2:30 PM - 2:45 PM
[ACC26-04] Statistical improvement of thermal resistance for debris-covered glaciers by means of whole available Landsat-8 scenes
Keywords:Himalaya, Glacier, Thermal resistance, Debris cover
In high-mountain Asia, supra-glacial debris cover introduces significant uncertainty in the melt rates; thinner layers promote melting, and thicker layers provide insulation. Debris-covered termini with melt ponds and ice cliffs accelerate the melt rates. Previous studies have utilized satellite observations to estimate debris thickness and thermal conductivity by calculating thermal resistance (TR). The remotely-sensed TR also enables spatial analysis for inaccessible glaciers, whereas criticisms have pointed out a linear temperature gradient assumption in the debris layer, leading to underestimations of the TR. This study, therefore, aims to improve TR by statistical approach, focusing on input energy budgets. We focus on the calculation process of TR (TR= Ts/G) and assess how surface temperature (Ts) corresponds to ground heat fluxes (G) among different dates.
Our target is Batal Glacier in the Chandra basin, western Himalaya [32.34°N, 77.58°E; 4250-5800 m a.s.l.]. Using all available Landsat-8 images from March 2013 to January 2024 (Path: 147; Row: 37/38; total 237 scenes) and corresponding hourly data from ERA5-Land reanalysis datasets, surface radiative balance is calculated for each moment. Considering the elevation difference within one grid, the downward longwave radiation is downscaled from ~11 km to 30 m, employing a digital surface model, AW3D30. TR is then determined from G and Ts at 86 locations where in-situ measurements of debris thickness were carried out. Cloud-covered and freezing pixels are removed. The entire process is conducted on an online cloud platform for geospatial analysis called Google Earth Engine, which facilitates high-performance analysis accessible to all.
As a result, by plotting the relationship between Ts and G for up to 237 dates at each point on the debris-covered surface, positive correlations between Ts and G are identified. Larger G and Ts are observed during warmer months, whereas most samples in the winter are excluded due to their freezing conditions. Among all in-situ points, the gradient of the linear approximation of Ts against G has a significant positive correlation with in-situ measured debris thickness. This study thus proposes the gradient of the linear approximation of Ts against G as an advanced TR.
Where the debris layer is thinner (~5cm), the distribution of Ts (versus G) is relatively smaller and tends to become linear. This feature suggests that sub-debris ice efficiently absorbs heat, making it difficult for the Ts to rise greatly. On the other hand, where the debris layer is thicker, Ts sharply increases as G increases. In both cases, the linear regression lines do not cross [0, 0], suggesting improvement of the formar TR definition without a constant value (TR = Ts/G).
In this study, the necessity and the possibility of an advanced TR is proposed. Statistical analysis demonstrated in this study will contribute to establishing a more realistic definition of TR under consideration of thermal properties among different debris-depth layers. Furthermore, ice cliffs and supraglacial ponds have more dominant roles in glacier melting than debris thickness. Other indirect approaches, such as terrain analysis using high-precision DEMs, are considered to be more effective for melt-rate estimation.
Our target is Batal Glacier in the Chandra basin, western Himalaya [32.34°N, 77.58°E; 4250-5800 m a.s.l.]. Using all available Landsat-8 images from March 2013 to January 2024 (Path: 147; Row: 37/38; total 237 scenes) and corresponding hourly data from ERA5-Land reanalysis datasets, surface radiative balance is calculated for each moment. Considering the elevation difference within one grid, the downward longwave radiation is downscaled from ~11 km to 30 m, employing a digital surface model, AW3D30. TR is then determined from G and Ts at 86 locations where in-situ measurements of debris thickness were carried out. Cloud-covered and freezing pixels are removed. The entire process is conducted on an online cloud platform for geospatial analysis called Google Earth Engine, which facilitates high-performance analysis accessible to all.
As a result, by plotting the relationship between Ts and G for up to 237 dates at each point on the debris-covered surface, positive correlations between Ts and G are identified. Larger G and Ts are observed during warmer months, whereas most samples in the winter are excluded due to their freezing conditions. Among all in-situ points, the gradient of the linear approximation of Ts against G has a significant positive correlation with in-situ measured debris thickness. This study thus proposes the gradient of the linear approximation of Ts against G as an advanced TR.
Where the debris layer is thinner (~5cm), the distribution of Ts (versus G) is relatively smaller and tends to become linear. This feature suggests that sub-debris ice efficiently absorbs heat, making it difficult for the Ts to rise greatly. On the other hand, where the debris layer is thicker, Ts sharply increases as G increases. In both cases, the linear regression lines do not cross [0, 0], suggesting improvement of the formar TR definition without a constant value (TR = Ts/G).
In this study, the necessity and the possibility of an advanced TR is proposed. Statistical analysis demonstrated in this study will contribute to establishing a more realistic definition of TR under consideration of thermal properties among different debris-depth layers. Furthermore, ice cliffs and supraglacial ponds have more dominant roles in glacier melting than debris thickness. Other indirect approaches, such as terrain analysis using high-precision DEMs, are considered to be more effective for melt-rate estimation.