Meteorological controls of gas exchange at a small English lake

Abstract

The relationships between gas transfer velocity, k600, wind speed, wind direction, rainfall, and relative humidity were examined using measurements of SF6 evasion from Coatenhill Reservoir, a small (0.017 km2), shallow (1.9 ± 0.1 m), man-made lake in northeast England characterized by predominantly low to intermediate wind speeds ∼1-10 m s-1. A graphical method was used to estimate the wind speed at a standard height of 10 m, U10, from wind speed measurements at 2 m and 3.8 m. Derived values of U10 normalized to remove thermal stability effects (U10-n) were significantly correlated with k600 (R2 = 0.71, p < 0.001). A detailed analysis of surface roughness lengths estimated with this procedure showed the k600 versus U10-n relationship to be sensitive to wind speed modification by the meteorological mast and sensors. With normalization to remove these effects, the correlation between k600 and U10-n substantially improved (R2 = 0.86, p < 0.001). In contrast to previous laboratory findings, relative humidity was not significantly correlated with k600 and rainfall rate (Rn) was only weakly correlated with k600, possibly as a consequence of the effects of these variables being largely masked by the time scales of data averaging and SF6 sampling. Similarly, k600 was not correlated with wind direction (i.e., fetch). An empirical gas exchange model accounted for 88% of the total variance in k600 (P = 0.01) at Coatenhill, with U10-n and Rn accounting for 86% and 2%, respectively. Future field investigations of the meteorological controls of k600 will require careful experimental design to allow for more detailed sampling than hitherto has been possible.