, 576 homes were damaged or destroyed and at least 20 businesses were destroyed due to landslide events. The agricultural land effects were also considered and 10 separate events of crop destruction were recorded and at least 10 vehicles were damaged or destroyed in such landslide events. As many as 234 roadblocks were recorded and 13 roads (roads and small bridges) were destroyed. In the second period, 2011–2019, 489 homes were damaged or destroyed, 87 businesses were lost, at least 5 events of crop damages or destruction were observed, 34 vehicles were destroyed, 745 roads (main roads and link roads) were blocked and 14 roads were destroyed in landslide related events. These damages are estimated to be in the multiple of millions of dollars and in a country that is not renowned for its economic superiority, landslides are indeed costly disasters (Note that these estimates also include the damages caused by the 2010 Attabad landslide which destroyed parts of KKH and new routes had to be carved inside the mountains).
Further analysis of the two timelines, 2003–2010 and 2011–2019, reveals that the prior time period is responsible for fewer number of fatal landslides, primarily because landslides were less frequent. In that same period, we see that rainfall-triggered fatal landslides were responsible for over 61% of the fatal landslides, whereas in the second period, the fatal landslides triggered by rainfalls accounted for only over 32%, which is a sharp decrease. However, the main triggers in the second period were construction and spontaneous events, which collectively saw an increase of 33.91% compared to the first time period. In Pakistan, earthquakes were involved in 3.33% of the fatal landslides (excluding 8 Oct 2005 earthquake event) and in the first time period they caused fatal landslides which accounted for 6.78% whereas in the second period it was only 1.65%. Furthermore, by analyzing the monthly data, December is the month that records the least number of landslides, but the least proportion of fatal landslides was observed in the month of June that had a share of just 3.3%.
While it has been observed that in the time period from 2011–2019 the main causes of fatal landslides were spontaneous and due to construction or mining along the slopes, it can not be overruled that rainfalls were not significant. As many as 60.53% of all the landslides in this period were caused by rainfall (rainfall and heavy rainfall) events, which is a significant percentage. The reduction in the fatal landslides in a rainy season might be due to an increasing awareness in the mountainous communities, but on the other hand spontaneous landslide events are even more alarming and need to be studied with geophysical tools.
Table 2
Distribution of Landslides and Fatal Landslides according to Months (Theses statistics are based on 204 months of data)
Month | Landslides | % Landslides | Fatal Landslides | Fatalities |
JAN | 47 | 4.31% | 16 | 89 |
FEB | 55 | 5.05% | 20 | 95 |
MAR | 48 | 4.41% | 19 | 101 |
APR | 291 | 26.72% | 32 | 252 |
MAY | 204 | 18.73% | 8 | 35 |
JUN | 55 | 5.05% | 6 | 26 |
JUL | 154 | 14.14% | 21 | 150 |
AUG | 58 | 5.33% | 14 | 132 |
SEP | 56 | 5.14% | 15 | 51 |
OCT | 81 | 7.44% | 13 | 71 |
NOV | 25 | 2.29% | 8 | 44 |
DEC | 15 | 1.38% | 8 | 26 |
Pakistan is a very active seismic region and has seen multiple high magnitude earthquakes in the near past, 2005, 2010, 2013, 2015 and 2017 etc. However, some scientific studies claim that the earthquake in 2005 could have killed more than 25000 people in Azad Jammu and Kashmir (AJK) which generated as many as 2500 landslides (Dunning et al., 2007; Mahmood et al., 2015), but it could not be verified which of the specific regions had recoded what number of landslides (fatal and non-fatal) and casualties (Note: The only landslides that were verifiable with date, time, damages and, more importantly, geolocation, were included in our study). Other landslide events caused by earthquakes within the 17 years of study timeline accounted for 10.19%, however, they were less deadly compared to other triggers in our study. In the period from 2003–2010, deadly landslides due to earthquakes had a proportion of 6.77% and in the last period of our study it was only 1.65%.
The increase in landslide events in Pakistan is related to extreme weather phenomena as the effects of climate change are already affecting the weather cycles and Pakistan is among the countries most effected by the climate change, it suffered economic losses worth 3.8b$ from 1999–2018 (Deccan-Herald 2019, Eckstein et al. 2019). Our study finds that rainfall and heavy rainfall events, which have increased in the recent years, are the major cause of trigger for the fatal landslides in Pakistan and they are also the main factor in deadly landslides in a study conducted for 128 countries worldwide (Haque et al. 2016, 2019). However, our study also finds that the second main trigger for the fatal landslides in Pakistan is the occurrence of spontaneous landslide events, which is of greater concern. The landslide hot spots are mapped according to the districts of Pakistan (Fig. 5a-c).
Figure 5a-c Landslide Hotspots in Districts of Pakistan (2003–2019) about to place here.