The ambiguity resolution (AR) significantly enhances the accuracy of precise orbit determination (POD). There have been numerous studies of different forms of POD: double-difference (DD), single-difference (SD), and un-differenced (UD) AR methods for global navigation satellite systems (GNSS) or low earth orbit (LEO). However, challenges persist in the integrated POD (IPOD) of the GNSS and LEO at regional ground stations. These challenges include the frequent selection of dual receiver-satellite pairs in DD methods, and time-varying hardware biases in LEO receivers for UD methods. In addition, the SD AR method has not been explored in IPOD, resulting in unfixed ambiguities. In this study, we investigated the feasibility and performance enhancement of AR in the BeiDou Navigation Satellite System (BDS) and LEO IPOD under regional ground stations using simulated ground and onboard observations. First, we introduce AR models applicable to BDS and LEO IPOD and analyze the applicability of different AR models for IPOD under regional ground stations. We designed a study to utilize SD ambiguity, which eliminates the time-varying hardware bias of the LEO receiver end, to estimate the uncalibrated phase delay (UPD) of the satellite end. Furthermore, we designed the BDS-3 and LEO constellations with 24 regional ground stations in China and simulated seven days of observations. Subsequently, the narrow-lane (NL) UPD quality and AR performance were analyzed, and a solution with satisfactory stability and residual distribution was obtained, enabling the implementation of SD AR. The daily fixed rate for wide-lane ambiguities exceeded 99%, while for NL ambiguities it surpasses 86%. After fixing ambiguities, the BDS-3 orbit’s along-track and cross-track components significantly improved. Simultaneously, LEO orbit solutions improved by over 20% in all three directions. Overall, the UPD estimation model using SD ambiguities yielded satisfactory UPD results, enabling AR and significantly enhancing the orbit accuracy of GNSS and LEO.