📄 earth science

SWOT-Based 3D Mapping of an Ice-Dammed Lake Outburst Flood: A Case Study of Merzbacher Lake

This study presents a novel satellite-based framework integrating Sentinel-2 imagery and SWOT observations to reconstruct the three-dimensional evolution of the July 2025 Merzbacher Lake outburst flood, enabling the quantitative estimation of a drained volume of approximately 9.84 million cubic meters and demonstrating a shift from two-dimensional detection to comprehensive stage–area–volume assessment of glacial lake outburst floods.

Original authors: Jiayi Zhang, Jiaqi Yao, Haohao Zhou, Teng Li, Meng Ding, Shuangxiao Luo, Xue Li, Haoxiang Xue, Jimin Chen, Changwei Chen, Hui Lu

Published 2026-10-16

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jiayi Zhang, Jiaqi Yao, Haohao Zhou, Teng Li, Meng Ding, Shuangxiao Luo, Xue Li, Haoxiang Xue, Jimin Chen, Changwei Chen, Hui Lu

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

High in the mountains, where glaciers carve deep valleys, water often gets trapped behind walls of ice. These ice-dammed lakes are beautiful but dangerous. When the ice wall fails, sometimes due to warming temperatures or shifting ice, the trapped water can burst out in a sudden, violent flood. Scientists have long known that these glacial lake outburst floods are among the most destructive hazards in high-mountain regions, threatening towns, roads, and power stations far downstream. However, measuring exactly how much water is released during such an event has been a persistent challenge. Traditional satellite cameras can see the lake from above, showing how the water's surface shrinks as it drains, but they cannot see how deep the water is or how much the water level drops. Without knowing the depth change, it is impossible to calculate the total volume of water that has escaped, leaving a critical gap in our understanding of these disasters.

A team of researchers has now bridged this gap by combining two different types of satellite data to create a three-dimensional picture of a massive flood. They focused on Merzbacher Lake, a famous ice-dammed lake in the central Tian Shan mountains, which experienced a sudden drainage event in July 2025. By pairing high-resolution optical images that show the lake's surface area with new radar measurements that reveal the water's height, the scientists were able to reconstruct the entire event in three dimensions. They found that the lake, which covered an area of 1.560 square kilometers before the flood, nearly vanished, leaving behind only tiny pools of water covering just 0.008 square kilometers. More importantly, the new radar data showed that the water level dropped by 12.55 meters. By combining the loss of surface area with this significant drop in height, the researchers calculated that the lake released a massive volume of water, estimated at roughly 9.84 million cubic meters.

The study relied on two distinct satellite systems working in tandem. One system, which uses standard optical cameras, provided a clear view of the lake's boundaries before and after the flood, allowing the team to map exactly how much of the surface disappeared. The other system, a specialized radar mission capable of measuring water height over vast areas, provided the missing vertical dimension. This radar data showed that the water level did not just recede at the edges; the entire lake surface sank by more than 12 meters. This vertical measurement was crucial because it transformed a simple map of a shrinking lake into a calculation of lost water volume. The researchers noted that the remaining water was concentrated near the outlet where the lake meets the glacier, suggesting that the drainage happened through a specific channel at the base of the ice dam rather than the entire lake simply evaporating or seeping away slowly.

This approach marks a significant shift in how scientists assess these hazards. Previously, estimates of flood volume often relied on guesses about the shape of the lake bottom or assumptions about how the water level might have changed. Now, with the ability to measure both the width and the height of the water loss directly from space, researchers can quantify the magnitude of these floods with much greater precision. The study confirms that the July 2025 event was a rapid, basin-scale drainage where the lake emptied almost completely in a matter of days. While the exact path the water took through the ice remains hidden from view, the satellite data provides a clear, observation-based record of the water loss. This method offers a practical way to monitor dangerous lakes in remote, glacier-covered regions where no human observers or ground sensors exist, providing vital information for predicting and preparing for future floods in a warming climate.

Technical Summary: SWOT-Based 3D Mapping of an Ice-Dammed Lake Outburst Flood

Problem Statement
Glacial lake outburst floods (GLOFs) represent a critical hazard in high-mountain regions, yet quantifying the magnitude of individual events remains challenging. Traditional satellite-based assessments rely primarily on optical imagery to map changes in lake surface area. However, these two-dimensional observations lack direct measurements of water-level drawdown. Consequently, estimates of released water volume often depend on empirical area–volume scaling, assumed basin geometries, or hydrodynamic back-calculation, all of which introduce significant uncertainties due to unknown bathymetry and poorly constrained post-event water levels. This limitation is particularly acute for ice-dammed lakes, where drainage is governed by transient englacial or subglacial conduits and rapidly changing outlet-control thresholds, rather than stable moraine spillways.

Methodology
To address the gap in vertical measurement, the authors developed a satellite-based framework integrating optical and radar altimetry data to reconstruct the three-dimensional evolution of a GLOF. The study focuses on the July 2025 drainage of Merzbacher Lake in the central Tian Shan.

The methodology combines two primary data sources:

Sentinel-2 MSI Imagery: Used to map the horizontal extent of the lake before and after the event. Water boundaries were delineated using the Normalized Difference Water Index (NDWI) combined with visual refinement to account for shadows and debris. This provided the pre-event lake area ( $A_{pre}$ ) and the post-event residual water area ( $A_{post}$ ).
SWOT (Surface Water and Ocean Topography) Observations: Used to retrieve water-surface elevation (WSE) changes. The study utilized L2_HR_Raster products (250 m resolution) to determine representative lake levels ( $H_t$ ) via the median of quality-screened pixels. Additionally, L2_HR_PIXC (pixel cloud) data were analyzed to observe spatial reorganization of valid radar returns, distinguishing between continuous lake surfaces and fragmented residual pools.

The drained volume ( $\Delta V$ ) was estimated using a first-order average-end-area approximation, as detailed bathymetry was unavailable:
$\Delta V \approx \frac{A_{pre} + A_{post}}{2} \times \Delta H$
where $\Delta H$ is the water-level drop derived from SWOT. Uncertainty was propagated from the mapped area uncertainties (estimated by perturbing shorelines by ±10 m) and the WSE uncertainty (conservatively estimated at ±3.0 m for post-event residual pools).

Key Results
The study successfully reconstructed the 3D drainage event of Merzbacher Lake between July 5–7, 2025:

Horizontal Collapse: Sentinel-2 observations revealed a drastic reduction in lake area from 1.560 km² to 0.008 km², representing a loss of 1.552 km² (>99% reduction).
Vertical Drawdown: SWOT measurements indicated a concurrent water-level decline from 3270.38 m to 3257.83 m, corresponding to a drawdown of 12.55 m.
Volume Estimation: Integrating the horizontal and vertical constraints yielded an estimated drained-water volume of $(9.84 \pm 2.4) \times 10^6 \, \text{m}^3$ .
Drainage Geometry: Post-event SWOT PIXC data and residual water distribution showed that remaining pools were concentrated near the glacier-contact outlet sector. This suggests the drainage was controlled by a specific outlet zone, likely associated with the ice dam, rather than a uniform basin emptying.

Significance and Claims
The paper claims that this approach enables GLOF assessment to move beyond simple two-dimensional lake-change detection toward quantitative stage–area–volume reconstruction. By combining Sentinel-2's horizontal constraints with SWOT's vertical constraints, the framework provides a practical method for estimating flood magnitude in remote, glacierized regions where field observations are unavailable.

The authors emphasize that their result is a "first-order, satellite-constrained estimate of source-lake storage loss" rather than a complete downstream flood-volume reconstruction. They note that their estimate of $\approx 9.84 \times 10^6 \, \text{m}^3$ differs from official downstream flood volume reports (approx. $6.0 \times 10^7 \, \text{m}^3$ ) due to differences in control volume, flood routing, and background meltwater input. However, the study validates the timing and hydrological significance of the event, demonstrating that SWOT-based 3D mapping offers a new opportunity for assessing cryosphere-related hazards under a warming climate without relying on uncertain empirical scaling laws.

More like this