10:45 AM - 12:15 PM
[ACC25-P02] Debris thickness estimation by unsupervised classification of time-series pattern of surface temperature
Keywords:Thermal imagery, Temporal change, glacier, debris
In general, glacier surface debris thickness largely affects ablation on debris-covered glaciers. Thin debris lower albedo and it strengthens solar heat absorption and ablation. In the case of thick debris, the insulation effect by the debris layer exceeds albedo lowering effect and it suppresses ablation. In this context, debris thickness distribution has a key role for evaluating mass balance of debris-covered glaciers.
Various proxies (ex. Thermal resistance) are proposed to estimate glacier debris thickness by remote sensing. Such proxies derived by remote sensing frequently have large variability and it makes it difficult to evaluate debris thickness. Though, specific consideration about thermal variation property on various surface conditions is essential.
In this study, we do not focus on remote sensing analysis as a snapshot but focus on temporal change patterns. And we have performed unsupervised classification analysis for the temporal change patterns as a proxy of debris thickness.
We selected Khumbu Himal basin in Nepal and Chandra basin in India as study areas. Landsat 8 level 2 product (surface temperature) is used as trial analysis at first. Temporal change pattern of surface temperature is extracted by each grid (30 m x 30 m), and unsupervised classification (KMeans) has been performed with 40 clusters. The result shows a clear boundary between thick debris-covered ice and non-glacier areas. And gradual classification from lower reach to upper reach on debris-covered areas is also clearly shown. We will perform more specific classification only on debris-covered areas and compare field based debris-thickness data on Chandra basin, India.
Various proxies (ex. Thermal resistance) are proposed to estimate glacier debris thickness by remote sensing. Such proxies derived by remote sensing frequently have large variability and it makes it difficult to evaluate debris thickness. Though, specific consideration about thermal variation property on various surface conditions is essential.
In this study, we do not focus on remote sensing analysis as a snapshot but focus on temporal change patterns. And we have performed unsupervised classification analysis for the temporal change patterns as a proxy of debris thickness.
We selected Khumbu Himal basin in Nepal and Chandra basin in India as study areas. Landsat 8 level 2 product (surface temperature) is used as trial analysis at first. Temporal change pattern of surface temperature is extracted by each grid (30 m x 30 m), and unsupervised classification (KMeans) has been performed with 40 clusters. The result shows a clear boundary between thick debris-covered ice and non-glacier areas. And gradual classification from lower reach to upper reach on debris-covered areas is also clearly shown. We will perform more specific classification only on debris-covered areas and compare field based debris-thickness data on Chandra basin, India.