Fig 1.
Map of Lake Tanganyika and its position in East Africa.
White circles with black outlines indicate locations of in situ temperature measurement. The size of the circle is proportional to the number of temperature measurements taken at each location. Most of the in situ data come from areas near major research centers (Uvira, Kigoma, and Mpulungu). The orange circle indicates the location of the TEX86 sediment core and the red square indicates the location of the satellite data extraction site. The north, central, and south basins are located between 3.4–5.8°S, 5.8–7.0°S, and 7.0–8.9°S, respectively.
Fig 2.
Modeled century-long warming rate estimates (1912–2013).
Each colored pixel on the maps is an estimate of the warming rate for that location in the lake based on the temperature model fit to all available in situ temperature data. Separate map panels show variation across seasons and depths in the estimated warming rate. The top row of warming estimate maps are for the dry season and the bottom row of warming rates are for the wet season. The columns indicate the depth gradient of warming rate estimates from 0–100 m. All temperature models were fit to data from more than 1.5 km from land and in locations where the water depth exceeded 100 m deep.
Fig 3.
Satellite temperature and TEX86 temperature as a function of modeled in situ temperature.
The black dashed line represents the 1:1 reference line. Small red square dots represent daily satellite temperatures as a function of the modeled in situ estimate at the satellite extraction site. Large red square dots with black outlines represent annual mean satellite temperatures as a function of modeled annual mean in situ temperatures at the extraction site. Annual mean satellite temperatures were calculated from raw satellite data linearly interpolated to daily timescales. Large orange circular dots with black outlines represent the TEX86 measurements as a function of the modelled annual mean temperature at the core site.
Fig 4.
Surface warming rates based on in situ, satellite, and TEX86 data.
Each bar represents a warming rate estimate based on one of the three temperature measurement methods (in situ, satellite, TEX86). Error bars represent 95% confidence intervals for each estimate. The two leftmost bars are warming rate estimates over the period from 1918–1996 based on TEX86 data and the in situ model output for the location and timeframe of the TEX86 core. The two rightmost bars show warming rate estimates over the period from 1985–2011 based on satellites and in situ model output for the location and time frame of the satellite data extraction.
Fig 5.
Long-term satellite temperature data compared to model estimates.
Each small red square dot is a raw, satellite measurement of the lake’s surface. The grey line is the modeled surface temperature at the location of the satellite extraction. Satellite temperatures closely track the seasonal, interannual, and long-term variation in temperature data predicted by in situ temperature data.
Fig 6.
Long-term TEX86 temperature data compared to model estimates.
Each large orange circular dot is a raw TEX86 surface temperature measurement. Each TEX86 measurement represents the annual mean surface temperature at the location of the TEX86 core site ± 0.4°C (95% confidence interval). The three lines are the modeled annual surface temperature at the location of the TEX86 core site at three different depths (0, 50, 100 m). There is a strong disagreement between TEX86 and modeled in situ temperatures, especially early in the time series. Deviations between the TEX86 temperature measurements and the modeled surface temperature may reflect model error, error in the TEX86 calibration procedure, or long-term shifts in the depth range of Thaumarchaeota (the microbes whose biomolecules are involved with generating the TEX86 proxy).