The conventional Huff model has a serious limitation with respect to over-smoothing of the rainfall peak because it is derived based on the mean temporal distribution of observed rainfall events. To overcome this limitation of the conventional Huff model, this study proposes a modified Huff model that considers the characteristics of the temporal distribution of observed annual maximum rainfall events. After normalizing and rearranging rainfall events around the storm center, the modified Huff model is formulated to consider their mean characteristics. Compared to the conventional Huff model, the modified Huff model has a unique characteristic of being a centered-type, which is also formulated using the rearranged rainfall events around the storm center. When we evaluated the suitability of the conventional Huff model, we found that the model is not representative of the temporal rainfall distribution of observed storm events. Though the modified Huff model is formulated as a centered-type that is found to perform better than the conventional Huff model formulated for a given quartile, the modified Huff model still has a limitation that the model is derived by considering the mean characteristics of the observed rainfall temporal distribution.