Lake Level Response to Climate in Minnesota

Abstract

We are interested in the variability of lake levels in Minnesota, and the relationship between lake levels and climate. We analyzed historical water levels in 25 Minnesota lakes. Eight were landlocked lakes and seventeen were flow-through lakes. The data were daily values, but substantial gaps existed. The longest record reached back to 1906 (Lake Minnetonka and Upper Prior Lake in Scott County). We determined statistical parameters such as mean annual lake levels and seasonal variations of the historical lake water levels. Linear regression and Mann-Kendall test were used to evaluate the presence of trends in daily, mean annual, spring (May) and fall (October) water levels. The majority of the 25 lakes showed rising water levels in the last century (1906 to 2007). The strongest upward trend was observed in a landlocked lake (Lake Belle Taine in Hubbard County) where the rate was 0.030 m/yr. The second largest increase was observed in a flow-through lake (Marion Lake in Dakota County) with a rate of 0.024 m/yr. Swan Lake (in Nicollet County) and Swan Lake (in Itasca County) were the only lakes that showed a falling trend with a rate of -0.011 and -0.002 m/yr, respectively. The analysis also showed that lake levels have been increasing in most of the 25 lakes in the last 20-years (1987-2006). One landlocked lake and eight flow-through lakes showed their strongest upward trends in the last 20 years. Five of the eight landlocked lakes and eleven of the seventeen flow-through lakes reached their highest recorded levels after 1990. Upward trends in recorded lake water levels were found in both spring and fall in the majority of the 25 lakes analyzed. We also attempted to understand how Minnesota lake levels have responded to climate changes in the past. Correlation coefficients were calculated between annual lake water levels and mean annual climate variables. The correlation of water levels with precipitation was moderate, and the correlation with dew point and air temperatures was very weak. 48- and 36-month antecedent precipitation was the strongest indicator of average water levels. Multivariate regression analysis of lake levels did not improve the lake level predictions. Numerical indicators for ground water and surface water inand out-flows appear necessary for further improvement. The correlation between mean annual water levels was strongest among lakes in the same climate regions and weakest among lakes in distant climate regions. Lake levels in the same Minnesota climate region (with identical precipitation and temperatures) had correlation coefficients as high as 0.78, while those in distant regions were not correlated. The average correlation coefficients among annual water levels in all lakes were 0.43 for the eight landlocked lakes and 0.41 for the seventeen flowthrough lakes. Overall, the analyses showed that changes have occurred in lake levels in Minnesota in the last century and in the last 20 years. The majority of the lakes have rising lake levels. The correlation between climate parameters and lake levels was weak. The consistency of water level variations in lakes of the same region is perhaps the strongest indicator of a climate effect. If the trends continue, lakes included in this study may experience significant water level increase by 2050.