Effect of DOC on evaporation from small Wisconsin lakes

•DOC enhanced surface water temperature, but did not explain differences in E.•High DOC was related to low E and high radiant energy outflux from lakes.•Wind speed was a weak correlate of E, but it was nonetheless an important driver.•Floating pan evaporation exhibited a strong diel cycle in all lakes.•Pan estimates and modeled E agreed reasonably well over daily time scales.

SummaryEvaporation (E) dominates the loss of water from many small lakes, and the balance between precipitation and evaporation (P-E) often governs water levels. In this study, evaporation rates were estimated for three small Wisconsin lakes over several years using 30-min data from floating evaporation pans (E-pans). Measured E was then compared to the output of mass transfer models driven by local conditions over daily time scales. The three lakes were chosen to span a range of dissolved organic carbon (DOC) concentrations (3-20 mg L−1), a solute that imparts a dark, tea-stain color which absorbs solar energy and limits light penetration. Since the lakes were otherwise similar, we hypothesized that a DOC-mediated increase in surface water temperature would translate directly to higher rates of evaporation thereby informing climate response models. Our results confirmed a DOC effect on surface water temperature, but that effect did not translate to enhanced evaporation. Instead the opposite was observed: evaporation rates decreased as DOC increased. Ancillary data and prior studies suggest two explanatory mechanisms: (1) disproportionately greater radiant energy outflux from high DOC lakes, and (2) the combined effect of wind speed (W) and the vapor pressure gradient (es − ez), whose product [W(es − ez)] was lowest on the high DOC lake, despite very low wind speeds (<1.5 m s−1) and steep forested uplands surrounding all three lakes. Agreement between measured (E-pan) and modeled evaporation rates was reasonably good, based on linear regression results (r2: 0.6-0.7; slope: 0.5-0.7, for the best model). Rankings based on E were similar whether determined by measured or modeled criteria (high DOC < low DOC). Across the 3 lakes and 4 years, E averaged ∼3 mm d−1 (C.V. 9%), but statistically significant differences between lakes resulted in substantial differences in cumulative E that were consistent from year to year. Daily water budgets for these lakes show that inputs were dominated by P and outputs by E; and our findings indicate that subtle changes in the variables that drive E can have measurable effects on water levels by shifting the balance between P and E.