Creeping slopes in NW Himalaya and Joshimath slide: constraints from GPS measurements

Abstract Slow slope movement is very common in the Himalayan region. In most cases, the topsoil, weathered rock mass, debris of paleo-landslide or of glacial origin, etc. slowly slides until it stabilizes or fails. In some cases, even the bedrock may be involved in sliding. Here we report four such sliding slopes on which GPS were actually installed for tectonic deformation monitoring in the Himalayan region after a careful investigation but they recorded slow slope sliding. The slow sliding motion on these slopes varied from 10-15 mm/year to 5-6 m/year. The slow slides did not show any anomalous seasonal variation due to hydrological loading. Amongst the four, the Joshimath slide exhibited a large variation in sliding motion which varied from less than a mm/day to 15 mm/day during 2022-2023, causing cracks on the ground and buildings and panic amongst the residents. Even the highly non-linear motion with a sudden change in sliding motion at Joshimath appears to be of viscous creep type. The exact time of initiation of this episode of slide is not known but the available InSAR analyses imply that it is moving at least since 2018. Its motion does not seem to be affected by the February 7, 2021 floods, however, whether the anthropogenic activities around the sliding zone influenced its motion and initiation, remains to be debated.


Introduction
The Himalayan region, being the convergent plate margin between the India and Eurasia plates, is prone to several geohazards.Earthquakes and landslides are the most common of them.Major and great earthquakes occur on the seismically active Main Himalayan Thrust (MHT) beneath the Outer and Lesser Himalaya, while the MHT beneath the Higher and Tethys Himalaya slips aseismically.Small and moderate magnitude earthquakes of the Himalayan seismic belt occur (Seeber and Armbruster 1981) occur in the transition from seismically active (MHT), to aseismically slipping MHT (Bilham et al. 1997).Landslides in the Himalayan region occur due to slope failure.Catastrophic landslides are the result of the sudden and rapid transport of rock, soil and water along a slope.They are strongly influenced by the slope, rock type, precipitation, anthropogenic activities, earthquakes, and tectonics.Many slopes in the Himalaya also slide slowly and their sliding speed (from a few mm/year to meter/year) may change with time.In many cases, they stabilize with time, either naturally or due to efforts taken to stabilize them.Speed of slide or creep may also be influenced by various factors, e.g.weathering, development of pore pressure, rainfall/snowmelt, irrigation and drainage, cutting of slope toe due to human activities, vegetation, earthquake shaking, loading due to construction of houses and other civil structures, tectonic uplift, etc. (Van Asch 1984;Lacroix et al. 2020).In the Himalayan region, slow slope movement is very common and their evidence in the form of the presence of curved trunks of pine trees on the slopes is abundant (Figure 1).Other than the influence of snow load on trees on higher elevations and their own weight, such bending in the lower part of the trunk depicts slow slope movement or creep (Harker 1996).Since the slope movement is highly variable with time and space, it's not necessary that all the trees present on the slope should exhibit such behaviour.
The slow movement of slopes sometimes may accelerate and fail catastrophically leading to runaway landslides.This can disrupt traffic and may also cause loss of property and lives.In case a town is located on such a slow-moving slope, then in the event of failure of the slope as a runaway landslide, it can cause a disaster.Joshimath in the Garhwal Himalayan region of India is one such recent case where the slow moving slope causing extensive cracks on the ground and in buildings in late 2022 and early 2023 has created panic amongst people and has led some of the population to relocate.Besides numerous landslides in the region, it also witnessed another disaster which occurred on February 7, 2021, about 30 km east of Joshimath town when a snow avalanche and rockslide caused a devastating flood in the Rishiganga and Dhauliganga rivers, destroying a power plant in which �200 people lost their lives (Cook et al. 2021, Rana et al. 2021;Meena et al. 2021;Pandey et al. 2022).In this article, we provide evidence of slow moving slopes from the Himalaya detected using ground based GPS measurements of crustal movement.Interestingly, all these GPS sites were identified for crustal deformation measurements due to tectonic processes after a careful examination to avoid any unstable slope.Despite this, at least four, out of �50, sites in the Himalayan region have inadvertently been installed on the slow moving slopes and have clearly documented slow motion along the slope which varies from a few mm/year to a meter/year.On the other hand, the sites not affected by the slope movement exhibit motion of about 10-20 mm/year towards SWS, i.e. in the Himalayan arc normal direction with respect to the Indian plate (Figure 1 of Yadav et al. 2019;Figure 2 of Yadav et al. 2020; Figure 1 of Mondal et al. 2022).Except for the site NAKO in Spiti valley, the other three sites are located in the Higher Himalaya which marks the physiographic transition with sudden changes in elevation, higher uplift and erosion rate (Hodges et al. 2001).

Measurements of slow motion using GPS observations
A large number of GPS sites have been installed in the Himalayan region for the purpose of monitoring crustal deformation due to tectonic and earthquake processes.The main purpose of the installation of these sites is to identify the extent of the region of strain accumulation along with its rate (Yadav et al. 2019).Such a network also provides valuable information on the earthquake source parameters in case of occurrence of a strong earthquake (Galetzka et al. 2015;Yadav et al. 2017).For this purpose, the geodetic grade multi-frequency GPS receiver and choke ring antenna have been used.In all cases, the antenna has been installed on a strong RCC pillar anchored to the bedrock, and unstable slopes and sites with loose soil or alluvium are avoided.Continuous GPS data from these sites in the Himalayan region have been analysed together with several International GNSS (Global Navigation Satellite System) Service (IGS) sites surrounding the Indian plate.We used GAMIT, version 10.60 (King and Bock 2005;Herring et al. 2010aHerring et al. , 2010b)), to estimate the time series of site coordinates and site velocity.Site position estimates and their rates were estimated in ITRF2014 (Altamimi et al. 2011) by stabilizing more stable continuous sites and core IGS reference sites using GAMIT/GLOBK, GLORG.The sites on stable ground in the Himalayan and adjoining regions exhibit a consistent north-eastward motion varying from 57 to 36 mm/year in ITRF14 and in the Indian plate reference frame (Ader et al. 2012) these sites exhibit arc-normal, predominantly southwestward motion of 1-15 mm/year (Yadav et al. 2019).However, four sites (RATH, GUPT, JOSH and NAKO, shown in Figure 2) identified here show anomalous motion in the direction of slope.The motion at these sites is different from the neighbouring sites and is inconsistent with the strain accumulation model on subsurface faults.As these sites also include contributions from plate motion and strain accumulation, we subtract the site motion of a nearby stable site located a few km from it, from the motion at these sites on slow moving slopes.Such a procedure allows us to estimate the slow slope motion, as the tectonic contribution arising from the strain accumulation is subtracted from the site motion.Although, as such there is no clear quantification of slow and fast movement of slopes, we refer slope movement as slow as long as its magnitude is comparable to the movement due to tectonics and earthquake processes at other nearby sites in the region.Higher slope movement is referred here as fast or accelerated movement.

Slow slide at Raithal and Bhatwari near Uttarkashi, Garhwal Himalaya
There are two continuous GPS sites (RATH and BHTW) on an east sloping hillock with a slope of �20 � in the Main Central Thrust (MCT) zone of Himalaya overlooking the Bhagirathi river (Figure 2).The region consists of a migmatite zone of mica schist and gnesiss of the central crystalline group with a thick capping of hill slope sediments.Site BHTW was operated by the Wadia Institute of Himalayan Geology, Dehradun and data from this site until 2013 were reported by Yadav et al. (2019).Site RATH is operated by CSIR-National Geophysical Research Institute, Hyderabad from year 2012 to 2021.The two sites are located on an eastward sloping hill with an average elevation of �2200 m (Figure 2b).The GPS site BHTW near the toe and RATH on the crown, show similar and predominantly eastward motion.All the nearby sites, including a campaign mode site, BHAT, which is located 1.5 km south of the two sites (Banerjee et al. 2008) on a different hillock across the river Bhagirathi river (Figure 2), show predominantly southward motion (towards N213 � ) at a rate of �7 mm/year in the Indian reference frame which basically corresponds to strain accumulation on the locked MHT (Yadav et al. 2019).As the sites BHTW and RATH also record tectonic motion due to strain accumulation, we removed the motion of BHAT from these sites to estimate the contribution of sliding motion only.Because of similar motion at these two sites which are only 1.4 km apart, we show a composite plot showing slope motion in Figure 3.The two sites now show an insignificant northward motion varying from 1 to 8 mm/year during 2006 to 2021.The eastward movement varies from 13 to 16 mm/year during the same period (Figure 4).Vertical component records subsidence of about 6 mm during 2011-2014 and then again towards the end of 2020.Thus, the motion observed at sites is in the direction of the slope and along the dip direction of the bedrock.All the above factors imply that this creep motion is due to the slow moving landslide.

Slow slide at Phata, Guptkashi, Garhwal Himalaya
The site GUPT is located �50 km southeast of BHTW and RATH in a similar tectonic and geological domain as that of site RATH and BHTW.The site at an elevation of 1440 m (Figure 2c) is located on a gently sloping hillock overlooking the Mandakini river (Figure 2a).The overall slope of the hillock is �14 � towards the northeast with a lesser slope of �8 � near the site.In the Indian reference frame, the site showed large NE motion whereas the nearby sites on stable ground, e.g.HRMN, a site located �35 km southeast of GUPT (Figure 2), exhibit predominantly SSW motion of 8 mm/year consistent with the strain accumulation in the region (Yadav et al. 2019).With reference to site HRMN, site GUPT exhibits almost linear motion of �13 mm/year towards the northeast in the slope direction (Figure 4) which implies that the site is located on a slow moving slope but without any subsidence.

Slow slide at NAKO, Spiti valley, higher Himalaya
The site NAKO at village Nako in the Spiti region was established in early 2018.The site at about 3700 m (Figure 2d) is located east of Kaurik Chango rift (Figure 2a) on a gently sloping hillock of cretaceous granite of Higher Himalaya with a local slope of �8 � towards west.However, the overall slope of the region from Nako town to Spiti river in the west is �18 � towards the west and the distance between the two is �1.8 km.The site shows extremely high velocity towards the west, in the direction of the slope.We use a nearby site LARI, �30 km northwest of NAKO (Figure 2a), as a reference site to estimate the sliding motion at NAKO.The westward sliding motion at NAKO is estimated as �62 mm/year during the year 2018-2021 (Figure 5).Although there is some data gap in the year 2019, the motion appears to be linear with usual seasonal variations.The site also exhibits large subsidence at the rate of �22 mm/year.

Fast, furious and non-linear slide at Joshimath, Garhwal Himalaya
Joshimath is located in the Higher Himalayan crystalline rocks and is very close to the MCT.It is located on a north sloping hillock with a slope of �22 � overlooking the confluence of the Dhauliganga and Alaknanda rivers (Figure 2a).It has both glacial and glacio-fluvial landforms and the slopes are covered with large fans of old landslide debris.This was first identified by Heim and Gansser (1939).The first documented evidence of subsidence and sliding was reported in the late 1960s when the government in 1976 constituted a committee to examine the cause of land subsidence at Joshimath.
After November 2022, the town of Joshimath reported widespread cracks in the majority of the houses and on the ground.The crisis intensified in the first week of January 2023 when more cracks developed.The InSAR analysis of deformation at Joshimath available at AGU landslide blogs pointed out slow sliding accompanying subsidence in the Joshimath town and in its immediate vicinity (https://blogs.agu.org/landslideblog/2023/01/18/joshimath-new-insar/ and https://blogs.agu.org/landslideblog/2023/01/23/joshimath-5/).These analyses suggest large land movement (up to 50 cm) since at least January 2018.Survey of India had established a GPS site, JOSH, in Joshimath town in Dec 2021 at an elevation of �1900 m.CSIR-NGRI had established a nearby site at AULI at an elevation of �1 km higher, �3 km south of JOSH and is operating since 2013 (Figure 2e).AULI, �3 km uphill towards the south on a hilltop, does not show any effect of landslide and moves towards SSW at the rate of �8 mm/ year (Yadav et al. 2020) in the Indian reference frame in response to the strain accumulation in the Himalayan arc.We use this site as the reference site and estimate the motion at JOSH with respect to it.The resulting motion at JOSH is mainly in the direction of slope, i.e. towards north along with subsidence (Figure 6) and does not contain tectonic motion or motion due to strain accumulation.Until mid-November 2022, the site exhibited a predominant northward motion of �1.5 mm/day with subsidence of �0.8 mm/day.This was followed by an intense movement of �7.4 mm/day towards the north with a subsidence rate of �4 mm/day until December 31, 2022.This was the period when residents started noticing cracks in the building and ground.The period between January 1 to 4, 2023 was even more intense when the site exhibited large northward movement of �15 mm/day with a subsidence rate of �11 mm/day.This was the period when the movement was most intense and local administration started the evacuation of people from damaged buildings and demolition of some buildings severely damaged due to cracks.However, after January 5, 2023, the site exhibits motion at a very slow rate of less than 0.5 mm/day (until April 8, 2023) which is still quite alarming.The ratio of horizontal and vertical movement may be used to estimate the dip of the slip surface, if we assume rigid body motion over a planar slip surface.In this case, it turns out to be �30 � .Considering that the slope of the hillock itself is �22 � , this estimate of the dip of the slip surface appears to be reasonable.It is not possible to estimate the depth of the slip surface, however, looking at the space-time heterogeneity in sliding and surface cracks, we speculate that it must be shallow, not more than a few tens of meters.

Seasonal variation in slide motion and influence of flood
All GPS sites in the Himalayan region exhibit seasonal variation in displacement in which during monsoon season they undergo subsidence of 10-15 mm and predominantly southward motion of �6 mm which is recovered in the following winter season.Such a motion is consistent with the hydrological and atmospheric loading effect (Gahalaut et al. 2017;Gautam et al. 2017;Yadav et al. 2019).Seasonal changes in the movement at sites affected by the slow motion, viz., RATH, BHTW, GUPT, and NAKO also show motion similar to other sites and is also consistent with the hydrological and atmospheric loading effect (Figure 7).Thus, it appears that seasonal rainfall did not have any additional influence on the landslide movement at these sites.
The heavy rainfall and ensuing floods of June 15-18, 2013 in the Garhwal-Kumaun region in Uttarakhand which also led to the Kedarnath crisis (Sati and Gahalaut 2013), caused huge devastation in most of the river valleys of Uttarakhand Himalaya.(Yadav et al. 2019;Mondal et al. 2022) and also with that due to hydrological and atmospheric loading (Gautam et al. 2017), shown here at each site (by continuous curves), thus implying that these variations are not unusual and the sliding movement remains unaffected by the seasonal changes.Because of discontinuous data during 2015-2018 at GUPT, that period is omitted.Histogram in each panel show monthly rainfall with the scale on the right-side Y axis.
The resulting floods in various valleys between the River Kali in the east to Yamuna in the west damaged several settlements along their banks.However, this event does not seem to have caused any increase in movement at RATH, BHTW and GUPT despite the extensive devastation even in Bhagirathi and Mandakini valleys where these sites are located.Thus, the slow but anomalous slope movement generally does not seem to be affected by seasonal rainfall and floods.
Due to extremely large and varying sliding motion and limited GPS data at JOSH, it is not possible to assess whether the displacement time series at the GPS site exhibits any anomalous seasonal variations.Even the inherent seasonal variations due to hydrological loading are masked by the large variation of sliding motion.However, the available InSAR analysis derived time series of line of sight (LOS) distance since 2018 may be used to infer the seasonal variations and the effect of flood.Interestingly, the InSAR analysis does not indicate any effect of seasonal variation on the LOS.The two available analyses dealing with the deformation using InSAR (AGU blogs on the Joshimath landslide of 2023/01/18 and 2023/01/23) indicate that the eastern part of the sliding zone started sliding at least since January 2018 which corresponds to the date of start of that analysis, while the motion in the western part probably became significant during September 2021.The LOS distance does not seem to be influenced by the February 7, 2021 flash floods in Dhauliganga, which occurred due to rockfall and avalanche and caused heavy damage at Tapovan, only �10 km from Joshimath.So, either the slide was not affected by the seasonal variations, particularly the monsoon, or the seasonal variations, if any, cannot be inferred due to the low resolution and limited accuracy of InSAR analysis.The former argument draws support from the deformation seen at the GPS sites exhibiting slow sliding which also shows the absence of such transients.

Similarities and differences in the slow moving slides
In all cases, the movement is in the direction of the slope, which may hint towards its destabilization.Further, as the movement is more than that expected from the tectonic or due to strain accumulation and is inconsistent with the expected direction of motion, i.e. towards SWS, it confirms that the motion is indeed the sliding motion of the slope.The magnitude of motion may vary from slope to slope depending upon the rheology, magnitude of slope, and magnitude of the external and internal factors influencing the sliding.However, in no case analysed here, there seems to be the influence of seasonal rainfall, which may imply that the depth of the sliding surface could be relatively deep in such cases.Since it is the first time that such data have been reported in the Himalayan region for the slow sliding movement, it is to be seen in what conditions the slow movement turns into a runaway catastrophic movement.

Viscous creep or stick slip motion
Large variation in the motion at JOSH allows us to explore whether the motion was smooth, involving viscous creep, or was through frictional slip, involving stick and slip.To check this aspect, we performed a kinematic data processing of GPS data at site JOSH which has a sampling interval of 30 s using the TRACK option in GAMIT/GLOBK.Thus, at each epoch, we estimated the coordinates with respect to a nearby site which is on stable ground and does not exhibit any slow motion.We chose a site at RUDR (at Rudraprayag) which is �30 km southeast of JOSH in a similar tectonic environment.We particularly focussed on the two episodes when the sliding motion changed dramatically (viz., November 10-20, 2022 andDecember 25, 2022 to January 5, 2023 in Figure 4a).We find that during these two episodes, the motion is essentially a creep motion exhibiting smooth variation, without any sudden change in sliding velocity.This is to be seen whether this creep motion caused any slidequake (Walter et al. 2013).Conventionally, it was considered that viscous creep (generally governed by the clay material and water) does not produce any slidequake.However, in recent times, it has been reported that the viscous creep motion may also cause slidequakes, implying some stick-slip motion (Walter et al. 2013).At Joshimath, there is a seismological station on the slide.The data from this site can be used to assess whether any tremor-like signals are recorded.

Influence of hydrological and anthropogenic activities
It is difficult to know what triggered the initiation of the slow slide movement at these sites.This could be natural as part of the mass wasting process in the mountainous region or could be influenced by hydrological and/or anthropogenic activities.Although the seasonal variation in the hydrological loading does not seem to affect the slow landslide motion at the three sites (RATH/BHAT, GUPT and NAKO), it is possible that the variation in slow slide motion at site JOSH could be linked with hydrological and anthropogenic activity.The viscous creep motion at JOSH implies the role of pore pressure.But the two episodes of accelerated slip leading to higher motion in 2022-2023 do not seem to be related to any apparent hydrological process or anthropogenic activities during that period.
It is likely that the slope of Joshimath town might have experienced such episodic accelerated movement intermittently in the past too.However, whether the current episode of rapid movement which certainly started at least in 2018 or earlier, was triggered by the monsoon or snowmelt from the top (draining out of Auli) with some delay, or it was due to some anthropogenic activities, is a matter of debate.Some of the known anthropogenic activities in this case include (i) construction activities in Joshimath which, as per google earth imageries, probably increased after 2012-2014, thereby increasing the load on the paleo-landslide debris of the slope and also disturbing the natural drainage system, (ii) development of several hydroelectric projects and construction of a few tunnels, some of them are alleged to have disturbed the aquifer, (iii) broadening of some highways connecting to the famous Hindu shrine, Badrinath, leading to toe cutting of some slopes, etc.However, in view of the lack of detailed and quantitative information on these activities, it is difficult to ascertain the cause of the initiation of this episode of sliding movement at Joshimath.

Risk of slow-moving landslides turning to catastrophic landslides
Process of landslide whether slow or runaway, is a process of mass wasting and is a natural process which brings equilibrium in the mountain building process due to tectonics, erosion, and other geological processes.However, in many cases, the process is influenced by the anthropogenic activities which impacts the safety and security of the local population.Catastrophic landslides certainly pose more risk and hence if the slow moving slopes change to catastrophic landslides, they may also pose a similar risk.Risk increases exponentially if the slow moving slope is inhabited.In some cases, depending upon the depth and geometry of the slip surface, rock properties, and the magnitude of sliding motion, the slope may develop cracks in the ground and buildings.Even if the slope is uninhabited, in case of catastrophic failure, it can obstruct the flow of a river which further poses a risk to the population living downstream.Unfortunately, all the above discussed slow moving slopes are inhabited, with the largest population in Joshimath.Thus, it is important to analyse the conditions under which a slow moving slope may turn into a catastrophic landslide.Surprisingly, both sliding modes are displayed by nearby slopes (and individual slopes at different times) subjected to almost identical environmental conditions (Handwerger et al. 2016).This has led to a search for mechanisms that can explain this transition.It is generally believed that pore pressure is responsible for the transition and catastrophic landslide (Handwerger et al. 2013).Variation in landslide motion has been explained using a rate-and state-dependent frictional model.These models imply that the failure mode depends on the material properties and the dimensions of the landslide slip surface relative to a critical size (Handwerger et al. 2016).If the slip surface exceeds this size and has rate-weakening frictional properties, then catastrophic failure can occur.
In the context of the Joshimath slide, a detailed analysis is not undertaken due to a lack of information on the required parameters.However, keeping in mind the large variation in the sliding magnitude with time (Figure 4) and space (available results of InSAR analysis), a few deductions may be drawn.It has been suggested that rate-weakening landslides can display both stable and unstable sliding.In fact, landslides exhibiting complex motion, consisting of smaller size high slip patches arising due to heterogeneity in basal strength may be less likely to fail catastrophically compared with landslides that exhibit more uniform motion (Handwerger et al. 2013).This scenario is similar to the Joshimath slide where InSAR analysis indicates that the sliding rate is varying spatially.Fortunately, the Joshimath sliding movement in the past two months (since Jan 5, 2023) has decreased significantly, bringing some temporary relief.However, considering the high risk involved in the Joshimath landslide, it is important that real-time slope monitoring at high spatio-temporal resolution is undertaken to understand the physics of the slide so that some scenarios for failure can be developed.
As a large population is exposed to the landslide risk at Joshimath (Sati et al. 2023), it is necessary to first characterize the landslide in terms of its movement, slip surface, rheological properties of the material, factors and processes affecting the motion, etc.This will help in undertaking engineering measures to stabilize the slope, if possible.

Conclusion
Out of more than 50 GPS sites in the NW Himalaya which were established for crustal deformation due to tectonic process, we found that four sites are located on slow moving slopes.Site motion in the Indian reference frame in the Himalaya due to the tectonic process is expected to be towards the southwest in the arc normal direction with a rate of up to 18 mm/year.But the four sites show large motion along the local slope of the hillock which varied from 10-15 mm/year to 5-6 m/year.Despite the reasonably long duration of measurements at these sites, the slow sliding movement did not show any anomalous seasonal variation due to hydrological loading or floods in the region.Among the sliding motion at four sites, the motion at Joshimath exhibits a large variation from <1 mm/day to 15 mm/day during 2022-2023.A detailed and high resolution analysis of the displacement time series at every 30second time interval, indicates that the sliding movement was of viscous creep type.Although the first report of slow sliding and subsidence was documented in the late 1960s, the initiation of the current sliding episode of Joshimath slide is not known but the available InSAR analysis implies that it is sliding at least since 2018.Fortunately, the sliding movement in the past few months (since Jan 5, 2023) has decreased significantly, bringing some temporary respite.However, it is important that slope sliding monitoring should continue with even better spatial resolution.

Figure 1 .
Figure 1.Curved lower trunks of pine trees in the Garhwal Lesser Himalaya, near Pipalkoti and Joshimath.

Figure 2 .
Figure 2. Permanent GPS sites for crustal deformation monitoring due to earthquake processes in the Garhwal-Kumaun and Spiti regions.(a) Sites showing slow sliding motion along the slope (NAKO, RATH, BHTW, GUPT, JOSH) are shown by red rectangles with their site names within a red rectangle.Site names with black rectangles are the reference sites (LARI, BHAT, HRMN, AULI, RUDR) used to derive the sliding motion at the slow moving GPS sites.Sites JOSH and RUDR are operated by Survey of India.(b)-(e) show terrain models of the GPS sites (a �0.25 � �0.25 � region) experiencing slow motion.STD-Southern Tibet detachment, MCT-Main Central Thrust, MBT-Main boundary Thrust, MFT-Main frontal Thrust.

Figure 3 .
Figure 3. Temporal variation in the three components of GPS sites at BHTW (from 2006 to 2013) and RATH (during 2012-2021) in reference to a neighbouring site (BHAT, Figure 2) which is not affected by the landslide.

Figure 4 .
Figure 4. Temporal variation in the three components of GPS site at GUPT during 2012-2022 in reference to a neighbouring site (HRMN, Figure 2) which is not affected by the landslide.

Figure 5 .
Figure 5. Temporal variation in the three components of GPS site at NAKO during 2018-2020 in reference to a neighbouring site LARI (Figure 2) which is not affected by the landslide.

Figure 6 .
Figure 6.Temporal variation of coordinates in the three components of the GPS site at JOSH during December 2021 to April 8, 2023, in reference to a neighbouring site AULI (Figure 2) which is not affected by the landslide.Note the two episodes of change in sliding rate (3-20 November 2022 and December 25-January 10, 2023) marked by the two strips.The average monthly rainfall (in mm) at Joshimath is shown with the histograms.(b) and (c) shows the results of kinematic data processing every 30 s with reference to a nearby site RUDR (Figure 2) during the two episodes of transition periods of anomalous change in sliding motion.In each case, the running average is shown in red colour.

Figure 7 .
Figure 7. Seasonal variations of N, E and up coordinates of RATH, GUPT and NAKO GPS sites.The variations are consistent with such variations at other stable GPS sites in the Himalayan region(Yadav et al. 2019;Mondal et al. 2022) and also with that due to hydrological and atmospheric loading(Gautam et al. 2017), shown here at each site (by continuous curves), thus implying that these variations are not unusual and the sliding movement remains unaffected by the seasonal changes.Because of discontinuous data during 2015-2018 at GUPT, that period is omitted.Histogram in each panel show monthly rainfall with the scale on the right-side Y axis.