Climatic changes with warmer temperatures in mid-latitudes require the need to improve the simplified vegetation scheme of the regional climate model COSMO-CLM, which is not capable of modelling complex processes depending on temperature, water availability and day length. Thus, we have implemented the physically based Ball-Berry approach coupled with photosynthesis processes based on Farquhar and Collatz models for C3 and C4 plants in COSMO-CLM (v 5.16). The implementation of the new algorithms includes the replacement of the “one-big leaf” by a “two-big leaf” approach. We performed single column simulations with COSMO-CLM over three observational sites with C3 grass plants in Germany for the period from 1999 to 2015 (Parc, Linden and Lindenberg domain, Fig.1). Hereby, we tested three alternative formulations of the new algorithms. The first formulation (COSMO_v3.5) is based on the Community Land Model (CLM v3.5) algorithms for stomatal resistance, which depend on leaf photosynthesis, CO2 partial and vapor pressure and minimum stomatal conductance. The second one is COSMO_v4.5, which is based on the phenology algorithms of CLM v4.5 including the soil water stress function. The third one is similar to COSMO_v4.5 but with additional equations for dry leaf calculations (COSMO_v4.5e). The results revealed major differences in the annual cycle of stomatal resistance compared to the control simulation (COSMO_orig) with the original algorithm (Fig. 1). The biggest changes are from October to April when stomata are closed. The summer values of experiments are closer to measured values, than COSMO_orig. Further, changes in the stomatal resistance algorithms improve the accuracy of calculated transpiration rate and total evapotranspiration. The results indicate that changes in stomatal resistance and photosynthesis algorithms can improve the accuracy of other parameters of the COSMO-CLM model by comparing them with FLUXNET data and meteorological observations at the sites, and GLEAM datasets. Figure 1: The stomatal resistance based on COSMO-CLM experiments (a - annual cycle; b - daily values from 01.06.2011 to 15.09.2011) for: I – Parc domain, II – Linden domain, III – Lindenberg domain. This research was funded by the German Research Foundation (DFG) through grant number 401857120