Changes in the Global Hydrological Cycle: Lessons from Modeling Lake Levels at the Last Glacial Maximum

Geologic evidence shows that lake levels in currently arid regions were higher and lakes in currently wet regions were lower during the Last Glacial Maximum (LGM). Current hypotheses used to explain these lake level changes include the thermodynamic hypothesis, in which decreased tropospheric water vapor coupled with patterns of convergence and divergence caused dry areas to become more wet and vice versa, the dynamic hypothesis, in which shifts in the jet stream and Inter-Tropical Convergence Zone (ITCZ) altered precipitation patterns, and the evaporation hypothesis, in which lake expansions are attributed to reduced evaporation in a colder climate. This modeling study uses the output of four climate models participating in phase 2 of the Paleoclimate Modeling Intercomparison Project (PMIP2) as input into a lake energy-balance model, in order to test the accuracy of the models and understand the causes of lake level changes. We model five lakes which include the Great Basin lakes, USA; Lake Petén Itzá, Guatemala; Lake Caçó, northern Brazil; Lake Tauca (Titicaca), Bolivia and Peru; and Lake Cari-Laufquen, Argentina. These lakes create a transect through the drylands of North America through the tropics and to the drylands of South America. The models accurately recreate LGM conditions in 14 out of 20 simulations, with the Great Basin lakes being the most robust and Lake Caçó being the least robust, due to model biases in portraying the ITCZ over South America. An analysis of the atmospheric moisture budget from one of the climate models shows that thermodynamic processes contribute most significantly to precipitation changes over the Great Basin, while dynamic processes are most significant for the other lakes. Lake Cari-Laufquen shows a lake expansion that is most likely attributed to reduced evaporation rather than changes in regional precipitation, suggesting that lake levels alone may not be the best indicator of how much precipitation this region receives. Our results indicate that the causes of hydrologic fluctuations are spatially diverse and that future projections will need to consider more than just thermodynamic changes for accurate regional predictions.