Retrieval of lake water surface albedo from Sentinel-2 remote sensing imagery

Of the many factors that affect heat flux, water surface albedo can affect the heat flux and energy balance at the water-atmosphere interface. As a result, lake water surface albedo can act on key climate-change factors through its impact on lake-atmosphere heat exchange. Most of the available global water cycle models, however, still use value specific water surface albedos and often produce estimation with relative errors of 5-15 % due to time and location differences. The lake albedo inversion model based on Sentinel-2 MSI sensor was obtained in this research. From the perspective of energy transmission, this research selects an inversion method of converting narrowband albedo to broadband albedo, and using Sentinel-2A spatial resolution image for the study area for 10 m lakes albedo product. Based on the measured albedo data, using the determination coefficient (R2), mean relative error(MRE) and root mean square error (RMSE) for validation, the result of the inversion results show that the R2 is 0.62, MRE is 0.25, RMSE is 0.039. The values of lake albedo in the study area ranged from 0 to 0.45, with an average value of 0.14. The high value of lake albedo appears in the south of the Songnen Plain lake area, with a size between 0.28 and 0.45, and the low-value gathering area appears in the Qiqihar Lake Group. The central lake area of Jilin also has a low value, with a size between 0 and 0.10. From the west to the middle of Jilin Province, the albedo value tends to decrease gradually, and some areas have higher values. In a typical lake, the average albedo is arranged as follows: Yangsha lake > Shitoukoumen reservoir > Chagan lake > Moon lake > Erlong lake > Xinlicheng reservoir.

Automated summary

Water surface albedo can affect the heat flux and energy balance at the water-atmosphere interface, thereby impacting key climate-change factors. Existing global water cycle models often use specific values for water surface albedo, leading to estimation errors of 5-15% due to time and location differences. This research developed a lake albedo inversion model based on the Sentinel-2 MSI sensor, using a method to convert narrowband albedo to broadband albedo.