Abstract: Acoustic tomography is an advanced and promising approach to reconstruct the aerodynamic field in furnace non-intrusively. A reliable and optimal reconstruction algorithm is critical. In this paper, a novel acoustic tomographic algorithm for 2-D aerodynamic field in furnace is proposed, which originally takes a set of wavelet scaling functions as basis to reconstruct the spatial field to be measured. A scaling function has intriguing characteristics to contain information of all signal frequencies beneath the defined scale, so that it is expected to effectively reconstruct an objective field with complex spatial frequencies. Nevertheless, Daubechies wavelet is applied to construct the 2-D scaling function basis, for the advantage of compact support, orthogonality etc., so as to achieve better identification of reconstruction. On the other hand, the gradient descent method is adopted to solve the linear equation sets in the model. The geometric parameters related to the model are normalized to fit for different size or aspect ratio of flow field to be measured. By reconstructing different typical numerical phantoms of cross-sectional flow fields in a furnace, the proposed acoustic reconstruction model is essentially proved to be valid and robust.