Skip to main content
. 2021 Mar 16;12:1688. doi: 10.1038/s41467-021-21971-1

Fig. 3. Long-term and decomposed trends in water temperature.

Fig. 3

a Lake surface water temperature trends based on annual GLSEA data at the thermistor location computed using two methods: simple linear regression (0.40 ± 0.26 °C/decade trend and 95% confidence interval) and a Theil–Sen line (0.41 ± 0.20 °C/decade trend and 95% confidence interval). b The decomposed long-term trend component of the seasonal trend decomposition (STL) method for the GLSEA lake surface temperature, a non-linear counterpart to the linear regression methods (trend from linear fit is given in Table 1). c Seasonal trend component from the STL for the GLSEA lake surface temperature. d Residual component from the STL for the GLSEA lake surface temperature. e Subsurface water temperature trends at 110 m depth based on a linear regression (0.05 ± 0.06 °C/decade trend and 95% confidence interval) and Theil–Sen line (0.04 ± 0.04 °C/decade trend and 95% confidence interval). f Long-term trend computed from the STL for the 110 m depth (trend from linear fit is given in Table 1). g Seasonal trend component from the STL for the 110 m depth. h Residual component from the STL for the 110 m depth.