Sensible heat flux by surface layer scintillometry and eddy covariance over a mixed grassland community as affected by Bowen ratio and MOST formulations for unstable conditions

Measurements of sensible heat flux for an extended period for unstable conditions using surface layer scintillometry (SLS) and eddy covariance (EC) and supplemented by Bowen ratio measurements for a mixed grassland community on the eastern seaboard of South Africa are presented. Measurements of SLS sensible heat flux density were compared with those obtained using EC for a wide range of Bowen ratio (β). Also presented is an analysis of the different forms of the Monin-Obukhov similarity theory (MOST) functions used in micrometeorology and suggested by various authors, done by comparing EC sensible heat and measured by SLS through the use of an iterative determination of Monin-Obukhov parameters. A comparison of the SLS-measured structure parameter of air temperature C_T² corrected for β and uncorrected was carried out, with the results showing good correspondence but with a slight bias indicating that not correcting SLS measurements of C_T² for β would also result in a slight bias in H. Eddy covariance estimates of sensible heat flux density (H_EC), obtained using averaging periods between 1 and 120 min and compared with scintillometer measurements, demonstrated that short-time averaging periods resulted in underestimated H_EC. The EC measurements for 60- and 120-min averaging periods were sometimes inconsistent with SLS measurements. A sensitivity analysis indicates that both the EC and SLS measurements of H are influenced by β. For 0 < β < 0.2, the correction to H_SLS amounts to more than 10% compared to more than 20% for H_EC, although the magnitude of the differences are small. A comparison of H_EC and H_SLS measurements for 0.1 intervals of β between 0 and 4.3 shows reasonable correspondence for β > 1. For 0 < β < 1, the H_SLS (y) versus H_EC (x) scatterplot linear regression slope decreased from 1.25 to close to 1 for β increasing from 0 to 1. A comparison between β-corrected H_EC and H_SLS measurements - the latter computed using various empirical stability functions used by MOST - shows significant differences compared to H_EC ranging from almost 20% overestimation for some methods to 20% underestimation for others. Long-term use of the recommended MOST stability functions for the SLS method is shown to result in reasonable correspondence between SLS and EC sensible heat flux for a wide range in atmospheric conditions, stability, and sensible heat magnitude.