A physics-based model to estimate source conditions for a tephra-dispersal model is developed. The source condition is generally expressed by a distribution of released particles along an eruption plume (referred to as source magnitude distribution" SMD).The present model (NIKS-1D) calculates the SMD and the column height for given vent conditions(e.g. mass eruption rate and magma properties) on the basis of an eruption column model below the neutral buoyancy level (NBL), a downwind gravity current model around the NBL,and a particle sedimentation model.It quantitatively reproduces the following features of the SMD for typical explosive eruptions: (1) a significant amount of coarse particles are released from the rising eruption column, whereas most of the fine particles are carried to the NBL, (2) in a downwind gravity current, the coarse particles tend to decrease more rapidly with distance from the vent than the fine particles,(3) the SMD from the downwind gravity current decreases with distance more slowly in a strong ambient wind than that in a weak ambient wind, and (4) the SMD from the downwind gravity current for eruptions with large mass eruptionrates decreases with distance more slowly than that for eruptions with smalleruption rates. NIKS-1D includes a new parameter, $\mu$, which represents the ratioof the volumetric flux at the source of the downwind gravity current to that ofthe eruption column model at the NBL. This parameter is determined by thephysics of the entrainment process around the connection between the eruptioncolumn and the downwind gravity current, and depends on the intensity oferuptions. We propose an empirical formula to calculate the value of $\mu$ as afunction of the mass eruption rate on the basis of the observation data fromthree well-studied eruption events. In a real-time tephra-dispersal predictionsystem, NIKS-1D estimates the mass eruption rate from the observed plumeheight, and calculates the SMD from the estimated mass eruption rate as asource conditions for a tephra-dispersal prediction immediately after an eruption.