The effect of 20th-century drought on groundwater and lakes in a climatically sensitive area of the Northern Great Plains (Grant County, western Minnesota) was investigated by analysis of lake sediments and historical air photos, in conjunction with groundwater flow simulations. Drought caused an observed 4.0 to 5.1m lake-head decline between 1923 and 1938; all but three either dried completely or declined to < 1.7 m depth. From the deepest (Elk Lake), a 210Pb-dated sediment core was analyzed for ostracode species distribution and geochemistry of C. Rawsoni shells (δ18O, δ13C, Mg/Ca, Sr/Ca). Mg/Ca and δ13C increased during drought at the same time that salinity-tolerant ostracodes thrived, suggesting increasing salinity. However, δ18O decreased during drought, anti-correlative with Mg/Ca. A numerical model for transient response of groundwater flow to drought, modeled after that of 1923-38, was constrained by drought-end recessional strand lines observed on air photos of one lake, with strand line elevations inferred from bathymetry. Simulated lake-water-level declines in the first 15 drought years were consistent with strand line elevations. The rate of decline was exponential at drought onset, slowing as water in lakes and wetlands receded to below land surface and lake evaporation declined. A near-steady state was attained between 40 and 50 simulation years after onset of drought, with the water table from 1-5 m below the level of dry lakebeds and only the two deepest lakes still holding water. They were, however, greatly constricted in area and depth. Simulated evaporation fluxes were reduced by over 50% within 15 years, despite increased ET rates, due to lake area constriction and water table declines. Both relate to land surface morphology beneath and around lakes. Model results suggest that lake-bed elevations exert control on the rate and ultimate level to which groundwater is depressed by drought. The deepest lakes in any region will tend to become the focus of groundwater flow during sustained drought once lakes around it dry out. The unexpected finding of decreasing shell δ18O in response to drought is interpreted as some combination of source and/or evaporation - reduction effects, such as seasonality of recharge/precipitation or reduced ET flux from wetlands that seep back into lakes.
Bibliographical noteFunding Information:
This research was funded by the National Science Foundation (NSF) Hydrologic Sciences grant EAR-9304741. The stable isotope analyses (Ito) were funded in part by grants from the NSF (BIR-9014277 and EAR-9406183) and the University of Minnesota. John Carney and Joan Puller (KSU), Brian Haskell and Eric Gong (UMN), Coburn Wightman (WVU), and Red Westrum (Elk Lake, MN) helped in the field and lab. This is a Paleoecological Reconstructions of Aridity: Interdisciplinary Research InitiativE (PRAIRIE) contribution, and LRC contribution #580.