After the revision of the International System of Units (SI) in May 2019, the Planck constant <inline-formula> <tex-math notation="LaTeX">$h$ </tex-math></inline-formula> became the top of the dissemination system for the kilogram. However, the Planck constant is difficult or impractical to use in daily calibrations. Thus, in many applications, alternative physical artefacts are used as mass standards for dissemination. Mass measurements of silicon spheres for the determination of Avogadro and Planck constants have shown that silicon spheres possess stability with high accuracy, thus can be utilized as mass standards. However, the material and the geometry properties of silicon spheres are different from classic mass standards commonly used for the dissemination of the kilogram. It is therefore necessary to determine mass corrections of silicon spheres to improve mass measurements. The effect of air density fluctuations in air buoyancy correction of silicon spheres mass measurements is investigated in this article. Six types of air density fluctuations for three different weighing processes are summarized. In addition, a method to minimize the impact of air density fluctuation by applying real-time air density data to estimate the air buoyancy correction is proposed. Mass comparisons using a 1 kg silicon sphere and classic mass standards in various materials were conducted for method verification. Results show that the improved method using the real-time air density of each alternative weighing process can reduce the measurement errors by <inline-formula> <tex-math notation="LaTeX">$8.7~\mu \text{g}$ </tex-math></inline-formula> to <inline-formula> <tex-math notation="LaTeX">$39.0~\mu \text{g}$ </tex-math></inline-formula> in the air buoyancy correction, which is not negligible when performing high accuracy mass measurements for the dissemination of the redefined SI mass unit using silicon spheres.