Although water cycles across North America have fluctuated in response to climate throughout geological time, recent climatic impacts and near-term forecasts are uncertain due largely to short and sparse hydrological records that are also often confounded by land-use or direct human influences, such as impoundment and diversion. To overcome these shortcomings and investigate long-term regional hydrologic dynamics, a collaborative effort among researchers and water resource managers across Wisconsin assembled a unique hydrologic data set of water levels and meteorology in northern Wisconsin and the Laurentian Great Lakes region. This assemblage of data was highly unusual in its length and diversity. Historic but discontinued lake level records for Buffalo Lake, WI were successfully harmonized with overlapping modern records from one of NTL’s core study lakes (Crystal Lake) to build a continuous time series that started in 1942. Further, data for undisturbed small seepage lakes were combined with information on groundwater levels from the NTL program as well as water surface levels for Lakes Superior and Michigan-Huron to evaluate the temporal coherence and climate sensitivity among highly divergent lake systems.
The overlapping time-series for Crystal Lake, Buffalo Lake and associated groundwaters indicate that a coherent, near-decadal oscillation has dominated water levels in the NHLD for most of the past 7 decades (Fig. 1A). The amplitude of oscillation approaches 2m, dwarfing the well-known annual cycles. And surprisingly, despite strong differences among systems, Lakes Superior and Michigan-Huron also experienced this same multi-decadal oscillation, although with a more dampened periodicity (Fig. 1B). These long-term cycles in water level corresponded to net atmospheric water flux (precipitation-evaporation [P-E]) in the region (Fig. 1C). These findings confirm that climatically driven, near-decadal oscillations have dominated the water cycle in this sector of mid-continental North America for most of the last century.
The second notable pattern from this 70-year time series is a distinct decline in water levels in the past decade, reaching the minimum level of record in 2010 for Northern Highland Lakes. Similarly, the water level of Lake Michigan-Huron recently dropped at a rate not seen since the 1930s mega-drought. These declines occurred as precipitation has remained relatively steady, but evaporation has increased, suggesting a strong link to an evolving climate system.
At least three future scenarios seem possible for this mid-continental sector: 1) the historical water cycle may resume in a few years, with the time period 1990-2012 emerging as an aberration in the historical record; 2) the recently altered cycle may propagate through future time as an amplified oscillation around the historical mean water level; or 3) a step-change (or series of step-changes) to new mean water levels may occur. Because of the magnitude of past oscillations, it remains challenging to determine which scenario is most likely. Nonetheless, recognition of these oscillations, and a potential departure from these oscillations, has only be recognizable through the construction of this unique and multifaceted long-term data set.