Simulations of Water Quality in Cisco Lakes in Minnesota

This report is the second in a series of three reports that describe model simulations of cisco (tullibee) lakes for the Minnesota Department of Natural Resources. This report describes the adaption, calibration and application of the year-round water quality model MINLAKE2010, used for the year-round simulation of daily water temperature and dissolved oxygen (DO) profiles in 28 lakes in Minnesota. The results of this report will be applied in the third report to make projections of potential (refuge) habitat for cisco, a cold-water fish, in Minnesota lakes under warmer future climate scenarios. Twenty-one cisco lakes and seven non-cisco habitat lakes were selected for the model simulations including adaptation, calibration and application of the model. Lake bathymetry data and measured water quality data (water temperature and dissolved oxygen profiles, Secchi depths, chlorophyll-a concentrations) were assembled and analyzed. Weather data were assembled for the periods 1961 to 2008 from six Class I weather stations, and 1991 to 2008 from three Class II weather stations. In order to more accurately project water quality conditions in cisco lakes, which are typically deep mesotrophic or oligotrophic lakes, modifications and refinements were made in the MINLAKE96 computer program, especially the simulation procedures and utility functions of the model. The model simulations of daily water temperature and DO profiles in the 28 lakes were calibrated using six calibration parameters. After calibration the average standard error of estimate (S.E.) against measured data for all 28 lakes is 1.47 oC for water temperature (range from 0.8 to 2.06 oC) and 1.50 mg/L for DO (range from 0.88 to 2.76 mg/L). The average regression coefficient (R2) is 0.92 for water temperature (range from 0.84 to 0.97) and 0.75 for DO (range from 0.12 to 0.91). Simulation results are presented as profile plots and time-series plots. Individual model calibration parameters were analyzed, and regional values were proposed for cisco lakes without data. The model performance using proposed calibration parameter values was examined.