Satellite altimeter observed surface water increase across lake-rich regions of the Arctic

Main text

On our planet, the Arctic region has numerous and widely distributed lakes. These lakes are fundamental components of regional water cycles, carbon cycles, and the energy balance and also play a crucial role in providing habitats for wildlife, maintaining biodiversity, and supplying freshwater. According to satellite observations, in situ measurements, and climate models, the Arctic is the fastest-warming region on Earth. However, up to date, it is still unclear how lakes in the Arctic have responded to the rapid warming climate over the past decades. This suggests a strong need for detecting temporal changes in lakes across the Arctic in order to better monitor regional water resource dynamics, understand the potential drivers, and evaluate ecological and socio-economic influence.

Recently, Webb et al.¹ investigated surface water area changes across lake-rich regions in the Arctic during the period 2000–2021 by detecting trends in the superfine water index (SWI) based on 500-m moderate resolution imaging spectroradiometer (MODIS) satellite imagery. Their results indicate that the net change in SWI was negative, and 82% of permafrost regions exhibited a decreasing trend in water areas.¹ However, Olthof et al.² argued Webb et al.’s study and concluded that results from Webb et al.¹ that show widespread Arctic surface water declines may not be reliable due to properties of the MODIS SWI index, including its coarse spatial resolution to depict the surface water extent, its sensitivity to vegetation changes and water turbidity, and a relatively weak relationship with higher-resolution surface water cover. Considering these potential limitations in Webb et al.’s study, more robust evidence for observing surface water dynamics in the Arctic is urgently required.

Here, we try to offer a new insight into satellite altimeters for tracking surface water dynamics across lake-rich regions of the Arctic, which significantly differ from the surface water extent perspective determined by optical satellite imagery in Webb et al.’s study and Olthof et al.’s study.^1,2 Satellite altimeters can provide a powerful tool to measure water levels for characterizing surface water dynamics, which can well avoid the inherent spectral-related issues in multispectral satellite imagery. Here, we detected water-level changes in lakes derived from satellite altimeters and further estimated the water storage dynamics. Satellite altimeter sensors can provide water-level observations in lakes, which can also be used to estimate water storage changes in lakes by combining water area changes derived from multi-temporal satellite imagery. For a lake, if its water level or water storage increases, then it can be defined as a surface water increase; otherwise, it can be defined as a surface water decline.

Specifically, we conducted the following steps to quantify surface water storage dynamics across lake-rich regions of the Arctic to further support our conclusions. First, we collected the global lake storage dynamics (GloLakes) database,³ providing water storages during 1984–2020 for more than 27,000 lakes globally derived from optical satellite imagery and satellite altimeter data, which have been validated by in situ measurements (R = 0.91) for 494 lakes in Australia, southern Africa, India, Spain, and the USA, and lake water levels were validated by in situ measurements (R = 0.98) for 160 lakes in Canada and Australia.

Second, for each lake in Webb et al.’s study, we obtained the yearly averaged water storages over 2000–2020 from the GloLakes database. Then, we applied the linear regression model to estimate the long-term trend and its significance (“statistically significant” refers to p < 0.1 unless stated otherwise). Third, we summarized the lake water storage dynamics within the study area. We calculated the proportion of lakes with declining and increasing water storages separately and calculated the proportions of significant decline and increase. For all lakes within the study area, we obtained the total water storages for each year over the study period and calculated the trend and its significance, then conducted a similar analysis for lakes with water storage decline and water storage increase separately. Moreover, we compared our results on surface water storage dynamics across the Arctic with Webb et al.’s results. We produced Figure 1 based on the above calculations. We also used the Mann-Kendall test for a comparison.

Figure 1.

Surface water increase across lake-rich regions of the Arctic

(A) Spatial pattern in surface water storage dynamics across lake-rich regions of the Arctic over 2000–2020.

(B) Annual surface water storages in lakes and the trend regression lines.

(C) Surface water storage trends in lakes.

(D) Number of lakes.

(B)–(D) share the same legend, and in (D), gray dashed boxes represent lakes with significant water storage trends.

We mapped the spatial pattern of water storage trends in the study area (Figure 1A). Lakes in North America show a relative balance of water storage increase and decrease, while lakes in the Eurasian continent show a significant water storage increase. As shown in Figures 1B and 1C, during 2000–2020, the lake water storages exhibit an increasing trend of 131.90 km³/year, and lakes with water storage increase and decline exhibit trends of 413.69 and −218.79 km³/year. For 5,689 lakes in the study area, 54.39% (3094) of them experienced increasing water storage trends during 2000–2020 (significant: 20.85%, p < 0.1; 16.68%, p < 0.05; 9.60%, p < 0.01) (Figure 1D), while only 45.61% (2,595) experienced declining water storage trends (significant: 10.23%, p < 0.1; 6.56%, p < 0.05; 2.94%, p < 0.01) (Figure 1D). Additionally, those expanded lakes had a total water storage increase of 8,687.45 km³ (6,338.45 km³ for significantly expanded lakes), and those shrunken lakes had a total water storage decline of 5,917.64 km³ (3,581.24 km³ for significantly shrunken lakes).

Additionally, according the Mann-Kendall test, the water storage trends in all lakes and lakes with water storage increase/decline are 200.56, 385.00, and −215.78 km³/year, respectively. Across the study area, 3,397 lakes (59.71%; significant: 22.10%) exhibit increasing water storage trends, contributing to the 8,085.19 km³ water storage increase (significant: 5,116.04 km³), while only 2,292 lakes (40.29%; significant: 8.60%) exhibit increasing/declining water storage trends, contributing to the 4,531.38 km³ water storage increase (significant: 2,461.84 km³). Both the linear regression model and the Mann-Kendall test indicated significant surface water increases across lake-rich regions of the Arctic.

We found that surface water across lake-rich regions of the Arctic experienced an increasing trend since 2000, contrasting Webb et al.’s conclusions.¹ Our results are based on surface water storages derived from 30-m Landsat imagery and the high-accuracy ICESat-2 altimeter, while Webb et al.’s results are based on coarse 500-m MODIS imagery. Thus, we inferred that, as a spectral index, the MODIS SWI can be easily influenced by vegetation and water turbidity, failing to accurately reflect surface water dynamics.⁴ Furthermore, we found that Webb et al.’s study lacked in situ measurement validation to ensure the robustness.

In summary, we provide evidence of surface water increases across lake-rich regions of the Arctic from satellite altimeter observations, strongly challenging the conclusion in Webb et al. The satellite altimeter is a robust tool for characterizing surface water dynamics by measuring lake water levels with good accuracy, which can well avoid the inherent spectral issues in MODIS imagery and provide a more accurate indicator for revealing lake expansion and shrinkage. However, satellite altimeters may also exhibit some disadvantages since they can only provide water levels along trajectories, which may omit small water bodies without satellite altimeter data. Also, lakes with dramatic seasonal changes may exhibit large uncertainties due to limited revisit frequency.

In the future, high-resolution satellite imagery and in situ measurements for validation need to be collected. Remote sensing data have been widely used in detecting Earth’s surface changes, and some problematic conclusions may be derived if we did not consider hidden uncertainties in remote sensing data and related algorithms. Considering the characteristics of different satellite data sources, we encourage the fusion of multi-source satellite data for better observations.⁵ We highlight the necessity to validate satellite observations with finer-scale measurements or in situ data to ensure the robustness of the findings. Moreover, we can integrate satellite altimeter observations and hydrological modeling to predict climate-driven lake-level changes in the Arctic region under different emission scenarios.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (42101343), the Second Tibetan Plateau Scientific Expedition and Research Program (2019QZKK0206), and the Natural Science Foundation of Jiangsu Province (BK20240258).

Declaration of interests

The authors declare no competing interests.

Published Online: October 11, 2024