View in EDS

A global analysis of ice phenology for 3702 lakes and 1028 reservoirs across the Northern Hemisphere using Sentinel-2 imagery.

Saved in:
Bibliographic Details
Title: A global analysis of ice phenology for 3702 lakes and 1028 reservoirs across the Northern Hemisphere using Sentinel-2 imagery.
Authors: Domart, Doris1 (AUTHOR) doris.domart@gmail.com, Nadeau, Daniel F.1 (AUTHOR), Thiboult, Antoine1 (AUTHOR), Anctil, François1 (AUTHOR), Ghobrial, Tadros1 (AUTHOR), Prairie, Yves T.2 (AUTHOR), Bédard-Therrien, Alexis1 (AUTHOR), Tremblay, Alain3 (AUTHOR)
Source: Cold Regions Science & Technology. Nov2024, Vol. 227, pN.PAG-N.PAG. 1p.
Subjects: Ice on rivers, lakes, etc., Random forest algorithms, Bodies of water, Latitude, Atmospheric temperature
Abstract: As existing global lake ice studies have predominantly focused on medium to large lakes, and reservoir ice studies have been limited to regional scales, very few studies of ice phenology have combined both lakes and reservoirs of different sizes. This study aims to characterize the freeze-up and break-up dates of 3702 lakes and 1028 reservoirs from 1 to 31,000 km2 across the Northern Hemisphere, and to analyze spatial patterns and relationships between ice phenological dates and driving factors. The freeze-up and break-up dates of these water bodies were retrieved from Sentinel-2 imagery using an ice detection algorithm through the Google Earth Engine platform from 2019 to 2023. The algorithm was verified by comparing phenology dates with an independent database based on observations from passive microwave sensors, with a mean absolute error of 18 days for both freeze-up and break-up dates. This newly established ice phenology database along with various geographic, morphometric, and climatic characteristics of the water bodies, was used to develop a random forest model for predicting ice phenology dates. While the predictive model performance is at a fair level (mean absolute error of 12 days for both freeze-up and break-up), challenges were encountered in certain high-elevation areas where cloudy conditions as well as black ice resulted in delayed freeze-up dates. Among the variables included in the random forest model, latitude and accumulation of freezing degree days were identified as the main drivers of ice phenology dates. Despite the challenges of applying a single, straightforward method on a global scale, this study has allowed the creation of a vast and comprehensive database of lake and reservoir freeze-up and break-up dates that can be used by the community to further analyze ice patterns. • Freeze-up and break-up dates were retrieved from 2019 to 2023 from Sentinel-2 imagery for 3702 lakes and 1028 reservoirs in the Northern Hemisphere. • Morphometric and climatic variables were used in a random forest model, with a 12-day mean absolute error for freeze-up and break-up predictions. • Key drivers for ice phenology dates were identified: latitude and freezing degree days related variables significantly influence ice phenology dates. • Break-up dates were more strongly correlated with air temperatures than freeze-up dates, which were also influenced by latitude and mean depth. [ABSTRACT FROM AUTHOR]
Copyright of Cold Regions Science & Technology is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:As existing global lake ice studies have predominantly focused on medium to large lakes, and reservoir ice studies have been limited to regional scales, very few studies of ice phenology have combined both lakes and reservoirs of different sizes. This study aims to characterize the freeze-up and break-up dates of 3702 lakes and 1028 reservoirs from 1 to 31,000 km2 across the Northern Hemisphere, and to analyze spatial patterns and relationships between ice phenological dates and driving factors. The freeze-up and break-up dates of these water bodies were retrieved from Sentinel-2 imagery using an ice detection algorithm through the Google Earth Engine platform from 2019 to 2023. The algorithm was verified by comparing phenology dates with an independent database based on observations from passive microwave sensors, with a mean absolute error of 18 days for both freeze-up and break-up dates. This newly established ice phenology database along with various geographic, morphometric, and climatic characteristics of the water bodies, was used to develop a random forest model for predicting ice phenology dates. While the predictive model performance is at a fair level (mean absolute error of 12 days for both freeze-up and break-up), challenges were encountered in certain high-elevation areas where cloudy conditions as well as black ice resulted in delayed freeze-up dates. Among the variables included in the random forest model, latitude and accumulation of freezing degree days were identified as the main drivers of ice phenology dates. Despite the challenges of applying a single, straightforward method on a global scale, this study has allowed the creation of a vast and comprehensive database of lake and reservoir freeze-up and break-up dates that can be used by the community to further analyze ice patterns. • Freeze-up and break-up dates were retrieved from 2019 to 2023 from Sentinel-2 imagery for 3702 lakes and 1028 reservoirs in the Northern Hemisphere. • Morphometric and climatic variables were used in a random forest model, with a 12-day mean absolute error for freeze-up and break-up predictions. • Key drivers for ice phenology dates were identified: latitude and freezing degree days related variables significantly influence ice phenology dates. • Break-up dates were more strongly correlated with air temperatures than freeze-up dates, which were also influenced by latitude and mean depth. [ABSTRACT FROM AUTHOR]
ISSN:0165232X
DOI:10.1016/j.coldregions.2024.104294