Correlations of Climate and Streamflow in Three Minnesota Streams

Correlations between four climate parameters and streamflow in three Minnesota streams were investigated. Runoff values measured over periods of up to 37 years were correlated with precipitation, air temperature, wind, and dew point temperature. The overall objective was to examine if relationships can be obtained which require only readily available input parameters without calibration. Such relationships would be of great use, e.g. to compute future lake water budgets without recourse to more detailed and complex hydrologic runoff models. Monthly, seasonal, and annual time frames were investigated. A seasonal time frame using 3 month averages gave the closest fit for the linear regressions without time lag. Although the watershed sizes varied from 360 to 49,600 square kilometers, the 3 month period seemed sufficiently long to average long term hydrologic processes such as infiltration, evaporation, and groundwater flow. An equation was found for each season (3 months) for each of the rivers. Winter (December, January, February) regressions required only precipitation data; spring regressions required air temperature and precipitation; summer and fall regressions were found with precipitation, air temperature, dew point temperature, and wind speed. The coefficients in the regression equations were related to the watershed characteristics. The r² values were highest for the Zumbro River in spring (0.69) and lowest for the Baptism River in winter (0.14). Root mean square error values ranged from 2.8 mm/mo for the Mississippi River in winter to 18 mm/mo for the Baptism River in spring. The coefficients of variability (CV) ranged from 0.24 to 0.52. Overall the results were disappointing but not all bad. Climate parameters without watershed parameters can characterize runoff only within limits. To project possible future runoff averages the GISS GCM-values for the 2 × CO₂ climate scenario were applied to the seasonal runoff regression equations. The projections were that the spring runoff values would decrease by up to 35% while in the other seasons streamflows would increase by up to 50%. Annual runoff would not change significantly enough to be predictable. The results were in the range of changes predicted by other investigations using very different techniques. Since predictions were based on equations found with past records, it was implied that the land cover would remain unchanged in the 2 × CO₂ environment. This may be unrealistic and needs further investigation.