Abstract: To address compatibility issues between meteorological and hydrological models in two-way land-atmosphere coupling, a highly efficient WRF/Grid-XAJ two-way coupling system was constructed based on the principle of mass conservation using soil moisture as a linkage. Analysis of two flood events in the Tunxi catchment revealed the following: the model weight parameter quantitatively assessed the compatibility of hydro-meteorological model coupling; after precipitation assimilation, the two-way coupling system slightly outperformed the WRF (Weather Research and Forecasting) model in peak precipitation accuracy (error is within ±5%); the two-way coupling system more accurately reflected soil moisture; both the two- and one-way coupling systems underestimated the hydrographs before precipitation assimilation; after assimilation, their simulation results improved and closely matched those of the Grid-XAJ model (all Nash-Sutcliffe efficiency coefficients E_NS > 0.85), with the two-way coupling system showing the best peak flood simulation results (error is within ±11%). This demonstrates that the WRF/Grid-XAJ two-way coupling system has the potential to promote the simulation and forecasting of rainstorms and floods, offering a new perspective for two-way feedback modeling between hydrological and meteorological models.