Spatio-temporal evolution of post-seismic landslides and debris flows: 2017 M_s 7.0 Jiuzhaigou earthquake

Abstract

Spatio-temporal evolution of post-seismic landslides and debris flows provides a new perspective to understand post-earthquake evolution of geological environments and landscapes, and to instruct cascaded catastrophic hazard mitigation and post-disaster reconstruction. However, limited earthquake events have been investigated for post-earthquake geohazard evolution. This work reports the geohazard evolution after the 2017 M_s 7.0 Jiuzhaigou earthquake considering the effects of the earthquake, geology, terrain, meteorology, hydrology, and human engineering activity. Some new viewpoints are suggested. (1) Landslide and debris flow activity intensified in the first year following the earthquake under the effects of the antecedent earthquake, precipitation, fault tectonics, human engineering activity, and fluvial networks. (2) Landslide and debris flow activity declined rapidly in the second year as a result of dramatically reduced sediments, declined rainfall, and self-healed slopes. (3) The significant decay of landslide and debris flow activity and the prominent reduction of loose deposits indicate that the geological environment was gradually restoring. (4) Although the hazard effect mitigation and geological environment restoration were ongoing (in the absence of rainstorm events) to attain a new balance, the geoenvironment has not returned to the pre-earthquake level because of widespread unrecovered geohazards and the remaining loose deposits on hillslopes or in channels. (5) The geological environment after the Jiuzhaigou earthquake may re-equilibrate and return to the pre-earthquake level more quickly than after the Kashmir, Chi-Chi, Gorkha, Wenchuan, and Murchison earthquakes. This work provides new knowledge pertaining to geohazard evolution after a strong earthquake and to profound impacts of a catastrophic earthquake on geological environment and landscape.

Introduction

Post-seismic landslides and debris flows (PSLDs) are characterized by large quantities, large scales, broad and concentrated distribution, great damage, and cascaded high risks (Xu 2018), and have received close attention from scientists (Keefer 1984; Zhang et al. 2014; Fan et al. 2019; Papathanassiou et al. 2021; Comert 2021; Porta et al. 2021; Koukouvelas et al. 2020; Litoseliti et al. 2020). PSLDs generate widespread mass wasting (Avouac 2007) and play a significant erosional and depositional role in relief reduction (Hovius et al. 2011) during the years to decades following earthquakes (Nakamura et al. 2000; Lin et al. 2006; Lin et al. 2008; Li et al. 2018). Thus, they exert a far-reaching influence on the long-term evolution of geological environment and landscape (Malamud et al. 2004; Fan et al. 2018; Broeckx et al. 2020). Moreover, the sudden and collective outbursts of PSLDs are conspicuously enhanced because of the relaxation and destruction of mountain rock mass caused by earthquakes (Huang 2011). These landslide and debris flow activities and impacts may extend several years (for example, the 2005 M_w 7.6 Kashmir earthquake) (Saba et al. 2010) or even over 50 years (like the 1929 M_w 7.7 Murchison earthquake) (Pearce and Watson 1986) before the impacted regions revert to stability. In addition, the aggravation of landslide and debris flow activity after earthquakes poses long-term dangers to human lives and critical infrastructures in large-scale earthquake-struck areas. This has always been an enormous challenge for long-term prevention and cure of geological disasters. Therefore, it is crucial to reveal the spatio-temporal evolution of PSLDs to better understand the geological environment and landscape development after earthquakes, to effectively mitigate the cascaded high risk of PSLDs, and to rationally instruct post-disaster reconstruction (Huang 2011). However, thus far, limited earthquake events have been investigated in terms of landslide and debris flow evolution after these catastrophic earthquakes, such as the 1929 M_w 7.7 Murchison earthquake, 1993 M_w 6.9 Finisterre earthquake, 1999 M_w 7.6 Chi-Chi earthquake, 2004 M_w 6.6 Niigata earthquake, 2005 M_w 7.6 Kashmir earthquake, 2008 M_w 7.9 Wenchuan earthquake, 2008 M_w 6.8 Iwate earthquake, 2008 M_w 6.4 Movri Mountain earthquake, and 2015 M_w 7.8 Gorkha earthquake (e.g., Pearce and Watson 1986; Hovius et al. 2011; Khan et al. 2013; Marc et al. 2015; Chang et al. 2015; Li et al. 2018; Yang et al. 2018a; Shafique 2020; Tian et al. 2020; Litoseliti et al. 2020; Basharat et al. 2021).

The evolution and restoration period of PSLDs varies distinctly in different earthquake-affected regions and are highly dependent on earthquake magnitude, geology, topography, geomorphology, material composition, and climate (Huang 2011; Zhang and Zhang 2017; Hu et al., 2019; Tian et al. 2020). In some earthquake-struck regions, the PSLDs were remarkably strengthened and highly activated for decades (Huang 2011). For example, the landslides caused by the 1923 M_w 7.9 Kanto earthquake were highly active for approximately 15 years and returned to stability after 40–50 years (Nakamura et al. 2000). The landslides induced by the 1929 M_w 7.7 Murchison earthquake supplied ~400,000-m³/km² stream channel sediments and at least 50–75% of the granitic debris was still trapped in the fourth-order catchment 50 years after the earthquake (Pearce and Watson 1986). After 7–8 years of evolution after the 2008 M_w 7.9 Wenchuan earthquake, the debris flows and landslides still remained active at higher levels than the pre-earthquake levels (Tang et al. 2016; Zhang and Zhang 2017; Fan et al. 2018a, 2018b; Liu et al. 2018a, 2018b) and would continue to remain active for a long time due to massive loose deposits and seasonal rainfall (Cui et al. 2011; Zhang et al. 2016; Zhang and Zhang 2017; Li et al. 2018; Yang et al. 2018a). However, in some earthquake-impacted areas, the activity of PSLDs rapidly decayed to pre-earthquake levels within years. For example, the enhanced landslide activity following the 2005 M_w 7.6 Kashmir earthquake remained at a high level for no more than 2 years and progressively approached a new stable condition within 2–5 years after the earthquake, with most slopes returning to equilibrium (Saba et al. 2010; Khattak et al. 2010; Khan et al. 2013; Shafique 2020). The density of rainfall-triggered landslides obviously increased 1–2 years after the 1999 M_w 7.6 Chi-Chi earthquake (Lin et al. 2006), and the mass wasting and fluvial sediment evacuation ever reached more than 5 times the background rate and gradually recovered to pre-earthquake levels after approximately 6 years (Hovius et al. 2011). Landslide rates after three shallow thrust earthquakes, i.e., the 1993 M_w 6.9 Finisterre earthquake, 2004 M_w 6.6 Niigata earthquake, and 2008 M_w 6.8 Iwate earthquake, peaked and then decreased to background values respectively within 1.5–2.7 years, 0.5–0.9 years, and 1.0–2.4 years, with the decay time scale possibly associated with the earthquake magnitude and decreased transient ground strength (Marc et al. 2015). Therefore, the evolutionary characteristics and period of PSLDs vary significantly in diverse earthquake-attacked regions, even under approximate seismic magnitudes.

The 2017 M_s 7.0 Jiuzhaigou earthquake occurred in the middle of the famous north-south seismic belt in China where earthquakes have intensively happened (Li et al. 2017). The earthquake occurred at the intersection of the Tazang, Huya, and Minjiang faults on the eastern margin of the Bayan Har block in the Tibet Plateau (Xu et al. 2017). Historically, 26 earthquakes above M_s 5.0 and 9 earthquakes above M_s 7.0 struck the area within 200 km from the Jiuzhaigou epicenter (The Earthquake Disaster Prevention Department of China Earthquake Administration 1995, 1999; Lei et al. 2018; Fang et al. 2018). The earthquake-attacked areas are characterized by extremely complicated geological and topographic conditions, such as developed folds and fractures, intensive neotectonic movements, widespread carbonates, large-scale karst tufa deposits, uplifted crust, and large terrain transformation from the Tibet Plateau to the Sichuan Basin (Lei et al. 2018). Therefore, research on the evolution of landslides and debris flows after the Jiuzhaigou earthquake has significant geological implications in revealing the PSLD development rule in an ultracomplex geological and topographic context. Moreover, the Jiuzhaigou Valley within the earthquake-influenced region, a world-famous scenic spot, has been selected as a UNESCO World Natural Heritage Site and a World Biosphere Reserve (UNESCO 1992; Fan et al. 2018). In 2016, the annual tourist number surpassed 5 million, accompanied by tourism revenue of CNY 805 million (SBNESD, 2016). In this world-popular tourist attraction, some scenic spots reopened on September 27, 2019. Thus, an insight into the PSLD evolution has important social and economic merits in safeguarding tourists, villages, roads, and human infrastructure with frequent human activities. Therefore, because of the unique and critical significance of geology, society, and economy, the evolution of landslides and debris flows after the Jiuzhaigou earthquake is studied to improve the understanding of PSLD development considering the effects of earthquake, geology, terrain, hydrology, meteorology, and human engineering activity.

Research on Jiuzhaigou earthquake-triggered hazards has covered four main aspects. (1) The spatial distribution patterns of coseismic landslides and debris flows were investigated mainly via remote sensing technique (e.g., Dai et al. 2017; Xu et al., 2018; Fan et al. 2018; Wu et al., 2018; Cheng et al. 2018; Tian et al. 2019; Dong et al. 2019; Chang et al. 2021). Some studies suggested that the spatial pattern of Jiuzhaigou coseismic landslides was substantially affected by topographic amplification of seismic waves, including back slope and thin ridge effects (Zhao et al. 2018; Wu et al., 2018; Wang et al. 2018; Shen et al. 2019; Liang et al. 2019). (2) The susceptibility of coseismic landslides was evaluated using remote sensing, seismic, geologic, topographic, and coseismic landslide data (Dai et al. 2017; Fan et al. 2018; Cao et al. 2019; Yi et al. 2019; Cao et al. 2020; Chen et al. 2020). Moreover, risk identification and evaluation were conducted to investigate the impact of post-earthquake landslides or rockfalls on roads (Yue et al. 2018; Li et al. 2019). (3) The variation trend and the potential danger of PSLDs were evaluated according to the damage from the coseismic geological hazards (Chen et al. 2018; Hu et al. 2019a). Chen et al. (2018) delineated the coseismic geohazards of 13 landslides, 70 collapses, and 25 potential debris flow gullies in the Jiuzhaigou Valley. Based on the rain intensity cycle in the Jiuzhaigou Valley and the geohazard variation tendencies triggered by the Kanto, Chi-Chi, and Wenchuan earthquakes, it was predicted that Jiuzhaigou PSLD activity may last over 10 years and would progressively attenuate back to pre-earthquake levels in 5 to 10 years (Chen et al., 2018). Hu et al. (2019a) highlighted 83 coseismic landslides in the Jiuzhaigou Valley and suggested that the PSLD activity recovery period in Jiuzhaigou would be shorter than that in Chi-Chi and Wenchuan because of the small hazard magnitude and better geological condition. (4) The serious destruction caused by coseismic geohazards to natural landscapes was quantitatively evaluated in the Jiuzhaigou Valley (Hu et al. 2019b).

A considerable amount of new knowledge has been acquired on the coseismic landslides and debris flows in the Jiuzhaigou area; however, the spatio-temporal evolution of PSLDs has not been specifically revealed. This work employs multisource data (seismic, geologic, topographic, meteorological, and remote sensing data) and field investigation to reveal the spatio-temporal development of PSLDs from August 2017 to August 2019 under the effects of earthquake, geology, terrain, rainfall, drainage system, and human engineering activity. New insights into the following questions are suggested: (1) How did the landslides and debris flows evolve after the earthquake? (2) Which influencing factors controlled or induced these evolutions? (3) What are the spatio-temporal evolution rule and trend of the PSLDs? Answers to these questions can improve our limited understanding of the geohazard evolution after a violent earthquake and supplement new knowledge on the profound impact of catastrophic earthquakes on landscape and geological environment.

Study area

On August 8th, 2017, a strong M_s 7.0 earthquake attacked Jiuzhaigou County in China with an epicenter at 33.20° N, 103.82° E and a focal depth of 20 km (CENC 2017). This earthquake caused a direct economic loss of up to CNY 22.45 billion and resulted in 25 deaths, 525 casualties, 6 missing people, 176,492 vulnerable people, and 73,671 damaged buildings by 8:00 pm August 13, 2017 (Fan et al. 2018; Xu et al., 2018; Hu et al., 2019). The intensive distribution region of the earthquake-induced landslides and debris flows is selected as the study area (Figure 1) to reveal the spatio-temporal evolution feature and rule of the PSLDs.

Figure 1

Seismotectonic setting of Jiuzhaigou County and the study area. a Jiuzhaigou earthquake and the complicated seismotectonic setting. b Study area in Jiuzhaigou County. The seismic intensity and the acquisition times and scopes of the high-resolution remote sensing images are indicated. c Seismotectonic setting of the study area and the distributions of roads and river systems. The inferred seismogenic fault is from Li et al. (2017). The historical earthquake data are from the United States Geological Survey, and the fault data are from National Geological Archive. The Earth picture is from Google Earth. The administrative boundary of Jiuzhaigou County is from Geographical Information Monitoring Cloud Platform.

The study area, covering a region of 938.9 km², is situated at the northern edge of the Minshan Block bordering the Longmenshan Range, Huya, Tazang, and Minjiang fractures in the south, east, north, and west, respectively (Zhao et al. 2018). A critical seismic belt, with multiple strong earthquakes (M_s≥7.0), spreads along the block boundaries making it one of the most active tectonic units in China (Sun et al. 2018; Zhao et al. 2018). Thus, the active fault structures are extremely intricate under the strong compression of shear stress in the study area (Xu et al. 2017; Li et al. 2017). The blind seismogenic fault, striking ~152° and dipping ~70° across the study area, is dominated by a NW-SE-oriented left-lateral strike-slip movement (CENC 2017; Li et al. 2017; Yi et al., 2017) and is attributed to an eastern branch fracture in the East Kunlun fault zone (Yi et al., 2017). The study area stretches across the regions of seismic intensities of IX, VIII, and VII, with the main shock peak ground acceleration (PGA) varying from 0.12 to 0.26 g. The neotectonic movement is intense and forms NE-oriented and NS-oriented tectonic systems composed of anticlinoria and synclinoria (Cheng et al. 2018). Limestone prevails regionally, accompanied by exposed Devonian, Carboniferous, Permian, Triassic, Tertiary, and Quaternary system strata (Liu G. et al., 2018).

The study area, abundant in deep-incised ravines and gullies, features alpine canyon landforms that were shaped and eroded by tectonic, glacial, and hydrological activities (Fan et al. 2018). The terrain elevation gradually varies downward from southwest to northeast with the highest elevation of 4570 m and an average slope angle of 31°. The high mountains, steep terrain, abundant precipitous surfaces, and developed river systems favor landslide and debris flow development (Liu et al., 2018). A mass of loose deposits and sediments, generated from the strong earthquake and secondary hazards, supply rich source materials for the post-seismic new landslides and debris flows. The annual rainfall (2017–2019) was relatively low in the study area (~677–~730 mm) mainly concentrated from May to October. Moreover, road construction and tourism development may play important anthropic roles in landslide and debris flow occurrences in the study area (Wang et al. 2018; Cao et al. 2019).

Data and methods

Multisource data

Five sets of multisource data are employed to investigate geohazards and mass wasting and to establish various influencing factors closely related to landslide and debris flow evolution (Table 1). (1) Multi-temporal high-resolution remote sensing images (Gaofen-1, Gaofen-2, Planet, Ziyuan-3, and Google images) are used to delineate the geohazards, hillslope deposits, and channel deposits generated during and after the earthquake. Thus, geohazard evolution and mass wasting can be revealed. Additionally, multi-temporal Sentinel-2 images, combined with the other high-resolution images, are adopted to establish the critical influencing factors of human engineering activity, environment, and hydrology in 2018 and 2019, including land use, distance to road, normalized difference vegetation index (NDVI), and distance to the river. (2) Seismic data are utilized to construct the seismic factors of PGA, seismic intensity, distance to epicenter, and distance to seismogenic fault. (3) Geological maps are used to build the geological and tectonic factors of lithology, fault density, and distance to the fault. (4) Digital elevation model (DEM) data are adopted to produce the topographic factors of elevation, slope angle, and aspect. (5) Meteorological data are employed to establish the precipitation factors of cumulative rainfall, maximum cumulative rainfall, minimum cumulative rainfall, and maximum cumulative rainfall in continuous precipitation events in 2017, 2018, and 2019. Therefore, by synthesis of geohazard development, mass wasting, and various influencing factors, the spatio-temporal evolution characteristics, rule, and trend of PSLDs can be revealed.

Table 1 Multisource data for coseismic and post-seismic landslides and debris flows

Methods

Establishment of geo-environmental factors and triggering factors

Two types of influential factors closely relevant to geohazard evolution are established: geo-environmental factors controlling hazard evolution and triggering factors inducing disaster development (Table 2). The geo-environmental factors comprise seismic, geological, tectonic, topographic, and environmental factors, in which environmental factors are embodied in vegetation growth and river system and comprise the indices of NDVI and distance to river. The triggering factors include precipitation and human engineering activity factors, in which human engineering activity factors are reflected by land utilization and road construction and contain the indices of land use and distance to the road.

Table 2 Geo-environmental and inducing factors associated with landslide and debris flow evolution in the study area

Classification of geohazard evolution

Post-earthquake vegetation growth and recovery illuminate the decay and restoration of PSLDs and highlight the stability tendency of the landscape and geological environment (Yang et al. 2018a; Li et al. 2018; Fan et al. 2019; Yunus et al. 2020). Vegetation regeneration, as an indicator of geohazard decay, has been applied in geohazard restoration investigations after the Wenchuan (Lu et al. 2012; Li et al. 2018; Domènech et al. 2019), Chi-Chi (Shou et al., 2011), and Kashmir earthquakes (Shafique et al., 2020; Basharat et al. 2021). The landslide (rockfall) evolution in the study area is divided into five classes (with reference to Li et al. 2018): (1) recovered or basically recovered landslides, (2) recovering landslides, (3) active landslides, (4) unchanged landslides, and (5) new landslides. Recovered or basically recovered landslides are those that are completely covered by thick vegetation. Recovering landslides mean those that are partly covered by vegetation without area expansion. Active landslides are those with increasing areas. Unchanged landslides signify the ones maintaining the same area with no vegetation growth. New landslides mean newly occurring landslides.

The debris flow evolution in the study area falls into four categories (with reference to Li et al. 2018): (1) active debris flows, (2) new debris flows, (3) recovered debris flows, and (4) unvaried debris flows. Active debris flows occur multiple times or undergo increased areas and abundant/increased source materials. New debris flows are the newly emerging ones generated from landslide deposits and/or from channel-bed failure, which arises from bed incision, lateral bank failure, bed erosion, and rainfall (Zhang and Zhang 2017). Recovered debris flows mean debris flow gullies where the vegetation is growing well. Unvaried debris flows are debris flow gullies unaccompanied by enlarged areas and vegetation growth.

The volume of coseismic geohazards is calculated according to Equation (1) (Guzzetti et al., 2009), and this equation was established according to 677 landslides from a global landslide database (Guzzetti et al., 2009). The relationship between geohazard area and volume is generally independent of the physiographic setting (Guzzetti et al., 2009).

$$ V=0.074\times {A}^{1.450} $$

(1)

where V and A are the volume and area of a geohazard, respectively.

Results

Landslide and debris flow investigation

The geohazards in the study area mainly consist of landslides (rockfalls) and debris flows, which are interpreted from multi-temporal high-resolution remote sensing images from October 2015 to September 2019 and validated by field investigation. The distributions and statistics of coseismic geohazards and PSLDs in both Jiuzhaigou County and the study area are shown in Table 3 and Figs. 2 and 3, respectively. As shown, the earthquake-triggered geohazards in Jiuzhaigou County were intensively distributed in the study area. The field survey route is shown in Figure S1 in the supplementary file, and 257 geohazards were validated, including 214 landslides and 43 debris flows. Moreover, rainfall had an important influence on coseismic geohazards and PSLDs, and the pre-earthquake and post-earthquake rainfall statistics in the study area are shown in Figure 4.

Table 3 Statistics of the coseismic landslides and debris flows in Jiuzhaigou County and in the study area

Figure 2

Geohazard distribution maps in Jiuzhaigou County and in the study area. a Distribution of the coseismic landslides and debris flows in 2017. b Distribution of the newly occurring landslides and debris flows in 2018. c Distribution of the newly emerging landslides and debris flows in 2019. CLOS, coseismic landslides outside of the study area; CDOS, coseismic debris flows outside of the study area

Figure 3

Geohazard statistics in the study area from 2017 to 2019. The geohazards in 2017 refer to the coseismic hazards, and the disasters in 2018 and 2019 indicate the newly occurring ones

Figure 4

Rainfall statistics from 2017 to 2019 in the study area. The cumulative rainfall in 2017, 2018, and 2019 signifies the accumulative rainfall from August 2016 to July 2017, from August 2017 to July 2018, and from August 2018 to July 2019, respectively. Each continuous precipitation event corresponds to a cumulative rainfall value, and the abbreviation of MCR implies the maximum cumulative rainfall among all the continuous precipitation events

The Jiuzhaigou earthquake triggered 5563 coseismic geohazards in the study area covering 10.83 km² and generated two hazard-dense regions: region A to the northwest of the epicenter covering 68.24 km² with an epicenter distance of 8.45 km and region B to the southeast of the epicenter covering 48.57 km² with an epicenter distance of 5.07 km (Figure 2a). These coseismic geohazards consisted of 5431 landslides with a total volume of 34.76×10⁶m³ and 132 debris flows with a gross volume of 7.85×10⁶m³. These coseismic geohazards were dominated by small sizes, falling in a region of 938.9 km² and generally spread along the seismogenic fault. Interestingly, 90.87% of the geohazards were located in the area with relatively abundant anterior rainfall, that is, with the accumulated rainfall larger than 590 mm from August 2016 to July 2017 (Figure 2a). The pre-earthquake rainfall had infiltrated into the rock and soil, sharply increasing pore pressures, decreasing shear strength, and facilitating the slope materials to slide and run out under the strong earthquake (Sassa et al. 2007; Faris and Wang 2014; Fan et al. 2019). Therefore, except for the triggering factor of earthquake, antecedent precipitation also played an important causative role in the Jiuzhaigou coseismic landslides and debris flows.

Within 1 year after the earthquake, 951 geohazards had newly occurred in the study area with a gross volume of 23.05×10⁶m³, consisting of 756 landslides over 6.4×10⁶m³ and 195 debris flows over 16.65×10⁶m³. The new landslides were dominantly (98.81%) characterized by small sizes and only occupied 27.77% of the total volume of the new PSLDs, whereas the dramatically increased debris flows accounted for 72.23% of the total volume. Coseismic geohazards provided ample source materials, and elevated rainfall was the leading trigger for the new hazards in 2018; thus, the new PSLDs prevailed in the regions with intensive coseismic hazards and sufficient rainfall. The region enriched in rainfall, with accumulation rainfall larger than 840 mm from August 2017 to July 2018, was typical of 94.05% of new landslides and 95.38% of new debris flows (Figure 2b). Therefore, significantly increased debris flow activities, associated with the enhanced rainfall, in the coseismic-hazard-concentrated area were the distinctive PSLD characteristic within 1 year after the earthquake.

During the second year after the earthquake, the number and area of new landslides and debris flows both significantly decreased. Only 31 new geohazards emerged in the study area covering 2.41 × 10⁶m³, composed of 16 landslides covering 0.61×10⁶m³ and 15 debris flows covering 1.80×10⁶m³, concentrated around roads or in bare lands (Figure 2c). Debris flow activities were still a dominant phenomenon, accounting for 74.69% of the total volume of the new PSLDs. Because the average cumulative rainfall (610 mm) during the second year was lower than the average (851 mm) during the first year, the significant mitigation of hazard activities highlights the evolution trend to a new balance in the absence of rainstorm events.

Furthermore, loose deposits generated by the coseismic and post-seismic landslides and debris flows are illuminated in distribution, transition, and mass wasting in the supplementary file. Both field survey and remote sensing technique show that loose deposits significantly reduced within 2 years after the earthquake. The sustained consumption of loose deposits and the prominent reduction in new sediments indicate that the geological environment was gradually restoring. (Refer to the supplementary file for the detailed loose deposit investigation and mass wasting rule.)

Evolution of post-seismic landslides and debris flows

The developments of landslides and debris flows after the earthquake are shown in Figures 5, 6, and 7 and Figure S8 in the supplementary file. During the first post-seismic year, 50.87% of the 5431 coseismic landslides in the study area remained active, accompanied by mass wasting and new loose deposits, with the volume increasing from 16.51×10⁶m³ in 2017 to 21.31×10⁶m³ in 2018. 47.76% of the coseismic landslides remained unchanged and unvegetated covering 13.21×10⁶m³, 1.34% were undergoing recovery and were partly vegetated, and only 1 coseismic landslide was completely vegetated. The average cumulative rainfall across the study area rose from 596 mm before the earthquake to 851 mm in 2018; thus, mainly under the actions of earthquake-loosened slopes and enhanced rainfall, 756 landslides newly appeared covering 6.40×10⁶m³.

Figure 5

a–d Distribution maps of various geohazard (landslide and debris flow) evolution classes in 2018 and 2019

Figure 6

Evolution characteristics of some representative landslides and debris flow. a Evolution histograms. b Evolution images of four selected geohazards

Figure 7

Statistics of various evolution classes in 2018 and 2019. a Statistics of landslide evolution in number and number proportion. b Statistics of landslide evolution in volume and volume proportion. c Statistics of debris flow development in number and number proportion. d Statistics of debris flow development in volume and volume proportion

Over the first year, among the 132 coseismic debris flows, 71.97% dynamically progressed or resurged, together with additional source materials, and extended from a volume of 4.44×10⁶m³ in 2017 to 6.58×10⁶m³ in 2018. 28.03% stayed inactive and unvegetated covering 2.75×10⁶m³, and no coseismic debris flow recovered and experienced good vegetation growth. The debris flow activity was significantly aggravated within the first year, and 195 new events occurred, accounting for 59.63% of the post-seismic debris flows (PSDFs).

Within the second year, the number of active landslides dramatically decreased from 2763 in 2018 to 45 in 2019, occupying only 0.73% of the post-seismic landslides (PSLs), covering 5.03×10⁶m³. The new landslides also significantly reduced from 756 in 2018 to 16 in 2019, making up 0.26% of the PSLs, covering 0.61×10⁶m³. The recovered and recovering landslides obviously increased from 74 in 2018 to 538 in 2019, constituting 8.68% of the PSLs, and the unchanged landslides without vegetation coverage also distinctly grew from 2594 in 2018 to 5604 in 2019, accounting for 90.34% of the PSLs covering 34.15×10⁶m³.

In the second year, debris flows exhibited a variation tendency similar to landslides. The active debris flows clearly dropped from 95 in 2018 to 6 in 2019, making up 1.75% of the PSDFs, covering 0.54×10⁶m³. Also, the new debris flows dramatically declined from 195 in 2018 to 15 in 2019, accounting for 4.39% of the PSDFs, covering 1.80×10⁶m³. In contrast, the recovered debris flows increased from 0 in 2018 to 11 in 2019, comprising 3.22% of the PSDFs and covering 0.14×10⁶m³. The unchanged debris flows soared from 37 in 2018 to 310 in 2019, comprising 90.64% of the PSDFs and covering 25.39×10⁶m³.

Therefore, over the first year, the PSLDs remained intense and developed. Most PSLDs, especially debris flows, stayed moving with enlarged areas, newly generated deposits, and mass wasting. Few PSLDs were recovering and only one PSLD was basically recovered. Moreover, the intensification of debris flow activity was another significant characteristic, and debris flows became an important hazard after the earthquake. Although the number of new debris flows was much lower than that of new landslides, the volume of new debris flows was 2.6 times the volume of new landslides. The volumes of new debris flows and new landslides accounted for 24.72% and 9.5% of the total PSLD volume, respectively. The enhanced rainfall in the first year may have to a certain degree facilitated the PSLD activity.

The PSLD evolution within the second year is in sharp contrast to the development in the first year. Both the active and new PSLDs dramatically reduced in number and volume, occupying a small portion of the PSLDs. Meanwhile, the recovering and recovered PSLDs significantly increased in number and volume. The vast majority of the PSLDs remained unchanged and unvegetated, and the unvaried ones attained 90.36% of the PSLDs in quantity ratio and 85.16% in proportional volume. Furthermore, recovery from debris flows was evidently slower than that from landslides. The number of restoring and restored landslides was 8.82 times the number of active and new landslides, whereas the number of restored debris flows was only 52.38% of the number of active and new debris flows. The restoring and restored landslides and the recovered debris flows accounted for 8.22% and 0.17% of the PSLDs in number, respectively, and comprised 3.24% and 0.2% of the PSLDs in volume, respectively. Accompanied by the decreased rainfall from an average of 851 mm in 2018 to 610 mm in 2019, the trends are obvious for the self-healing of the geological environment and hazard mitigation. Therefore, the geological environment was rapidly recovering in the absence of extreme rainfall events.

Furthermore, the geohazard chains in the 2 years mainly include (1) evolution from landslides to debris flows and (2) lake-level uplift and dam break soon after the earthquake. (See the supplementary file for the concrete illumination of the geohazard chains.)

Discussions

Relationship between geohazard evolution and influencing factors

The maps of various influencing factors are shown in Figures S10 and S11 in the supplementary file, and the relationships between geohazard evolution and influencing factors are shown in Figures S12–S14 in the supplementary file and Figures 8 and 9. The road network and river system remained almost unchanged during the 2 years after the earthquake. Among the influencing factors, only 12 factors played important roles in landslide and debris flow evolution, including distance to seismogenic fault, distance to epicenter, PGA, seismic intensity, fault density, distance to fault, aspect, NDVI, distance to river, cumulative rainfall, distance to road, and land use. Therefore, the PSLD evolution is controlled or induced by the antecedent earthquake, fault tectonics, topography, vegetation growth, hydrology, precipitation, and human engineering activity.

Figure 8

Aspect effects on the active and new landslides and debris flows in 2018. a Back-slope and east-slope effects on the active and new landslides. b Back-slope and east-slope effects on the active and new debris flows

Figure 9

Statistics of the aspect effects on the active and new landslides and debris flows in 2018. a Statistics of the back-slope and east-slope effects on the active and new landslides. b Statistics of the back-slope and east-slope effects on the active and new debris flows. The abbreviations of NB, AB, DB, and ADB mean the number, area, number density, and area density of the geohazards on the back slopes, respectively. The abbreviations of NF, AF, DF, and ADF represent the number, area, number density, and area density of the geohazards on the front slopes, respectively. The abbreviations of NE, AE, DE, and ADE indicate the number, area, number density, and area density of the geohazards on the eastern slopes, respectively. The abbreviations of NW, AW, DW, and ADW signify the number, area, number density, and area density of the geohazards on the western slopes, respectively

Recovering and recovered geohazards and influencing factors

The important factors influencing PSLD recovery included distance to seismogenic fault, distance to epicenter, PGA, seismic intensity, fault density, distance to fault, distance to river, NDVI, cumulative rainfall, and distance to road. Thus, the anterior earthquake, fault tectonics, hydrology, vegetation growth, precipitation, and human engineering activity controlled or induced the restoration of PSLDs. There are three characteristics of PSLD recovery. First, the PSLDs, which were less influenced by the earthquake and fault development and far away from rivers and roads, more easily resumed to a new balance. Second, the geological environment and precipitation condition conducive to vegetation growth were beneficial to PSLD restoration. Vegetation development provided a good capacity for soil and water conservation, reinforced the slope stability (Swanson and Dryness 1975; Woo and Luk, 1990; Sidle and Ochiai 2006), and contributed to PSLD recovery. Third, the newly emerging geohazards after the earthquake were more inclined to resume than the coseismic hazards. The influencing factors and the three characteristics are specifically highlighted in the supplementary file.

Active geohazards and influencing factors

The dominant factors for active PSLDs included distance to seismogenic fault, seismic intensity, distance to fault, fault density, aspect, distance to road, distance to river, and precipitation. Therefore, the anterior earthquake, fault tectonics, topography, road construction, river system, and precipitation led to active PSLDs. There are three distribution features of active PSLDs. First, the active PSLDs primarily appeared in the areas relatively close to faults, roads, and rivers and with abundant rainfall. Compared with the first year, in the second year, the influences of human engineering activity and river erosion intensified, and the active PSLDs that were impacted by human engineering activity and river development increased in quantity proportion. Second, the back-slope effect (Xu and Li 2010) was obvious for the active PSLDs, especially for the active landslides in the first year after the earthquake (Figures 8 and 9). Seismic waves propagated from northwest to southeast (CENC 2017; Li et al. 2017), and the active landslides relatively clustered on the slopes nearly perpendicular to the seismogenic fault and backing to the seismic wave propagation. Third, the east-slope (and southeast-slope) effect was interesting for the active PSLDs in the first year (Figures 8 and 9). Some ridges were almost parallel to the seismogenic fault or intersected the seismogenic fault at small angles. Active PSLDs were relatively concentrated on the east-facing slopes of the ridges. The dominant aspects of the back slope and east slope may have been due to stress enlargement when seismic waves were compressed and multiplicatively reflected from the free surfaces of the back and eastern slopes, which caused slope spalling, shattering, and relaxing (Tang et al. 2009; Xu and Li 2010). Tension cracks and failure planes tended to occur in the superficial rock mass of the back and eastern slopes (Tang 1997; Sang et al. 2003). Moreover, the seismogenic fault was dominated by a NW-SE-oriented left-lateral strike-slip movement (CENC 2017; Li et al. 2017; Yi et al., ..), and the left-lateral mechanism may have influenced the refraction and reflection of the seismic waves between the eastern slope surfaces and the internal slopes and may have been involved with the eastern slope shattering (Wang et al. 2018). Slope relaxation led to instability of the back and eastern slopes and made them tend to become active hazards within a short period after the earthquake. The dominant controlling and triggering factors and the distribution characteristics of active PSLDs are concretely elucidated in the supplementary file.

New geohazards and influencing factors

There are two types of new geohazards. The first type is the re-sliding of the loose body on the original landslides or debris flows to form a new large landslide or debris flow. The second type is newly emerging landslides and debris flows independent of the re-sliding of the original loose deposits. The first type was closely associated with the distribution of coseismic landslides and debris flows and concentrated in the areas with dense coseismic geohazards. Regarding the second type, the critical factors for the new PSLDs included distance to the seismogenic fault, seismic intensity, PGA, fault density, distance to fault, aspect, distance to river, distance to road, land use, and rainfall. Thus, the antecedent earthquake, fault tectonics, topography, river system, human engineering activity, and precipitation controlled or triggered the occurrence of the second type of new PSLDs. There are three distribution characteristics of the second type of new PSLDs. First, in the first year, abundant rainfall played an important role in newly occurring PSLDs, and new PSLDs also tended to appear near faults. Human engineering activity and river development to some degree contributed to the new emergence of PSLDs. Second, within the second year, human engineering activity, for example, road construction and land use, had an important influence on newly emerging PSLDs. The new PSLDs were mainly triggered by human engineering activity. Third, the distribution of new PSLDs (including the two types) in the first year was not evidently controlled by the back-slope effect but was greatly affected by the east-slope effect (Figures 8 and 9). New PSLDs were relatively concentrated on the eastern slopes of the ridges parallel to the seismogenic fault or intersecting the seismic fault at small angles. Thus, the stress difference between the eastern and western slopes caused by the seismic waves was relatively large, which resulted in the difference of the slope-loosened degrees and then led to the difference in PSLD occurrence levels. The critical influencing factors and distribution features of new PSLDs are specifically clarified in the supplementary file.

Comparison with some earthquake events

Certain earthquake events, for which PSLD evolution and coseismic geohazard magnitudes were specifically investigated in the previous works, are selected for comparison with the Jiuzhaigou earthquake (Tables 4 and 5).

Table 4 Comparison of coseismic hazard magnitudes in five earthquake events. The earthquake magnitudes of the Jiuzhaigou, Gorkha, and Wenchuan earthquakes are from the National Aeronautics and Space Administration and China Earthquake Administration. The earthquake magnitudes of the Kashmir and Murchison earthquakes are from Xu (2015). The M_w and M_s values of the Chi-Chi earthquake are from National Aeronautics and Space Administration and Xu (2015)

Table 5 Comparison of fault characteristic, precipitation, and recovery period of geological environment in five earthquake events

For fault characteristics, the Jiuzhaigou earthquake was triggered by a blind left-lateral strike-slip fault (CENC 2017; Li et al. 2017; Yi et al., 2017), whereas the other earthquakes in Table 5 are induced mainly by thrust or reverse faults (Berryman 1980; Lee et al. 2001; Sato et al. 2007; Zhao and Xu 2008; Kargel et al. 2016; Xu 2018). A reverse (including thrust) earthquake results in significantly stronger surface damage than a strike-slip earthquake of similar magnitude (Xu 2018). Moreover, four of these thrust or reverse earthquakes caused surface rupture or surface displacement (Henderson 1937; Lin et al. 2006; Zhao and Xu 2008; Kaneda et al. 2008). However, there was no obvious fault fracture or trace generated during the Jiuzhaigou earthquake (CENC 2017; Xu et al. 2017; Yi et al. 2017). Therefore, there was no obvious damage caused by the Jiuzhaigou earthquake to the Earth’s surface, and the destruction was less than that of the other five earthquakes.

With regard to coseismic hazard magnitudes, because of the strike-slip focal mechanism (Xu et al. 2017; Yi et al. 2017, ..) and no obvious surface rupture (Fan et al. 2018), the geohazards triggered by the Jiuzhaigou earthquake were dominated by small scales and shallow depths; thus, they possessed much smaller areas and volumes than those caused by the five thrust or reverse earthquakes (Berryman 1980; Lee et al. 2001; Sato et al. 2007; Zhao and Xu 2008; Kargel et al. 2016; Xu 2018). The landslide volumes triggered by the four earthquakes (Gorkha, Wenchuan, Kashmir, and Chi-Chi) were 22–272 times the volume induced by the Jiuzhaigou earthquake, and the landslide area resulting from the Murchison earthquake was nearly 18 times the area caused by the Jiuzhaigou earthquake. Therefore, the loose deposits generated by the Jiuzhaigou earthquake should be much less voluminous than the sediments generated by the other five earthquakes.

For rainfall, Jiuzhaigou County is characterized by a plateau monsoon climate (Gu 2019) with relatively low rainfall. The rainfall in Jiuzhaigou County is significantly less than that in the affected areas of the Gorkha, Kashmir, and Chi-Chi earthquakes. The average annual rainfall in the earthquake-influenced areas of Gorkha, Kashmir, and Chi-Chi is nearly 3 times, 2 times, and over 3 times the mean annual rainfall in Jiuzhaigou County, respectively. Although the rainfall in Jiuzhaigou County is similar to that in the partially affected area of the Wenchuan earthquake, the fault generated during the Wenchuan earthquake propagated approximately 240 km from the epicenter (Chen 2008; Zhao and Xu 2008), and the coseismic landslide volume triggered by the Wenchuan earthquake was over 272 times the volume caused by the Jiuzhaigou earthquake.

For the recovery period, the landslide activities following the Kashmir and Chi-Chi earthquakes quickly returned to equilibrium and the landscapes resumed new stabilities within 2 to approximately 6 years after the earthquakes (Saba et al. 2010; Khattak et al. 2010; Hovius et al. 2011; Khan et al. 2013; Shafique 2020). In contrast, the landslide and debris flow activity after the Wenchuan earthquake may be restored to pre-earthquake levels within two decades (Yang et al. 2018a) or even longer under the action of extreme rainfall (Li et al. 2018), due to enormous landslides (Xu et al. 2016a), extremely large rock avalanches (Xu et al. 2014; Fan et al. 2018), and massive destruction to the geological environment and landscape. Similarly, the landslide evolution following the Murchison earthquake was characterized by the long-term sediment retention after the earthquake (Pearce and Watson 1986). After 50 years of evolution, at least 50 to 75% of the granitic debris still remained in the fourth-order 92-km² catchment (Pearce and Watson 1986). In addition, within 3 years following the Gorkha earthquake, 56% of the landslides along the two earthquake-affected highways proceeded to deform and expand and may require over 3 years to restore (Tian et al. 2020). For the Jiuzhaigou earthquake, landslide and debris flow activity intensified in the first year after the earthquake. Most geohazards, especially debris flows, dynamically progressed or resurged under enhanced rainfall, accompanied by additional loose deposits, expanded areas, and mass wasting. However, in the second year, the earthquake influence quickly decayed as the rainfall decreased to the pre-earthquake level, and only a small amount of landslides and debris flows stayed moving. The landslide and debris flow evolution in the second year was primarily controlled by fault development and induced by rainfall and human engineering activity, for example, road construction and land use.

Therefore, compared with the above five earthquakes, the Jiuzhaigou earthquake is characterized by a smaller earthquake magnitude, no obvious surface rupture (Fan et al. 2018), much smaller hazard magnitudes, obviously less rainfall, and rapid attenuation of the earthquake influence. Thus, the geohazard activity and landscape after the Jiuzhaigou earthquake may potentially be restored to equilibrium more quickly than after the Kashmir, Chi-Chi, Gorkha, Wenchuan, and Murchison earthquakes.

Evolution rule and trend of post-seismic landslides and debris flows

The evolution rule and trend of PSLDs are revealed by integrating geohazard investigation, loose deposit investigation, PSLD evolution, influencing factors, and mass wasting analyzed above.

• (1) The Jiuzhaigou earthquake triggered 5563 coseismic geohazards in the study area, including 5431 landslides covering 34.76×10⁶m³ and 132 debris flows covering 7.85×10⁶m³. Of these coseismic hazards, 97.43% possessed small scales and 2.52% had medium scales. Of the coseismic hazards, 90.87% occurred in the area with relatively abundant anterior rainfall; thus, infiltration of the anterior precipitation and water softening may make the rock mass more prone to instability during a strong earthquake (Sassa et al. 2007; Fan et al. 2019).

• (2) The landslide and debris flow activity became aggravated in the first post-earthquake year. The antecedent earthquake shattered and relaxed the slopes (Tang et al. 2009; Xu and Li 2010), and the coseismic landslides and debris flows generated a mass of loose deposits. Moreover, rainfall significantly increased in the first year, and the developed faults caused broken rock mass, benefited rainfall penetration and groundwater seepage, and created favorable tectonic conditions for slope instability (Koukis et al. 2009). In addition, road construction and fluvial scouring eroded slope toes, which reduced the counterpressure of slope feet (Maharaj 1993; Kamp et al. 2008) or caused hydrodynamic pressure via underground water infiltrating, rising, and declining (He et al. 1998; Miao et al. 2014), and then destroyed slope balance. Therefore, the PSLDs intensified within the first year under the joint effect of loosened slopes, sufficient source materials, developed faults, enhanced rainfall, anterior road construction, and river erosion. However, human engineering activity and river erosion contributed less to the PSLDs than the anterior earthquake, fault tectonics, and precipitation. Most PSLDs stayed moving, highlighted by active and newly emerging landslides and debris flows, accompanied by enlarged areas, newly generated sediments, and mass wasting. A handful of PSLDs were undergoing recovery, and only one landslide was restored.

• (3) Landslide and debris flow activity quickly decayed in the second year and may continue several years before returning to a new balance. Driven by gravity and water erosion (Lin et al. 2008; Zhang and Zhang 2017), loose deposits were transported from high altitude to low elevation, and a few were delivered into rivers and lakes. The dissipative deposits were much abundant than the newly generated deposits; thus, the sediments dramatically reduced with time. In the second year, rainfall reduced to the pre-earthquake level, and slopes experienced self-healing by grain coarsening, material consolidation, and vegetation growth (Fan et al. 2018). Therefore, PSLDs rapidly resumed in the second year because of dramatically reduced sediments, decreased rainfall, and self-healing slopes. The recovering and recovered PSLDs significantly increased, and the vast majority of the PSLDs remained inactive. However, the anterior earthquake, fault tectonics, human engineering activity, and meteorology have far-reaching influences on PSLD evolution, and a few PSLDs remained active or newly appeared under the combined action of earthquake-relaxed slopes, fault development, road construction, bare soils and rocks, and rainfall. Moreover, over 90% of the PSLDs remained unchanged and unvegetated and may present two trends in the future. One trend is that without extreme rainfall events, these geohazards may gradually recover and be vegetated, and the sediments may consolidate as part of the landscape (Fan et al. 2019). The other is that in rainstorm events, these geohazards constitute important hidden dangers and may become active again and generate numerous loose deposits. Although the PSLDs were rapidly decaying and the geological environment was promptly restoring, the geological environment and landscape had not returned to the pre-earthquake levels, because the unrecovered geohazards were still quite significant both in number and volume (69.66×10⁶ m³), and abundant sediments still remained on hillslopes or in channels 2 years after the earthquake.

• (4) Intensified debris flow activity was a prominent feature shortly after the earthquake. In the first year, 71.97% of the coseismic debris flows remained active, and the newly emerging debris flows dominated the new geohazard activity, mainly due to the rising rainfall and sufficient source materials. No coseismic debris flow recovered in the first year. In the second year, the activity of debris flows dramatically reduced with small amounts of active and new debris flows; however, debris flow activities were still prominent and occupied 74.59% of the total volume of the new PSLDs. Moreover, over 90% of the debris flows remained unchanged and unvegetated and may become hidden dangers in the future when experiencing heavy rains. In addition, debris flows were restored more slowly than landslides, and the number of the restored debris flows was only half of that of the active and new debris flows in the second year.

• (5) The geohazard chains within the 2 post-seismic years mainly consisted of (1) evolution from landslides to debris flows and (2) lake-level uplift and dam break soon after the earthquake. Except for chain (2) that occurred shortly following the earthquake, there was no dammed lake generated during the 2 years. With regard to development from landslides to debris flows, loose deposits generated by landslides remained on slopes or were transported to channels by gravity and water erosion (Lin et al. 2008; Zhang and Zhang 2017). These loose deposits provided sufficient source materials for debris flows. Under rainfall events, these sediments, carrying other erosion materials and water, rushed down slopes or to ravine mouths, and then hillslope debris flows or channel debris flows occurred (Curry 1966; Griffiths et al. 2004; Chen et al. 2012; Zhang and Zhang 2017). With respect to dam breaks, vast loose deposits generated by the coseismic landslides and debris flows moved into the lakes and raised the water levels together with the action of the earthquake-induced geomorphology change (Yang et al. 2018b). The pressure on the dam walls correspondingly increased, causing dam breaks (Yang et al. 2018b). The dam breaks generally occurred soon after the earthquake; for example, water in Huohua Lake ran dry on August 9, 2017, shortly after the earthquake.

• (6) Aspect controls geohazard evolution and distribution within a short period after the earthquake. Some previous works (e.g., Wang et al. 2011; Zhao et al. 2018) indicated that the topography amplification effect on the PGA of seismic waves intensifies on the back slopes; thus, back slopes are inclined to fail more easily, which is called the back-slope effect on coseismic landslides (e.g., Xu and Li 2010; Xu et al. 2011; Wu et al. 2018). In addition, under the integrated influence of fault movement direction and valley orientation, east and southeast were the dominant aspects of the Jiuzhaigou coseismic landslides (Wang et al. 2018). This work reveals that the fractured and loosened rocks on the back, east-facing, and southeast-facing slopes not only influenced the coseismic landslide distribution but also controlled the PSLD evolution and distribution in a short period. The reinforced destruction to the back, east, and southeast slopes was sourced from the coupled effect of topography amplification and fault movement. The slope destruction caused the active and new PSLDs to tend to occur on the slopes on the three aspects. Moreover, it seemed that active landslides were related to both the topography amplification effect and fault movement direction, and new landslides and debris flows were mainly controlled by fault movement direction.

• (7) The geological environment and landscape after the Jiuzhaigou earthquake may return to new balances and restore to the pre-earthquake levels more quickly than after the Kashmir, Chi-Chi, Gorkha, Wenchuan, and Murchison earthquakes. The strike-slip focal mechanism and the blind seismic fault of the Jiuzhaigou earthquake did not generate obvious fault fractures or traces on the Earth’s surface (Xu et al. 2017; Yi et al. 2017) and restricted the degree of destruction to the surface. The coseismic and post-seismic geohazards were dominated by small scales, and the meteorological conditions were not favorable for geohazard development due to relatively low rainfall. Thus, although the PSLD activity was aggravated accompanied by the enhanced rainfall in the first year, the earthquake influence rapidly decayed in the second year with rainfall declining to the pre-earthquake level. The newly occurring and dynamically expanding landslides and debris flows comprised only 1.25% of the PSLDs in the second year, and these hazards were mainly controlled by fault development and triggered by rainfall and human engineering activity, such as road construction and land use. Therefore, the recovery period of the PSLD activity and landscape in the Jiuzhaigou area may be shorter than previously expected in the absence of extreme rainfall events.

Conclusions

In this work, the comprehensive investigation of landslide and debris flow evolution after the 2017 M_s 7.0 Jiuzhaigou earthquake is reported. This work focuses on the spatio-temporal evolution of PSLDs under the effects of the earthquake, geology, terrain, precipitation, drainage system, and human engineering activity. The Jiuzhaigou region possesses unique and significant merits in geology, society, and economy, which is characterized by the famous north-south seismic belt in China (Li et al. 2017), the M_s 7.0 strong earthquake (CENC 2017), ultracomplex geological and topographic context (Lei et al. 2018), UNESCO World Natural Heritage Site (UNESCO 1992; Fan et al. 2018), and World Biosphere Reserve (Fan et al. 2018). Thus, the Jiuzhaigou region is selected as the study area in this work. Multisource and multi-temporal data, including seismic, geological, topographic, meteorological, and remote sensing data, are employed to provide new insights into the PSLD distribution, loose deposit distribution, PSLD evolution, influencing factors, hazard chain, and mass wasting in the Jiuzhaigou area from 2017 to 2019. This work may provide new knowledge pertaining to landslide and debris evolution after a strong earthquake and profound impacts of a catastrophic earthquake on geological environment and landscape. Six main conclusions are drawn as follows.

• (1) The landslide and debris flow activity became intensified within the first year after the earthquake under the combined effect of loosened slopes, sufficient source materials, developed faults, enhanced rainfall, anterior road construction, and river erosion. Moreover, intensified debris flow activity was a prominent feature shortly after the earthquake due to the rising rainfall and sufficient source materials. Debris flows were restored more slowly than landslides within a short period after the earthquake.

• (2) Landslide and debris flow activity quickly decayed in the second year because of dramatically reduced sediments, decreased rainfall, and self-healing slopes. The activity may continue several years before returning to a new balance. The PSLD activity was rapidly decaying and the geological environments were promptly restoring in an evolution trend to a new balance in the absence of rainstorm events. However, the geological environment and landscape had not returned to the pre-earthquake levels because of the vast unrecovered landslides and debris flows covering 69.66×10⁶ m³ and the remaining loose deposits trapped on hillslopes or in channels. Furthermore, the PSLD activity level may fluctuate and significantly increase in the future when experiencing extreme rainfall events.

• (3) The geohazard chains within 2 years after the earthquake mainly consisted of (a) evolution from landslides to debris flows and (b) lake-level uplift and dam break soon after the earthquake.

• (4) The earthquake-generated sediments were quickly consumed by water erosion, transport to rivers and lakes, and artificial removal. The sustained and rapid consumption of coseismic loose deposits and the prominent reduction in new sediments indicate that the geological environment was gradually restoring.

• (5) The geological environment and landscape after the Jiuzhaigou earthquake may return to new balances and restore to the pre-earthquake levels more quickly than after the Kashmir, Chi-Chi, Gorkha, Wenchuan, and Murchison earthquakes, as a result of a smaller earthquake magnitude, no obvious surface rupture, much smaller hazard magnitudes, obviously less rainfall, and rapid attenuation of the earthquake influence.

• (6) The activity and new emergence of post-earthquake landslides and debris flows are chronically influenced by the anterior earthquake and pre-earthquake human engineering activity (especially road construction) due to the sustained effects of the fractured and loosened rocks and slopes. Therefore, human construction activity has a profound impact on geohazard evolution, and the balance between economic progress and the natural environment should be more carefully considered.