Underwater autonomous gliders (UAG) such as Slocum electric gliders provide an effective platform and tool for marine and coastal scientists to conduct prolonged missions over weeks or months. However, localization of these gliders underneath ocean surface is challenging due to the lack of inexpensive, effective underwater positioning sensors. In this paper, we present a dynamic model-based localization scheme to improve the positioning estimation of UAG. The new localization approach is built on an experimentally validated dynamic model and fused with onboard depth and yaw angle sensor measurements. We compare and validate the proposed scheme with dead reckoning and the ground truth.