[1] Utilizing global auroral images obtained by Polar Visible Imaging System Earth Camera, we have analyzed the UV emissions from 116 auroral substorms. The events selected were fairly isolated and had to expand from a localized onset. It was found that essentially all auroral events fulfilling these simple criteria expand into the Akasofu (1964) bulge-type aurora. Averaged auroral emission patterns were compiled for 11 time steps of the substorm covering 20 min prior to the onset until well into the recovery phase. This compilation required a three-step normalization technique, one temporal based on the expansion time and two spatial, magnetic local time and latitude. These average patterns were then fit to either a single or double Gaussian distribution in latitude for each of 24 magnetic local times. On the basis of this analysis we made the following conclusions. The normalization technique is highly efficient in maintaining the key features in the individual auroral emission patterns, even though the individual events varied significantly in intensity, size, position, and lifetime. Thus our normalization results quantitatively validate the Akasofu (1964) assumption that key auroral features exist in the bulge-type auroral substorm. After the substorm onset the auroral oval becomes clearly bifurcated consisting of two components: the oval aurora in the latitude range of the preonset oval and expanding primarily eastward postmidnight, and the bulge aurora, which emerges out of the oval, expanding poleward and both east and west in MLT. The oval aurora decay faster than the bulge emissions indicating that the decays have separate time constants. Owing to the pronounced difference in the spatiotemporal behavior of the two auroral components we suggest that their sources are quasi-independent.