Understanding the intricate spatiotemporal dynamics of merging flames holds significant importance for advancing the efficiency and stability of gas turbine combustion systems. In the present study, we conducted a series of experimental investigations using a three-candle configuration to emulate the merging behavior observed in such systems. High-speed imaging at a remarkable frame rate of 250 frames per second (fps) was employed to capture the rapid flame interactions. To visualize the flame structures at various time instants, we utilized Shadow imaging techniques, enabling the discernment of fine details in the images. Subsequently, we performed an indepth analysis of the acquired flame image sequences using the Spectral Proper Orthogonal Decomposition (SPOD) technique. This analysis allowed us to identify and quantify the dominant spatiotemporal coherent structures present in the merging flames. Through the SPOD analysis, we successfully extracted eight distinct SPOD modes along with a mean image, effectively representing the ensemble behavior of the merging flames. Further, we investigated the variation of absolute eigenvalues with respect to frequency for each identified mode. This investigation enabled a comprehensive understanding of the frequency characteristics associated with different flame structures, revealing essential insights into their temporal evolution and mutual interactions. The outcomes of our study significantly contribute to the fundamental understanding of merging flames within gas turbine combustion systems. The identified spatiotemporal modes offer valuable information for the design, optimization, and control of combustion processes, thereby facilitating enhanced flame stability, efficiency, and reduced emissions. The utilization of high-speed imaging in conjunction with SPOD analysis demonstrates its effectiveness as a robust methodology for comprehending the complex dynamics of merging flames, paving the way for further advancements in this field.