Today, Global Navigation Satellite System (GNSS) has been wildly applied in many fields, such as navigation, map-making, surveying, and geotagging. However, the safety of such heavy reliance on the GNSS is often ignored. The ionospheric delay contributes the largest and most unpredictable error to GNSS observations, and the corresponding ionospheric error turns out to be a major source of inaccuracy in user position, navigation and timing information. Moreover, severe ionosphere scintillation in low latitude areas implies serious impacts on GNSS performance in these regions. Fatal accidents are therefore possible if the degradation on GNSS performance is neglected. For this reason, GNSS could not be used alone for safety-of-life applications. To compensate the effects GNSS errors have on GNSS service and enhance integrity, GNSS augmentation systems are developed. One of the augmentation systems is the Satellite Based Augmentation System (SBAS). While the ultimate goal of the SBAS is to provide integrity assurance through integrity information, it also increases accuracy by offering the wide area thin shell planar fit ionospheric delay model to remove the ionospheric delay from GNSS signal. Currently, a variety of ionospheric delay models are available for moderate ionospheric delay and quiet ionospheric activity. Most of the models fail to capture severe ionosphere scintillation and ionospheric storm in areas, Asia Pacific region for example, where the magnetic latitude is low. The five-degree resolution for the ionosphere grids provided by the SBAS using current models is too large to reflect the variation introduced by local ionospheric activity within the grids. Yet the severe scintillation poses another requirement on computation time to catch the dynamic changes. Implementation of a proper wide area ionospheric delay model in Asia Pacific region is a challenging task and with the ionospheric activities reaching the peak for their ten-year cycle in 2012/2013, the problem is even more imminent. The goal of this paper is to develop a proper ionospheric delay model to correct ionospheric error in low magnetic latitude areas, particularly the Asia Pacific region. The proposed ionospheric delay model uses four dual frequency GNSS reference stations distributed in Taiwan as gird points in place of the traditional grid points generated by ionospheric pierce points. The ionospheric delays observed in the four stations are processed and provided to user, who then is capable of constructing the local ionospheric delay by Weighted Least Squares (WLS) with the distances between user and the stations as weightings. In this sense, computation burden is relieved by eliminating the process to convert the delays at the ionospheric pierce points to those in the grid points. Further, the reference stations in Taiwan afford good descriptions on the dynamical variations due to local ionospheric activities while the tradition grid points fail to do so because they are too far away. In this paper, the details on the proposed ionospheric delay model are explained and experiments using data collected from the four reference stations are presented. The effectiveness of the proposed is validated by comparison with the conventional delay model provided by Japanese Multifunctional Satellite Augmentation System (MSAS).