GNSS Precise Point Positioning (PPP) has been widely recognized as one of the most efficient approaches to retrieve zenith tropospheric delay (ZTD) / precipitable water vapor (PWV) for the climate and weather monitoring. Nowadays, ground- based GNSS receivers have been widely investigated with good agreements of Numerical Weather Prediction (NWP) model, Radiosonde observations, and satellite measurements. However, this approach is further expected to be applied to the vast ocean to extend the coverage of GNSS derived ZTD/PWV where other observations are sparse. The estimated ZTD/PWV can not only be contributed into NWP for better accuracy but also be employed for the validation and calibration of onboard microwave radiometers to provide precise tropospheric delay corrections, for example, for the altimetry satellites where calibrated tropospheric delays are needed in order to retrieve the sea surface height (SSH) with high accuracy. In this study, GNSS PPP using IGS (real-time) products is applied to a shipborne data set over 20 days in the Fram Strait and the data of a number of GNSS ground stations close to the ship trajectory in kinematic and static mode, respectively. The shipborne and ground-based tropospheric delays or water vapors are compared firstly and the shipborne ones are further validated with numerical weather models and nearby radiosonde data, and satellite onboard microwave radiometer data as well. The ZTD estimate shows good agreement with the coastal ground reference GNSS stations with an RMS value of 0.62 cm, and for the whole trajectory an RMS of 0.65 cm was achieved compared to ERA-Interim based ZTD. We demonstrated that the shipborne GNSS PWV agrees with the ERA-Interim surface product with a linear fit RMS of 0.87 mm and a correlation coefficient of 97.18%, while an RMS of 1.12 mm was achieved compared to nearby radiosonde observations. Moreover, the shipborne GNSS PWV achieved an RMS of ~2 mm with an average bias of 0.85 mm compared to SARAL PWV observations, and ~5% of the SARAL footprints were detected as outliners. With the multi-GNSS constellation, including Galileo and GLONASS in our case, the ZTD/PWV estimates could be improved by ~10%. As conclusions, our research demonstrated that shipborne GNSS PWV could achieve 1~2 mm accuracy, which is useful for the validation and calibration of altimeter satellite observations. Moreover, if utilizing the huge number of ships with GNSS receiver installed, it is expected to achieve a better NWP model with shipborne GNSS PWV included.