The fluid-filled crack model has been considered as a model of the resonator at the source of long-period (LP) seismic events. Crack geometry and fluid acoustic properties have been estimated by the comparison between observed peak frequencies of LP events and resonance frequencies of the crack model. Recently, an analytical formula for the resonance frequencies of the crack model was proposed by Maeda and Kumagai (GRL, 2013; GJI, 2017). Taguchi et al. (AGU meeting, 2016) showed that observed several peak frequencies in LP events at Kusatsu-Shirane and Galeras volcanoes were successfully explained by those predicted from the analytical formula. In this study, we further analyzed LP events at both the volcanoes, and systematically estimated the crack model parameters at the LP sources using the analytical formula, in which we assumed misty and dusty gases at Kusatsu-Shirane and Galeras, respectively. Our estimates indicated that the crack geometry and fluid properties largely changed in our analysis period between August 1992 and January 1993 at Kusatsu-Shirane and that between 6 and 10 January 1993 at Galeras. We found that the crack volume increased with increasing gas-weight fraction at both the volcanoes. Although an increase in gas-weight fraction was indicated by Kumagai et al. (JGR, 2002) at Kusatsu-Shirane within our analysis period, our estimates indicated that the significant volume increase (from 10-1 m3 to 103 m3) was associated with the gas-weight fraction increase. At Galeras, we found that the crack volume systematically decreased with decreasing gas-weight fraction. An eruption occurred several days after the LP activity analyzed in this study. Our estimated trend may be caused by precursory processes leading to the magma eruption. This study demonstrates that the approach using the analytical formula is useful to constrain the source process of LP events and to diagnose the state of fluids in magma and hydrothermal systems.