Hapke radiative transfer model has been widely used in the field of soil remote sensing, such as characterizing soil reflectance characteristics, and inverting soil particle size, water content, and surface roughness. However, the latest development of the Hapke model used piecewise fitting of soil spectral reflectance, which brought great difficulties to soil parameter inversion. This paper presents a method to calculate the imaginary part of the soil complex refraction index using the spectral reflectance of dry soil, which can effectively solve the problem of piecewise fitting the soil spectral reflectance by the Hapke model, and then improve Hapke radiation transfer model. Finally, the improved Hapke model is coupled with the MARMIT-2 model to invert soil water content. The results show that the improved Hapke can effectively characterize the spectral characteristics of soil and show higher fitting accuracy (RMSE < 0.012), especially with high soil water content (>30%). Therefore, the improved Hapke radiative transfer model can better understand soil physical properties, improve the inversion accuracy of soil-vegetation physical parameters, which can be use to enhance agricultural water use efficiency, promote optimal allocation of water resources and ensure food production security.