Empirical and Process Models for Acidification and Alkalinity Regulation in Lakes of the Upper Great Lakes Region

Abstract

A large data base was assembled from surveys conducted by several federal and state agencies on approximately 1500 inland lakes in the Upper Great Lakes Region (UGLR) -- northern Minnesota and Wisconsin and Upper Michigan. Data were scrutinized carefully by a variety of quality control procedures, and outliers were eliminated. The quality-assured data base was used to characterize lakes in the three state region according to parameters related to the potential sensitivity of lakes to acidification (e.g alkalinity, pH, conductivity, hydrologic type). A trend of increasing numbers of acidic and very low alkalinity lakes across the region (from west to east) correlates with a similar trend in increasing acidity of precipitation. Drainage lakes are the dominant hydrologic lake type in northern Minnesota, but seepage lakes are most common in northern Wisconsin and Upper Michigan. Most of the lakes in the data base (78%) have organic color levels below 50 chloroplatinate units and are classified as "clearwater" systems. Correction of ion balances for contributions of organic anions was unnecessary for these lakes but was useful in improving ion balances of more highly colored lakes. A factor, CF, defined as the ratio of the average chloride concentration in a lake to the average chloride concentration in precipitation, was used as a surrogate measure for hydrologic data on evaporative concentration. Half the lakes had CF values between 2.3 and 4.4 (mean:3.8). A sulfate enrichment factor (SEF, defined as the sulfate/chloride ratio in a lake divided by the analogous ratio in precipitation) was used to determine whether or not sulfate behaves conservatively in lakes. SEF > 1 indicates the occurrence of terrestrial sources of sulfate in a lake's watershed; SEF < I indicates net loss of sulfate in a watershed or lake (presumably by sulfate reduction), if chloride is assumed to behave conservatively. Only 43t of the lakes exhibited nearly conservative behavior (0.75 < SEF < 1.25), and 40% of the lakes showed evidence of significant sulfate sinks. Two measures of acidification were defined for each lake: change in sulfate (ASO4 ) and change in alkalinity (AAlk), both parameters being the difference between measured (present) lake values and background (pristine levels), which were estimated for each lake by a variety of methods. Several lines of evidence suggest that background sulfate levels in regional precipitation (bSO4p) were between 10 and 20 u.eq/L; multiplying these values times CF for a lake gives the lake's bSO4. Good correlations were found between the two measures of lake acidification (ASO. and AAlk) among the UGLR lakes and especially among precipitation-dominated seepage lakes. A separate data base on chemical quality of atmospheric precipitation across the region was obtained from the Minnesota Pollution Control Agency and used with subsets of the lake data base to explore acid-loading lake response relationships. Significant relationships were found between ASO4. and precipitation acidity (H+p ) but not between AAlk and H+n for all lakes in the region. A weighted regression procedure showed that- H+p had significant relationships with boch ASO4 and AAlk for seepage lakes,however, and these relationships were used to develop estimates of acid loading criteria designed to prevent the acidificacion of the most acid-sensitive lakes in each state. The critical H+p loading value estimated this way for Minnesota lakes is about 11 kg/ha-yr. A model to predict in-lake alkalinity generation (IAG) was developed based on CSTR (continuous-flow stirred reactor) kinetics. The model describes budgets for each ion involved in alkalinity regulation by a differential equation that includes terms for inputs and outputs and a first-order source/sink term. The equations are linked to an alkalinity balance equation that includes inputs, outputs, IAG by sulfate and nitrate reduction, and internal alkalinity consumption by ammonium assimilation. Calibration of the model was accomplished using ion budget data obtained from studies on 14 softwater lakes in diverse geographic areas. Rate coefficients generally are similar among softwater lakes: kSO4,, = 0.5 m/yr; kNO3.= 1.3 yr-1; kNH4 + = 1.5 yr.-t. Sensitivity analysis showed that predicted alkalinity is sensitive to water residence times but not very sensitive to moderate changes in rate coefficient values. According to the model, IAG is important in regulating the alkalinity of lakes with water residence times greater than about 2 years. The model reflects the homeostatic nature of IAG: the process increases with increasing inputs of HNO3. or H2SO4 and decreases as loadings of these acids decrease.