Temporal variations in lake water temperature: Paleoenvironmental implications of lake carbonate δ18O and temperature records

δ18O and “clumped isotope” measurements of lacustrine carbonates provide important records of past terrestrial climate and paleoelevation. However, one of the primary challenges interpreting these data is constraining the relationship between mineral formation temperature and seasonal and mean annual climate. We examined surface water temperature records from 88 lakes across the globe to develop transfer functions that relate seasonal water surface temperature to mean annual air temperature. These transfer functions provide a means of reconciling proxy measures of water temperature with annual climatic conditions. Mean annual surface water temperature is related to mean annual air temperature (MAAT=−0.0318×TWater2+2.195×TWater−12.607; R2=0.96) independent of lake size, with consistently higher mean water temperatures than air. N. Hemisphere Spring (April–June) mean lake surface temperatures (MAAT=−0.0097×TWater2+1.379×TWater−8.227; R2=0.94) are only slightly warmer than MAAT for lakes in all climate zones, while summer water temperatures (JJA) (MAAT=−0.0055×TWater2+1.476×TWater−18.915; R2=0.90) may be 10–20 °C warmer than MAAT for cold climate lakes. Comparison of carbonate stable isotopic data from sites with different timescales of carbonate formation, such as found along latitude or elevation gradients, can lead to large over- or underestimates of past temperature. Accurate paleoclimatic interpretation of isotopic proxies or use of temperature proxy constraints (i.e., fossil leaves) to estimate lake temperature and calculate water δ18O thus requires appropriate application of water temperature–air temperature transfer functions, particularly when considering ancient cool, high elevation lake systems.