Using all observed information in a variational approach to measuring z(0m) and z(0t)

A variational method is developed to estimate the aerodynamic roughness len gth and roughness scaling length for temperature based on wind and temperat ure measurements conducted routinely at an observational network. The probl em is formulated to find optimal estimates of roughness lengths for momentu m and heat transfer through a minimization of a cost function with respect to these two roughness lengths that measures the errors between observed an d predicted wind and temperature profiles. The method has been applied to d ata collected in two experimental campaigns. Some results are compared with other methods used to compute the aerodynamic roughness length. The variat ional computations show that the aerodynamic roughness lengths agree well w ith the estimated z(0m) in the experimental campaigns. The roughness scalin g lengths for temperature z(0t) are in most cases one order of magnitude sm aller than z(0m). It was found that the variations of z(0m) and z(0t) durin g the course of a day are not likely to follow a simple functional relation ship, especially during the daytime, during which both z(0m) and z(0t) are highly oscillatory. The error test shows that z(0m) and z(0t) generated fro m the variational method are not very sensitive to measurement errors.