Barbagallo et al. (2025) Integrating Satellite and Field Data for Glacier Melt Modeling in High-Mountain Asia: A Case Study on Passu Glacier

Identification

• Journal: Remote Sensing
• Year: 2025
• Date: 2025-12-02
• Authors: Blanka Barbagallo, Davide Fugazza, Guglielmina Diolaiuti, Antonella Senese
• DOI: 10.3390/rs17233907

Research Groups

• Department of Environmental Science and Policy, Università degli Studi di Milano, 20133 Milan, Italy

Short Summary

This study developed an integrated remote sensing and ground-based approach to model bare ice melt on Passu Glacier, High-Mountain Asia, accurately estimating a total melt volume of 16 million cubic meters water equivalent with a 9% uncertainty against field measurements.

Objective

• To develop and validate a physically based, accurate, and scalable enhanced T-index glacier melt model for bare ice on Passu Glacier, High-Mountain Asia, by integrating satellite-derived products (albedo, surface temperature, topography) with minimal ground-based meteorological and glaciological observations, and to identify the dominant topographic drivers of melt.

Study Configuration

• Spatial Scale: Passu Glacier, Hunza Valley, Karakoram, Pakistan. The glacier covers an area of 53.47 km² with a length of approximately 38 km and an elevation range from 2686 m to 7638 m above sea level. Modeling was performed at a 30 m spatial resolution.
• Temporal Scale: The primary study period for melt modeling and validation was the ablation season from 5 August to 13 October 2023. Meteorological data from 2022 was used for baseline forcing.

Methodology and Data

• Models used:
  • Enhanced T-index melt model (based on Pellicciotti et al. [15] and Senese et al. [16]), explicitly incorporating net shortwave and longwave radiation.
  • Air temperature distribution: Mean tropospheric lapse rate of –6.5 °C/km applied to daily mean air temperature.
  • Incoming shortwave radiation (SWin-point): Multi-step approach by [29], accounting for topography (slope, aspect, shading), geographic parameters, astronomical factors, and atmospheric transmissivity.
  • Surface albedo (αpoint): Narrowband-to-broadband conversion algorithm by Liang [35].
  • Incoming longwave radiation (LWin): Estimated using air temperature and cloudiness from AWS, following [29,38], with atmospheric emissivity derived from Konzelmann et al. [40].
  • Outgoing longwave radiation (LWout-point): Stefan–Boltzmann law, with surface temperature estimated using the empirical relationship by Oerlemans [38] (constrained to ≤ 0 °C).
  • Model validation: Leave-one-out cross-validation (LOOCV) for ice melt, and statistical metrics (bias error, mean absolute error, root mean square error, bias-removed RMSE, R-Squared) for meteorological variables.
  • Melt driver analysis: Standardized multiple linear regression.
• Data sources:
  • Ground-based:
    • Meteorological data: Hourly measurements from two Automatic Weather Stations (AWSs): Passu AWS (2945 m a.s.l.) measuring air temperature, relative humidity, atmospheric pressure, wind speed/direction, incoming solar radiation; Borith Lake AWS (2680 m a.s.l.) with a complete set of WMO-standard sensors including a net radiometer.
    • Glaciological observations: Ablation measurements from four bamboo stakes (12 m length) installed in the ablation zone of Passu Glacier (2916–2932 m a.s.l.) from 5 August to 13 October 2023.
  • Satellite:
    • Glacier perimeter: "New Pakistan Glacier Inventory" [20].
    • Digital Surface Model (DSM): ALOS/PRISM AW3D30 (30 m resolution) from JAXA, used for topographic parameters (slope, aspect, shading) and spatial distribution of air temperature and solar radiation.
    • Surface albedo: Landsat 8 OLI (30 m resolution) surface reflectance data, processed via Google Earth Engine (GEE). Nine usable scenes from August to October 2023.
    • Surface temperature (for outgoing longwave radiation validation): Landsat 9 TIRS (100 m spatial resolution, resampled to 30 m) Band 10 (Collection 2 Level 2). Three scenes from August to October 2023.

Main Results

• Total Ice Melt: Passu Glacier experienced a total ice melt volume of approximately 16 million cubic meters water equivalent (m³ w.e.) during the monitoring period (5 August to 13 October 2023).
• Average Melt Depth: The median cumulative melt for melting grid cells was 3.60 m w.e., with values ranging from 0.00 m w.e. to a maximum of 14.11 m w.e.
• Model Performance: The melt model accurately reproduced observed ablation stake measurements with an uncertainty of 0.48 m w.e., corresponding to 9% of the average observed melt.
• Dominant Melt Drivers: Elevation was identified as the primary topographic factor controlling ice melt (standardized regression coefficient β = –0.501, unique R² = 0.199), with south-facing aspect exerting a secondary positive influence (β = +0.164, unique R² = 0.022). Slope had a small negative effect (β = –0.101, unique R² = 0.008), while east-facing aspect was negligible.
• Albedo Influence: Satellite-derived albedo variability significantly influenced melt estimates, and the use of multi-date satellite imagery for albedo improved the accuracy of ice melt simulations.
• Meteorological Variable Validation:
  • Modeled air temperature showed good agreement with Borith Lake AWS data (Bias Error (BE) = +0.63 °C, Root Mean Square Error (RMSE) = +0.99 °C, R² = 0.97).
  • Modeled incoming shortwave radiation showed acceptable performance (BE = −96.54 W m⁻², RMSE = 102.32 W m⁻², R² = 0.81).
  • Modeled incoming longwave radiation had moderate agreement (BE = −49.69 W m⁻², RMSE = 72.61 W m⁻², R² = 0.45).
  • Modeled outgoing longwave radiation showed good agreement with satellite-derived values (BE = −2.67 W m⁻², RMSE = 20.98 W m⁻², R² = 0.48).

Contributions

• Developed an integrated, physically consistent, and scalable framework for glacier melt estimation in remote, data-scarce regions of High-Mountain Asia (Third Pole/Karakoram) by combining minimal but strategically located field data with multi-date satellite products.
• Achieved a significant reduction in ice melt estimation error (from 17% to 9%) compared to previous studies in the region, by incorporating more physically based approaches for incoming shortwave radiation, spatially distributed multi-date satellite-derived albedo, and including all key radiative components (shortwave and longwave).
• Demonstrated the reliability and scalability of a simplified melt model for bare ice melt in high-altitude, limited-accessibility, debris-free conditions.
• Quantified the dominant influence of elevation on ice melt distribution, with secondary effects from aspect and slope.
• Provides a transferable methodology that supports effective glacier monitoring, hydrological forecasting, and climate adaptation strategies in the Karakoram, with relevance for Glacial Lake Outburst Flood (GLOF) risk management and water resource planning.

Funding

• "Glaciers and Students" project (project number: 00144462)
• "Water for Development (W4D)" project (Award ID: 1274884)
• Funded by the Ministry of Foreign Affairs and International Cooperation and the Italian Agency for Development Cooperation (AICS).
• Executed by the United Nations Development Programme (UNDP).
• Implemented by EvK2CNR.

Citation

@article{Barbagallo2025Integrating,
  author = {Barbagallo, Blanka and Fugazza, Davide and Diolaiuti, Guglielmina and Senese, Antonella},
  title = {Integrating Satellite and Field Data for Glacier Melt Modeling in High-Mountain Asia: A Case Study on Passu Glacier},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs17233907},
  url = {https://doi.org/10.3390/rs17233907}
}