Towards an Improved Surface Roughness Length Parametrization: Optimising Vegetative Drag in JULES

The momentum flux is the fundamental process that determines the amount of turbulence in the atmosphere closest to the surface and hence impacts not only on the transfer of momentum, but also on scalars such as heat and moisture. An initial analysis of the surface momentum flux in the JULES land surface model shows that there are substantial biases for all land cover types. By utilising site observations, it is possible to constrain these errors with the choice of values used for the roughness length. Hence, there should be the potential to improve the parametrization and remove the biases.

Data from the FLUXNET2015 database have been used to identify an appropriate roughness length for each site considered, although the caveat is that some key assumptions related to the observations have to be made that may limit the confidence in these values. Both site-specific and IGBP class-specific roughness lengths have been determined. The seasonal and site variations in these optimal values suggest additional dependence on leaf area index (LAI). Hence, including LAI in the parametrization of roughness length is also considered.

These new roughness lengths have been compared to the original default values within the JULES model. Parameters used by ECMWF in HTESSEL both before and after being optimised to synoptic observations for 10 m wind speed have also been considered. Simulations using these new roughness lengths in offline JULES have been compared to both the FLUXNET2015 data and the original simulations for each site in terms of errors in the surface momentum flux and energy balance components. Results will be presented to show that important improvements can be achieved through the more appropriate values for the roughness length. Moreover, it will be shown that these indicate widespread enhanced roughness compared to typical classical literature values for particular land cover types, suggesting other roughness elements within the footprint area influence the overall surface drag.