An inverse wave modeling-based method is proposed for globally reconstructing the sound field in room environments. The method builds the wave model of the sound field as prior knowledge to support the reconstruction under strong reverberation. In this method, the whole space is divided into a set of subdomains. Based on the theory of discretization-based numerical simulation, a wave model that can describe the transfer characteristic between any subdomain and the source is built. Supported by this model, the sound source is recovered based on spatial sound pressure sampling and the global sound field reconstruction can be further accomplished in the reverberant environment. In particular, the shape function with the property of sparsity is constructed in building the wave model. Then, the intensity on the point source is represented by a sparse vector over the subdomains, and then, the sparse method can be used to achieve the recovery of this vector, which reduces the sampling burden in the space. Numerical verifications are performed to evaluate the performances of the proposed method. It demonstrates that the proposed method is capable of obtaining accurate reconstructions in a strong reverberant environment. It also shows that the method is applicable to problems with complicated excitations in the low-frequency range.