Analyzing hydrological dynamics for water balance estimation of landslide dammed lakes in Pakistan

Landslide-dammed lakes are formed by natural blockages of river channels. These lakes pose significant hydrological risks downstream, especially under changing climate conditions. Monitoring the surface area extent and modelling the hydrology of such lakes is important to assess the stability of landslide dams and the downstream flood risk. In this study we aim to link observed lake water balances changes with the driving hydrological processes in the upstream catchment.

The present study focuses on two lakes in Pakistan, formed by landslide dams, both still standing many years after their formation: Attabad Lake (formed in 2010 by a rockfall triggered by rainfall) and Zalzal Lake (formed in 2005 by a landslide triggered by earthquake). Over the years, after an initial phase of increasing trend and large fluctuations, both lakes have seen a consistent decline in area and volume, apart from some remaining seasonal fluctuations. Remote sensing images from Landsat 5, 7, and 8 were integrated to determine lake surface area based on the Normalized Difference Water Index (NDWI). Data Gap filling techniques were applied to estimate missing months with cloudy images. Digital elevation models (DEM) prior to lake formation were used to derive volume over time for the two lakes.

The estimated variations of lake volumes were subsequently modelled based on the water balance of the upstream catchments. We considered precipitation, snowfall, snow accumulation, snowmelt, ice melt, springs, and groundwater recharge. Hydrometeorological data (including precipitation, snowfall, snowmelt, temperature, runoff, and actual evapotranspiration) was collected from various sources (GRACE, TerraClimate, ERA5-Land) by utilizing Google Earth Engine. Groundwater recharge was calculated by analyzing variations in terrestrial water storage collected from GRACE data for both lakes. Additionally, we used lumped hydrological models (such as the Budyko framework) to quantify the interplay between climatic inputs and hydrological fluxes.

We conclude that using hydrological models helps understand the role of hydrological processes in lake inflows, outflows and storage changes. This approach facilitates the assessment of the sensitivity of lake hydrology to changes in climatic variables. The analysis showed seasonal variations in lake inflow and outflows driven by snowmelt, and precipitation. This study will contribute to the assessment of the hydrology of landslide-dammed lakes in data scarce catchments.

Keywords: Landslide dams, hydrological modeling, water balance, climate change, remote sensing