A Coupled WRF-CLM5 Framework for Simulating Carbon-Nitrogen Dynamics and Biogeochemical Feedbacks

Abstract. Neglecting biogeochemical cycles in regional climate simulation compromises the representation of land-atmosphere interaction and the exchanges of materials and energy. The WRF (Weather Research and Forecasting) model coupled with CLM4 (Community Land Model, version 4) (WRF-CLM4) was widely used to simulate regional climate and environment. However, this open-source model lacks advanced biogeochemical cycle module. To address this, we develop an online coupling framework that embeds the advanced biogeochemical processes of CLM5 (Community Land Model, version 5) into WRF model (WRF-CLM5), integrating the fixation and uptake of nitrogen (FUN), leaf utilization of nitrogen for assimilation (LUNA), and Medlyn stomatal models. By introducing the pre-stabilized initial fields, this system achieves full biogeochemical simulations across all subgrid patches. Three sensitivity experiments: WRF-CLM4 (Exp_CTL), WRF-CLM5 with LUNA and Medlyn (Exp_LUNA); and WRF-CLM5 with LUNA, Medlyn, and FUN (Exp_FUN), are designed to evaluate model's capacity in representing carbon-nitrogen (C-N) processes. Compared with Exp_CTL, both Exp_LUNA and Exp_FUN markedly reduce gross primary production (GPP) overestimation. The Exp_FUN, which incorporates flexible C:N ratios and novel nitrogen uptake strategies, yields the best performance. The three experiments yield magnitudes and spatio‑temporal patterns of net primary production (NPP) and net ecosystem exchange (NEE) align with the reference data. However, their predicted leaf area index (LAI) values are considerably higher than observation, a discrepancy primarily arising that the model represents total LAI while the observations represent effective LAI. The Exp_FUN reveals that nitrogen acquisition is dominated by active uptake and symbiotic pathways, with carbon costs generally ranging 5–20 gC/gN, and shows pronounced spatial and temporal variations. Compared with flux site observations, Exp_FUN improves GPP simulations, lowering the averaged RMSE by 3.74 gC/m2/d. The improved representation of C-N processes enhances latent heat simulation at the expense of sensible heat simulation, since it alters surface energy partitioning without changing the total surface energy budget. Overall, the coupled model advances regional climate modeling by capturing critical C-N interactions, yet further refinements in initial fields and parameter calibration are essential to reduce residual biases.