LAUSR

Abstract Using AmeriFlux data from 74 eddy covariance sites, we present a synthesis of the dependence of thermal roughness length on friction velocity across five land covers types, including deciduous broadleaf forests, evergreen needleleaf forests, croplands, grasslands, and shrublands. To analyze patterns in thermal roughness length, we estimate the ratio of momentum to thermal roughness lengths ( k B − 1 ) at half hourly scales. Because kB−1 cannot be directly measured, we infer kB−1 from surface flux and meteorological measurements using Monin-Obukhov similarity theory. The inferred kB−1 is moderately correlated with the measured friction velocity over two of the land cover types—grasslands and shrublands—but weakly correlated at croplands and forests, suggesting that kB−1 varies with friction velocity over bluff body surfaces, but not over tightly packed, permeable canopies, such as forests and croplands. This is consistent with previous work showing contrasting turbulent structures over sparse and dense vegetation canopies. To determine how this degree of correlation affects the modeling of kB−1, we estimate kB−1 using four common models, specifically those with friction velocity dependencies, and find that no single model works the best across land covers.