Remote sensing of changes in lake-calving glacier area and dynamics

Abstract

The combination of glacial retreat and climate change are increasing glacial lakes globally. Satellite remote sensing provides a way to accurately monitor glacial lake changes through automated methods. In this study, we developed a new Object-Based Image Analysis (OBIA) approach in Google Earth Engine (GEE) by undertaking 48 classification experiments in the Southern Alps of New Zealand to investigate the impact of classifier parameters, input features and training data on classification accuracy, and four additional experiments to identify the optimal combination of simple non-iterative clustering segmentation parameters for Landsat OBIA classification. Our results show that the best-performing parameters was 100 random trees, 250 training points per class, 4 for connectivity and 128 for neighborhoodSize. Using random point sampling, the accuracy for OBIA generated class outlines were evaluated by comparing 100 random points per class to a manual classification. This resulted in an overall accuracy of between 91.8% and 96.8% for New Zealand and Svalbard.

We applied this approach to 14 lake-terminating glaciers in the Southern Alps of New Zealand and discovered that the combined glacial lake area increased by 136.4 ± 4.8%, from 9.7 ± 0.21 km² to 22.9 ± 0.49 km² between 1990-2024. We hypothesise a combination or different climatic, topographic, geometric, and glaciological factors influenced individual lake area changes. In Svalbard, we identified 217 individual lake-terminating glaciers between 19852024. The total glacial lake area increased from 59.2 ± 1.4 km2 to 200 ± 4.3 km2, an increase of 255.9 ± 4.1%. By 2024, 136 lake-terminating glaciers remained observable. This demonstrates that our methodology is a powerful tool for efficient and accurate glacial lake mapping in different glaciated regions. We emphasise the importance of monitoring these glacial lakes in the coming century, as many glaciers exhibit early stages of proglacial lake growth, which may speed up ice-thinning and glacier retreat.

Thesis is embargoed until 31st December 2027

Date of Award	Dec 2025
Original language	English
Supervisor	Frank Lyons (Supervisor), Brian Irvine (Supervisor) & Brian Bridges (Supervisor)