This paper investigates the buoyancy effect on near-field evolution of triple equal starting thermal plumes from volumetric heat sources, with a focus on merging process, velocity evolution, and stream-wise penetration. Instantaneous velocity fields and corresponding vorticity distribution, temporal evolution of centerline axial velocities, temporal penetrating rates and heights are examined by 2-D particle image velocimetry (PIV) at three different source heat strengths of 180 W, 90 W, and 30 W. A four-stage merging process is demonstrated to be independent of heat strength. Normalized starting and terminating times of each developing stage are also unified regardless of the heat strength, i.e., t/t∗ = 1.7–2.5 for stage i (relatively isolated development), t/t∗ = 2.5–3.3 for stage ii (bending wall flows), t/t∗ = 3.3–3.9 for stage iii (development after a self-merged state), and t/t∗ > 3.9 for stage iv (evolution with global merging). The axial velocity at a specific centerline point usually involves three distinct developing periods: a “centerline-silent” period ending at a normalized transportation time tt_n when heat is just transported to the measuring points, a fast rising period, and a fluctuating period starting at a normalized endurance time te_n. The buoyancy-independent tt_n is approximately 1.5, 2.2, 3.0, and 3.2 at a normalized downstream distance of 1.5, 2.5, 3.5, and 4.5, respectively. The time te_n, independent of buoyancy and axial height, is about 3.6 for all measuring points in this study. The maximum axial velocity above the middle heat source exhibits three linearly developing periods with different acceleration rates successively: a quicker rate as the wall flow develops, a slower rate due to enhanced lateral interaction, and the fastest rate as the global merging happens. In different periods, the proportional coefficients of linear fitted functions vary from 0.48 to 2.98. The downstream distance where axial velocity is maximum shows similar power-law increasing during t/t∗ = 1.5–3.2 at Q = 180 and 90 W, but exhibits significant fluctuations at Q = 30 W. Normalized overall penetration based on global velocity fields is notably faster and higher at the lowest buoyancy (Q = 30 W) than that at the higher buoyancy (Q = 180 and 90 W), probably due to the weaker lateral interaction and turbulent mixing together with the more dedicated vertical rise at Q = 30 W.