Abstract: One-dimensional radiative transfer model has been widely used in atmospheric numerical studies. However, the one-dimensional model neglects the radiation transport process in the three-dimensional cloudy atmosphere and thus produces inaccurate numerical simulations and weather forecasts. In order to address uncertainties in the one-dimensional radiative transfer model and optimize the radiation parameterization in numerical weather forecast system, effects of three-dimensional radiation interaction among clouds and their impacts on the thermodynamic structure of clouds are studied. I3RC Ⅱ cumulus and stratocumulus fields are selected for the present study, and their central areas are taken as the target domain while the surrounding areas are treated as the adjacent domain. Spatial distributions of shortwave and longwave radiative heating rates are simulated by the broadband three-dimensional radiative transfer model SHDOM to illustrate the impact of the adjacent-cloud on the thermodynamic structure of the target-cloud. Simulation results show that the adjacent-cloud imposes longwave radiative warming effects on the entire target domain. The warming zone is mainly located in the surface layer of the cloud, the depth of which is negatively correlated with the liquid water content; the warming rate is proportional to cloud coverage and the reciprocal of the distance between the target and adjacent clouds, and its maximum value is 3.08 K/h. For shortwave bands, the adjacent-cloud has both scattering-induced warming effects and shadowing-induced cooling effects for the target-cloud. With a small solar zenith angle, warming effects are dominant although they are quite weak with a small spatial inhomogeneity. As the solar zenith angle increases, cooling effects become more important and cause significant temperature reductions (the maximum reduction is 1.72 K/h) in the surface layer of the target-cloud on the side sheltered by the adjacent-cloud. The shading effects may even exceed the longwave radiative warming effects. In conclusion, the adjacent-clouds could greatly change the local heating rate of the target-clouds. Therefore, it is crucial to consider the three-dimensional radiative effects in the numerical weather forecast system to improve the radiation parameterization.