Paleoseismology is the study of past earthquakes and paleosiesmologists study them to know what might happen in future. They collect clues on the surface along active faults and dig deeper for evidence.

When earthquakes occur, layers of top earth surface including sedimentary rock slip past each other along the fault, with one side dropping down and the other side jutting out as a small cliff. This creates fault scarps.

Jayan Perumal, a paleoseismologist at the Wadia Institute of Geology in Dehradun, first collects evidence of fault scarps’ locations using high-resolution satellite imagery, plugs in coordinates into the GPS, and heads out into the Himalayan foothills for two months every year. An earthmover follows him. 

While seismologists study earthquakes, their behaviour and characteristics mostly based on numerical data through seismometers, paleoseismologists dig across the fault scarp and study the faulted layers along with debris and rock fragments that pile up—called colluviums.

They also look for charcoal and cultural artifacts, and date them through carbon dating or fallout isotopes. It gives them the size and magnitude of past earthquakes, the behaviour of the fault, and preview of what might be in store. Moreover, the study also says how mountains got built.

Perumal graduated from Chennai’s Presidency College and did postgraduation in applied geology at the University of Madras. After completing his Ph.D during 94-97, he worked as mining geologist for five-and-a-half years, and came to the Wadia institute as a scientist in 2002. He got interested in past earthquakes and joined the American teams working then.

Edited excerpts from an interview with Fountain Ink:


Why is paleosiesomology important?

 The Himalayan frontal fault system is the longest and fastest moving continental convergence system, with the potential for producing devastating, large-magnitude earthquakes.

The Indo-Gangetic fertile plains lie along the HFT (Himalayan Frontal Thrust) at the base of the sub-Himalaya, where large populations live and farm. Establishing the timing of earthquakes along the HFT is critical in assessing regional seismic hazards for these areas near the fault that have both large populations and potentially inadequate infrastructures.

However, determining the seismic hazards in this region has been difficult due to uncertainties in basic fault parameters, including the locations and rupture lengths of historical earthquakes, earthquake recurrence intervals, segmentation, and co-seismic slip. Each of these parameters can be established through paleoseismic investigation, and can then be used to generate predictive, probabilistic models for both the magnitude and likelihood of future earthquakes.


What is a ‘Great Quake’.

 Magnitude 8 or more. Major quake is 7 to 7.9


How do you go about searching for past earthquakes?

 We search fault scarp along the Himalayan frontal fault system that lies in foothills of Sub-Himalaya, where plain and mountain front meets. Fault scarp is the fault-generated landforms and they’re sensitive indicators of the style and timing of tectonic/earthquake activity. Various types of fault scarps exist along the active fault system, and they tell us the near-surface deformation and duration of faulting.

In the Himalayas, along the foothills there are several fault scarps that look like a baby mountains in front of the Sub-Himalayan frontal fault system that grew like a hill… and then mountain due to repeated earthquake activity and thus mountain building process.

We use high resolution satellite imagery to detect the fault scarp. It is an escarpment, somewhat linear in trend and having same strike to that of regional Himalayan structural grains and it will be perpendicular to the major or minor river direction.

We process the satellite imagery data to produce a bare-earth model in the form of digital elevation model and then look for the scarp, and also we construct series of topo profiles across the fault scarp to check if there are any deformed features such as back tilting etc., and then we choose the representative GPS points along this fault scarp in the lab and plug those waypoints (latitude/longitude) into a handheld GPS and then we navigate to reach the site to examine whether it is a fault scarp.

Does evidence in fault scarp survive through time? What do you look for to establish indeed it’s a fault scarp?

 Care should be taken as sometime at the mouth of the small to major rivers the river flow direction can swing parallel for a while along the strike of mountain front. This process can either destroy the fault scarp and it pushes the scarp (i.e. scarp retreating) toward the mountain side leaving a pseudo-site for paleoseismology investigation. Sometime it can form a typical scarp but if you place a trench you may not able to see the fault that we refer to as fluvial scarp.


After you confirm fault scarp, what are the next steps?

Upon confirmation of the fault scarp, we place a trench perpendicular to the strike of fault scarp (20 m length, 5 m width and 5-7 m depth). The strike of fault scarp generally parallels the Himalayan structural grains. The dimension is variable and it depends upon the occurrence of fault and its tip associated folds in the exposure. Having opened a trench, we then clean the wall and a  1mx1m grid will be deployed on the exposure of the trench wall using nail and string. Strings are leveled using a string level. Using the graph sheet we make a hand-log. It is nothing but transforming all information into a graph sheet. Additionally we use robotic total station (electronic distance meter to map the trench wall at the same level). And also we take photos of each 1m grid to produce a photo-mosaic log. Then we collect radiocarbon samples and sediment samples from the deformed units as well as in the underform units of the trench exposure to  know  the timing of an earthquake event.


What does surface faulting signify?

If the tip of the fault reaches the surface it indicates that the accumulated stored energy has been released. If it is blind—that is to say it does not cut the top surface instead it folds the surface and  produces a fold scarp—it may be said that stored energy is not fully released, but in the case of blind fault the full strain can also be consumed in the form of folding. However, general consensus is that surface faulting or fault reached to the surface is indicator that accumulated energy is total or partially released.


Could you elaborate on how the rupture travels, given that it starts down below the earth?

A Rupture—earth surface breaks suddenly due to earthquake—travels from the hypocenter of the earthquake. That is the origin of the earthquake. But so far we don’t have good understanding on the starting point of the earth’s breaking the top most part of the crust at a large scale that is an earthquake.

In general, it’s not clearly known how the ongoing convergence breaks the  surface in the form of an earthquake.

However, based on case studies of strong earthquakes it has been suggested that usually the rupture propagates from the region of high gradients of stresses toward the crust characterised by low effective compression.

The hypocenter is generally at a depth of 17-20 km, but a deep focus earthquake will reach more than this. The rupture breaks the locked portion of the brittle crust materials where strain was already accumulated during the inter-seismic period. A simple example is if you throw a coin on a moving car’s windshield, you see how the rupture propagates.


Could you talk about the faults underlying the Himalaya?

We have crustal scale faults that are deep seated and lengthy and they emerged from the main detachement plane. The main detachement is nothing but the underthrusted part of the top of the Indian lithosphere.  (See graphic above)

From North to South are Main Central Thrust (MCT), Main Boundary Thrust (MBT) and Himalayan Frontal Thrust or Main Frontal Thrust (HFT or MFT). All these thrusts merge at a common plane at variable depth know as Main Himalayan Thrust (MHT). Generally MHT is referred to as the top of the underthrusting Indian plate. Above which the materials are accreted (orogen) and deformation takes place above the MHT. This is known as thin-skinned deformation.

 The deformation of the crust takes place above the MHT, but this deformation does not penetrate the MHT hence we say it is thin skinned tectonics. In general, the earthquakes that have happened so far in the Himalaya damage zone is confined to south of the High Himalaya (i.e South of MCT and north of HFT). The locked portion of the down going Indian slab also lies between these two zones. The GPS gradient also increases toward north of HFT.

The GPS velocity reveals the present day accumulation of stress due to ongoing convergence between the Indian and Eurasian plates. The velocity vectors become larger toward the inner part of the Himalayas between the sub and Higher Himalaya. This portion is referred to as the locking width or locking area. If the locking area is large then the earthquake will be great. In Himalayas locking width is approximately 80 km at some places. It is more than 100 km (in the Jammu and Kashmir Himalayan segment).


How far down do quakes in the Himalayas emanate from?

In general, at 17-20 km depth. Deep focus earthquakes are seen on the either of the Syntaixal bends of the Himalayas (eastern and western Syntaxis).  In general, Himalaya arc is bow shaped towards the plains but at both ends of it, the curvature is towards the mountain side. Like a hair pin on either side of a bow-shaped mountain.


What signs, marks, changes do the past earthquakes leave behind?

It leaves the fault scarps that are the permanent features on the earth surfaces along the active mountain fronts. Since the Himalayan region is monsoonal terrain, the fault scarp generated by the earthquake activity could be modified or eroded depending upon its size. Surface process can erode or can be removed off these scarps by erosion.

Besides this, it changes river flow direction, and in the case of historical monuments it leaves shear related deformational features (Archeoseismology).


If historical records are not available, what’s the way forward to finding past quakes?

If fault scarp is removed off by erosion then we have to look for disjointed different levels of terraces that was were cut off by the active fault or preserved along the mouth of the river exit at the termination of the mountain front.


Could you give your findings on past earthquakes? When they happened, where they happened, each one’s magnitude, what each one left behind, how many people killed and damage.

In NW Himalaya between Bhatpur (Punjab Plains) and Ramnagar (near Kashipur in Kumaon region) the 1344  earthquake occurred. The scarp of this quake is seen along the mountain front for a length of about 300 km— hence we infer a greater than eight magnitude. Its epicentre lies  somewhere north of Kumaon Himalayas. It damaged several 10th and 11th century’s monuments and temples. Even Qtub Minar in New Delhi was damaged by this event, but historians relate that damage to lightening. But I believe the lightening could be seen in the great earthquake. The historical documents are available in Nepalese Himalayas. 

Surface faulting of the 2005 Kashmir earthquake was seen along already mapped active faults in Muzaffarabad.

 In the Garhwal region, 1803 earthquake damaged ancient monuments. It  occurred near the Badrinath temple. It has even damaged monuments in Delhi.

In the Nepal Himalayas, the 1255 earthquake has been discovered in a trench study and this matches with historical accounts. In this quake, the king of Nepal was killed.

In the Eastern Himalaya, historically reported earthquakes have so far not been discovered in trench studies except the controversial (dispute on date) 1100/1255 earthquake.

Recently, we indentified the 1950 surface faulting in Pasighat.


Any issues or differences among scientists about the numbers of past quakes. Why?

Yes, controversies exist.

For example, in NW Himalaya some researchers suggested a great earthquake occurred between 1200 and 1700 CE and have related it to the 1505 earthquake. Others suggested that the 1505 earthquake is greater than it is believed to be.

Had it been a great event it would have damaged monuments such as the Qutb Minar. But they were not damaged, hence 1505 was not related to a big quake. Further, the 1803 earthquake damaged those monuments. The quake in 1505 didn’t reach to the mountain front and was of a small magnitude.

Some have suggested twin earthquakes of 1255 and 1344 occurred at closer intervals. There are some interpretations about the 1255 quake: one suggested that the rupture propagated toward east of Nepal and not to the west of Indian side i.e. Ramnagar, where, the other interpretation placed it. Still others suggested the 1255 earthquake reached the Indian side up to Nameri, which was questioned.

 So the timing of earthquakes is controversial in the eastern Himalayas and also the style of rupture of 1255 is controversial. The controversy in timing is due to dating limitations. To overcome this issue multi-radiometric dating should be deployed.


What are the underlying forces that make the quake hit the same zone or place again.

Yes it will happen again. Due to ongoing convergence between Indian and Eurasian plates.


Earthquakes release the strain; most of them may not release all of the strain. Does it make the area more prone to another earthquake, and does it remain quiet. What happens then to the fault that causes an earthquake? How does the fault behave before and after one?

In general the earthquake does not release all the stored energy. Yes that will make it prone. But in the Himalayas, the recurrent interval is not known well. Some empirical studies say no characterised time interval exists for earthquakes. When an earthquake happenes,  generally it triggers the adjoining region, sometimes it may advance or postpone the next event. To understand the timing, recurrent interval, parameters of faul needs more research along the Himalayan arc. Such studies are at a nascent stage.


How to connect the past earthquakes to the present scenario. Could you give some examples?

The presence of fault scarp reveals the earthquake happened along this fault. For example, the 2005 Kashmir earthquake occurred along the already-mapped active faults at Muzaffarabad. So a study of active faults (characterisation using paeloseismolgy) has great potential to generate input parameters for seismic hazard assessment of the region.


Have you found legends, stories, or mythology of the past earthquakes passed down the generations of people, and are they helpful to you.

Yes, several stories have been heard from old people at in the sites where we have done excavation, but we don’t find any sound ground truths.


Average repeat time of the quakes in the Himalaya?

No firm results are available. Most of the trenches we excavated did not show more than one event. This may be due to insufficient depth. We need a couple of mega trenches along the HFT to provide recurrent intervals for Himalayan earthquakes. Only two events were reported in Muzaffaradbad, and Kala Amb trench sites where excavations reveal evidence of two events. These results suggest an interval of about 2,000 years. In Nepal, it has been said 650 years using paleoseismology. Using GPS data it has been inferred as 250 yers to 500 years, in general. In NW Himalaya it is speculated to be 1500-3000 years.

In my opinion every magnitude has different recurrent intervals, it would be site specific and you can’t generalise as it is very complicated.  For instantce, a great earthquake will have different recurrent intervals. Uttarkashi 1991 and Chamoli 1993 earthquakes occurred within a two-year interval, these earthquakes are  six plus  in magnitude. So each magnitude will have different time.


Please talk about hazards map, or hazard potential of the Himalayan arc you’re working on.

Hazard maps are based on occurrence of historical and new earthquakes; it is not based on deterministic data sets. Geological Survey of India has a hazard map based on qualitative data set. For making hazard maps we need several inputs such as active fault mapping, characterisation of active fault that includes fault length, fault slip rate, rupture length, magnitude of earthquake, and repeat time. These inputs are useful for making deterministic seismic hazard assessments. Other data such as soil characterisation or landform liquefaction potential index, site amplification and so on…. These data forms  enable us to proudce a seismic hazard map of an area. In the Himalayas, the paleoseismology has come in late in the 19 century and so far we have rudimentary data on how often a great earthquake can come, if so when, where, it is not clear. What is the real rupture length of the Himalayan earthquake? The general consensus is that the Himalayan arc is prone  to get earthquake due to interplate seismicity.

We can’t take intra-plate earthquakes lightly. Intra-plate earthquakes such as Bhuj, Jabalpur earthquakes are also equally important, but their recurrence is slower compared to Himalayan earthquakes.


Which fault really concerns you? Why? Could you describe how a quake could travel up that fault.

In my opinion all faults that lie south of the Main central thrust or south of  High Himalaya (where there is a sudden topographic rise) to the main frontal thrust in the south  are major concerns as the locking width of the Himalayan wedge is nearly 80 km wide. 

Of these faults the studies done so far suggest that the Himalayan frontal thrust can host a great earthquake as we see fault scarp along the Himalayan frontal thrust. North of MFT and South of MCT can host medium to large size earthquakes (e.g. Uttarkashi and Chamoli, 2005 Muzaffarabad earthquake and 2015-Gorkha earthquake of Nepal)

Generally in the Himalayas the earthquake occurs at upper flat or lower flat or ramp of the MHT and some occur (1934  Bihar-Nepal earthquake) in the accretionary wedge itself (e.g. 1905 Kangra or 2005 Kashmir earthquake).


Why do you think it’s so likely to pop off.

Due to continuous ongoing convergence of Indian plate underneath the Eurasian plate. Owing to this process during the inter-seismic state the energy will be accumulated in the locked portion of the brittle crust.  The locked portion of brittle crust along its strike, meaning all along the length of the Himalaya, varies from 80 to 130 km in width. It is just like a spring you compress, during the interseismic strain and release it during the earthquake.

If this comes to pass, how many people would be affected?

The population of fertile plains is huge. The Kashmir earthquake (Mw 7.6)struck north of the plain in the mountain ranges and killed 75,000-100,000 people. The Nepal Gorkha earthquake (Mw 7.8) which was confined to the hinterland of the mountain range and much north of Bihar Plains, killed about 9000. The Indo-Gangetic alluvial and Brahmaputra plains are heavily populated so the death toll will be enormous and will be much higher than the 2005 Kashmir earthquake.