Remote sensing of ice albedo using harmonized Landsat and Sentinel 2 datasets: validation

Shunan Feng; Joseph Mitchell Cook; Yukihiko Onuma; Kathrin Naegeli; Wenxia Tan; Alexandre Magno Anesio; Liane G Benning; Martyn Tranter

doi:10.1080/01431161.2023.2291000

Remote sensing of ice albedo using harmonized Landsat and Sentinel 2 datasets: validation

Int J Remote Sens. 2023 Dec 26;45(19-20):7724-7752. doi: 10.1080/01431161.2023.2291000. eCollection 2024.

Authors

Shunan Feng¹, Joseph Mitchell Cook¹, Yukihiko Onuma², Kathrin Naegeli³, Wenxia Tan^{4

5}, Alexandre Magno Anesio¹, Liane G Benning^{6

7}, Martyn Tranter¹

Affiliations

¹ Department of Environmental Sciences, Aarhus University, Roskilde, Denmark.
² Earth Observation Research Center, Japan Aerospace Exploration Agency (JAXA), Tsukuba, Japan.
³ Department of Geography, Remote Sensing Laboratories, University of Zurich, Zurich, Switzerland.
⁴ Key Laboratory for Geographical Process Analysis & Simulation of Hubei Province, College of Urban and Environmental Sciences, Central China Normal University, Wuhan, China.
⁵ State Key Laboratory of Geodesy and Earth's Dynamics, Innovation Academy for Precision Measurement Science and Technology, CAS, Wuhan, China.
⁶ GFZ German Research Centre for Geosciences, Section Interface Geochemistry, Telegrafenberg, Germany.
⁷ Department of Earth Sciences, Free University of Berlin, Berlin, Germany.

Abstract

Albedo plays a key role in regulating the absorption of solar radiation within ice surfaces and hence strongly regulates the production of meltwater. A combination of Landsat and Sentinel 2 data provides the longest continuous medium resolution (10-30 m) earth surface observatory records. An albedo product (harmonized satellite albedo, hereafter HSA) has already been developed and validated for the Greenland Ice Sheet (GrIS), using harmonized Landsat 4-8 and Sentinel 2 datasets. In this paper, the HSA was validated for various Arctic and alpine glaciers and ice caps using in situ measurements. We determine the optimal spatial window size in point-to-pixel analysis, the best practices in evaluating remote sensing algorithms with groundtruth data, and cross sensor comparison of the Landsat 9 (L9) and Landsat 8 (L8) data. The impact of the spatial window size on measured ice surface homogeneity and albedo validation was analysed at both local and regional scales. Homogeneity statistics calculated from the grey-level co-occurrence matrix (GLCM) suggest that the ice surface becomes more homogeneous as the image resolution becomes coarser. The optimal spatial window size was found to be 90 m, based on maximizing the statistical and graphical measures while minimizing the root mean square error and bias. HSAs generally agree closely with in situ albedo measurements (e.g. Pearson's R ranges from 0.68 to 0.92) across various Arctic and alpine glaciers and ice caps. Cross sensor differences between L9 and L8 are minor, and we suggest that no harmonization is necessary to add L9 to our HSA product.

Keywords: Google Earth Engine; Ice albedo; arctic and alpine; data harmonization; spatial window size; validation.

Grants and funding

This publication is part of the Deep Purple Project. The project receives funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program under grant agreement No. 856416. Kathrin Naegeli is supported by the ESA PRODEX Trishna T-SEC project under Grant PEA C4000133711, Yukihiko Onuma is supported by JSPS KAKENHI under Grant Number JP20K19955, and Wenxia Tan is funded by the State Key Laboratory of Geodesy and Earth’s Dynamics under Granter Number SKLGED2023-5-3.