A probability distribution model of the local voidage was proposed to describe and simulate dynamic gas-solid distribution in the bubbling and turbulent fluidized bed reactors. Experiments were carried out in an air-fluidized bed. The bed materials were FCC particles (Geldart A) and irregular sand particles (Geldart B). A cross-optical fiber probe was employed to measure dynamic voidage. The minimum probability method was introduced to identify the division between the emulsion phase and the bubble phase. The statistical analysis indicated that the two particle types employed have extremely different dynamic behaviors corresponding to different gas-solid distributions and the interaction between the bubble and emulsion phases. For the FCC particles, the voidage of the emulsion phase is very close to that at the minimum fluidization with little effect from the formation and motion of bubbles in bubbling regime, and deviates a little from epsilon(mf) in turbulent regime. For the sand particles, the voidage of the emulsion phase differs far from that at the minimum fluidization, and the bubble phase gradually becomes more dilute from bubbling to turbulent regime. However, for both particles the dynamic voidage fluctuations in the emulsion phase and the bubble phase followed beta distribution under various operating conditions. The probability density functions of the local voidage from epsilon(mf) to 1 showed the continuous double-peak phenomena, one peak for the emulsion phase and another for the bubble phase, and evolved with changing operating conditions and bed position. A particular distribution, called coupled beta distribution, was developed to describe and simulate such probability density function with double peaks and its complex evolution from bubbling to turbulent regime. The quantification of the probability density function then statistically introduced the spatiotemporal two-phase flow structure.