Liquefaction Risk in Kathmandu: Clear Study Summary Part I

July 8, 2026 | By Thethoughtpole

This first article explains why Rama M. Pokhrel and co‑authors studied liquefaction in Kathmandu Valley and what basic methods they used. Kathmandu sits on soft river and lake deposits above a powerful earthquake fault system, so strong shaking is likely in the future. Yet the 2015 Gorkha earthquake showed less liquefaction than experts expected, which raised doubts about old maps and assumptions. The authors used borehole soil data, earthquake hazard models, and dry vs monsoon groundwater levels to calculate a liquefaction potential index at many sites. These results help show how season, geology, and location control liquefaction risk in different parts of the valley.

For Full Research Paper Visit https://link.springer.com/article/10.1007/s10518-021-01198-7

Liquefaction potential for the Kathmandu Valley, Nepal: a sensitivity study

1. Kathmandu Valley: Setting and Risk

Kathmandu Valley is a basin filled with soft sediments sitting over older hard rock.
The fill is made of river, lake, and delta deposits of sand, silt, and clay.

In the north, the Gokarna Formation is mostly sandy.
In the south, the Kalimati Formation is mainly clayey silt.

The valley is close to major Himalayan faults, including the Main Himalayan Thrust under the region.This geological system can produce very large earthquakes, possibly close to magnitude 9.Big damaging events already happened in 1833, 1866, 1934, 1988, and 2015.

2. What Is Liquefaction and What Happened Before?

Liquefaction occurs when loose, wet sand or silt loses strength during shaking.The soil can behave like a liquid for a short time.

It can cause:

Sand boils
Ground cracks
Tilting or sinking of buildings

In the 1934 Nepal–Bihar earthquake, photos from Tundikhel showed sand boils and fissures in central Kathmandu.In the 2015 Gorkha earthquake, liquefaction signs were seen at Bungmati, Kausaltar, Imadol, and some other sites.But the number of such cases was small compared with early expectations.

This led to a key question:

Were old liquefaction maps and assumptions wrong or too simple?
Were we ignoring important factors like groundwater season?

3. What This Study Tried to Do

The authors set three main goals.

They wanted to:

See how dry vs monsoon groundwater changes liquefaction potential.
Identify which parts of the valley (north vs south, near rivers vs centre) are more at risk.
Test which simple liquefaction index method best matches real 1934 and 2015 evidence.

They also wanted to create better liquefaction maps for planning and loss assessment.

4. Soil and Groundwater Data Used

The study used the SAFER/GEO‑591 database, which collects geotechnical borehole data.
From this, the authors chose 75 deep, good‑quality boreholes with SPT and soil descriptions.

Grain‑size curves from these boreholes mostly fall within known “liquefiable” or “possibly liquefiable” zones from Japanese work (Tsuchida).
This means many valley soils can liquefy if they are saturated and shaking is strong.

To model groundwater, they relied on data from 239 wells in the northern valley.
From these, they defined two simple levels:

Wet (monsoon): water table about 1.6 m below the surface
Dry season: water table about 5.1 m below the surface

These values match observed seasonal differences and long‑term pumping trends.
Earlier studies showed groundwater drawdown of up to 7.5 m between 2000 and 2008, and about 1 m per year in some areas.

5. Earthquake Shaking Levels (PGA)

The authors used probabilistic seismic hazard analysis (PSHA) results for Kathmandu.
They studied two standard probabilities of exceedance:

2% in 50 years (rare, near‑collapse level)
10% in 50 years (more frequent, life‑safety level)

They combined this with two hazard models:

AVERAGE of four ground motion models → about 1.2 g and 0.65 g.
AB03 subduction model → about 0.48 g and 0.3 g, near the current code value of 0.35 g.

These values were used together with soil and groundwater data at each borehole.

6. From Factor of Safety to Liquefaction Index

At each borehole, the authors:

Used the Seed and Idriss simplified method to calculate factor of safety against liquefaction, FL, for each layer.
Then converted all those FL values into a single liquefaction potential index (PL) for the top 20 m.

They tried two PL formulas:

Iwasaki
Sonmez (a modified version)

PL is larger if:

More layers have FL < 1
Liquefiable layers are closer to the ground surface

The Sonmez method also includes “marginal” layers with FL between about 0.95 and 1.2, giving a more detailed scale (no, low, moderate, high, very high).

7. Checking Against Real Liquefaction Evidence

To see which method worked better, they compared PL with real cases.

They collected:

7 locations with observed liquefaction (e.g., Bungmati, Manamaiju, Kausaltar, Imadol, Tundikhel).
3 locations with no liquefaction but with structural damage.

For each case, they identified a nearby borehole (often within a few hundred metres) and compared the computed PL to what was seen in the field.

Results:

Both methods roughly matched the observed pattern.
Sonmez gave slightly higher PL and better separation among classes.

So they chose Sonmez PL to build the final maps.