The acoustic tomographic (AT) approach aims to obtain information in a non-invasive way. Multidimensional and comprehensive information about temperature distribution can be used to optimize the performance of industrial processes. AT is a promising method for visualizing the 2D temperature field. The key challenge is to improve the accuracy of the relative spatial distribution and the absolute temperature value. Thus, this paper presents an alternative way to improve the reconstruction of the temperature field distribution by analyzing the effect of the grid structure and acoustic paths. In order to achieve this spatial moving average technique (MAT) have been studied and implemented across different grid structures (hexagon, square, and polar). Several numerical simulations were conducted to analyze the accuracy of the reconstructed temperature field using these three grid shapes. The effect of grid size was also analyzed. It was found that the hexagonal grid is more sensitive to the shift of hot spots and more accurate in the peak temperature. A hexagonal grid whose radius of the circumscribed circle is equal to 7 % of the sample diameter could efficiently detect a 3 % shift of the hotspot with less than 5 % error in the peak temperature. This simulation result was supported experimentally by 12 ultrasonic transceivers arranged in a circular pattern with 1 m in diameter.