Abstract. The estimation of latent heat fluxes in semi-arid regions faces several challenges, such as human intervention in the water cycle through irrigation, sudden changes in vegetation state due to crop harvesting, the still evolving knowledge of the physical processes governing plant transpiration and soil evaporation, and the lack of measurements to develop and test models. Representing the wide range of evapotranspiration values presents difficulties for both simulations and measurements, owing to strong soil/plant spatial heterogeneity at relatively small (e.g. hectometric) scales. The ability to accurately predict the partition of evapotranspiration into evaporation and transpiration from observation is still very limited, but improved estimates are required so that better water use decisions can be made. Land surface models (LSMs) can be used as a tool in this regard, when their validation is possible, simulations tend to overestimate soil evaporation in most models.

The simulations in this study make use of the LSM ISBA, which represents the land component within the surface coupling platform SURFEX. They include two field sites with contrasting soil moisture and vegetation characteristics during the summer of the Land surface interactions with the atmosphere over the Iberian semi-arid environment (LIAISE) campaign. The first site corresponds to a full cutting and growing cycle of one month in a flood irrigated alfalfa field. A detailed examination of the parametrization suggests that several parameters determine the amount and tendency of transpiration change. In particular, a higher quantum efficiency and maximum assimilation are marked as the driving model parameters together with a mesophylic conductance value closer to C4 behavior. The second site is an uncultivated rain-fed area of natural grass close to senescence. As the parametrization of the vegetation proved to be insufficient to characterize the evapotranspiration, for this study the implementation of a dry surface layer (DSL) resistance within the LSM ISBA was developed. The consideration of this process characterizes the transfer of vapor in a physical way that has proved successful in improving the partitioning of evapotranspiration in other models. The implementation of a DSL resistance led to an improvement in the simulated latent heat flux by reducing bare soil evaporation compared to simulations without a soil resistance. This approach resulted in a reduction in daily latent heat flux RMSE of 29 % and 32 % for the alfalfa and natural grass site respectively, while increasing slightly the correlation by 0.02 and 0.01 at both sites. Sensible heat flux and net radiation are improved on the order of 10 W m^-2 whereas the ground heat flux is deteriorated within the same order. The resulting DSL simulations reduced the overall global error compared to a simulation without a DSL resistance. Sensitivity tests of the parameters that drive a DSL resistance in ISBA further improve the simulations, reducing excessive damping after rain events. The new DSL parameterization helps overcome current problems of ET modelling by reducing bare soil evaporation within LSMs.