The Galileo Reference Center (GRC) provides independent monitoring and assessment of the quality of Galileo system and services. It consists of a core facility operated by the European GNSS Agency (GSA) and is supported by contributions from entities located in EU member states, Norway and Switzerland (GRC-MS). One of the GRC-MS consortium is coordinated by CNES and includes inputs from 20 partners in 12 countries. Monitoring activities include field measurements campaigns in different transportation domains (road, aviation, maritime) and the analysis of the GNSS performances in the associated environments. The Romanian Space Agency (ROSA) is a partner in the GRC-MS consortium and is responsible for the assessment of GNSS performance in the maritime environment. Five measurement campaigns were performed on a quarterly basis from November 2018. This paper presents the main findings of these campaigns. Data collection equipment, main features of the adopted vessels, and data processing techniques are also discussed. The data collection equipment included two types of GNSS antennas, i.e. Septentrio PolaNT-X MF, a regular, maritime grade antenna and Septentrio Choke Ring antenna. Their combined used allowed for the assessment of the multipath generated on board the vessel. The antennas were connected to two identical multi-constellation multi-frequency receivers (Septentrio AsteRx-U), designed for applications requiring centimeter-level position accuracy. In particular, one of the receivers was connected to one maritime antenna and to the choke ring antenna, while the other was connected to two maritime GNSS antennas. Two ships have been used: GEOECOMAR’s Mare Nigrum, an oceanographic multidisciplinary research vessel, and Cpt. Cdor. Alexandru Catuneanu, a maritime hydrographic vessel of the Romanian Maritime Hydrographic Directorate. Mare Nigrum is 82 m long with 3200 t displacement, while Catuneanu is 65 m long with 2665 t displacement. Their missions include the hydrographic and oceanographic surveying of the Black Sea, beaconing activities to support the safe navigation in the Romanian jurisdiction area, and morphological studies of the Black Sea bottom. The deployment of the GNSS antennas on board these vessels considered several aspects. These include the need for power supply, protection from water, cable length limitations and a proper distance from the masking structures of the vessels (including other antennas). In particular, the best identified location for the GNSS antennas was, in the case of Catuneanu, the lighting mast structure from the bow of the ship and, for Mare Nigrum, a small bridge between the aft side of the chimneys. The following data was collected and analyzed by the receivers for both the GPS and the Galileo constellations: observations (pseudo-range, carrier-phase, Doppler shift, carrier-to-noise ratio and lock-time of each tracked signal), navigation messages, and differential corrections (from both local DGPS and EGNOS). Data binary files from the receivers were converted into RINEX observation files. These, in turn, were used to generate a text file with estimated pseudo-range measurements, information on possible cycle slips, and elevation and azimuth of the tracked satellites. Six different PVT solutions are then evaluated, i.e.: (1) Galileo-only Single Frequency (SF); (2) Galileo-only Dual Frequency E1/E5a (DF); (3) GPS-only SF; (4) GPS-only DF; (5) Combined GPS/GALILEO SF; and (6) Combined GPS/GalileoALILEO DF. The performance of each of these PVT solutions are then evaluated using a reference trajectory for the vessels. Such a reference was obtained using the RTK or Precise Point Positioning (PPP) algorithms implemented in the gLab Tool (“The GNSS Laboratory Tool Suite (gLAB) updates: SBAS, DGNSS and Global Monitoring System”,Ibanez et.al, NAVITEC 2018). For each of the five campaigns, measured data were collected over a period of two hours. The vessel was always navigating in the Romanian coastal waters. The only exception is the campaign performed in March 2019, when the ship was navigating in the Aegean Sea, near the Greek island of Kea. All the PVT solutions were calculated for the two aforementioned antenna configurations. The reference trajectory for the campaigns performed near the Romanian coast was computed using RTK corrections from either a reference station located in the Port of Constanta or with a station located in Mangalia. A PPP reference positioning solution was used for the campaign in the Aegean Sea. The analysis was performed for both Cartesian coordinates and ENU coordinates and the provided results were the Position Error (PE) statistics (standard deviation, mean value, absolute mean value, 95% accuracy, absolute minimum and maximum values) and the number of samples. Only the samples when a PVT solution was available for both the frequency combination under analysis and the reference were considered. If a PVT reference was not available, the samples were discarded. The geometry of the GPS and Galileo satellites, in terms of dilution of precision (position, time, horizontal and vertical), as well as the number of satellites used in each of the six PVT solutions were also analyzed. This supported the comparison of the different PVT solutions in terms of accuracy performance. A summary of the results obtained for each data collection campaign are presented in the following paragraphs. The first data collection campaign was run in November 2018, with Mare Nigrum, in the Black Sea near the Romanian coastline. The PVT solutions analyzed were GPS single and dual frequency and GALILEO single and dual frequency. These solutions were compared to an RTK reference solution. During the data collection period of two hours, the number of GPS satellites in view varied between 8 and 11, whereas only 6 Galileo satellites were in view (out of the 20 operating at that time). The 95 percentile ENU accuracy recorded was 2.4 m, 2.3 m and 5.2 m for Galileo SF, and 1 m, 1.95 m and 3.5 m for GPS SF. For DF, the 95 percentile ENU accuracy was 2.5 m, 1.8 m, 1.95 m for Galileo, and 1 m, 2.3 m, 3.2 m for GPS. The results revealed that GPS had an overall better performance as compared to Galileo, with some exceptions in the Up direction, for which GPS dual frequency had a worse standard deviation than Galileo dual frequency of about 35 cm. It was also observed that the impact of a second frequency for PVT was not always positive. However, Galileo dual frequency solutions in the East and North directions had a standard deviation of about 10 cm lower than the single frequency results. In terms of the antenna configurations, Galileo had better results in the East and North directions on the PolaNT-x MF marine antenna , while the GPS PVT solutions were improved by the use of a choke ring antenna. This could indicate a better multipath rejection for Galileo than GPS. The second data collection campaign was performed in the Aegean Sea, with Cpt. Cdor. Catuneanu, in March 2019. The analyzed PVT solutions were GPS L1 and L1 L2 and Galileo E1 and E1 E5a E5b, while the reference positioning solution was computed using PPP. There were 10 GPS satellites and 8-9 Galileo satellites in view, out of the 23 Galileo satellites in operation at that time. The 95 percentile ENU accuracy recorded was 1.7 m, 2.8 m and 4.2 m for Galileo SF, and 2 m, 3.1 m and 7 m for GPS SF. For DF, the 95 percentile ENU accuracy was 1.6 m, 3 m, 4.2 m for Galileo, and 2 m, 3.1 m, 7 m for GPS. This show a generally better accuracy for Galileo, especially on the Up direction. There were also small improvements of the PVT solutions when using multiple frequencies. The largest standard deviation for Galileo E1 was 1.6 m in the Up direction, while in the same direction the standard deviation for GPS L1 was 1.7 m. Comparing the two antenna configurations, GPS had similar performances for both configurations, while the Galileo PVT solutions provided better results for the regular marine antenna . In the third data collection campaign, from June 2019, Catuneanu was navigating in the Black Sea, close to the Port of Mangalia. The computed PVT solutions were Galileo E1 and E1 E5a E5b and GPS L1 and L1 L2, which were compared considering an RTK reference solution. There were 8-10 GPS satellites and 8-9 Galileo satellites used in PVT processing. The Galileo constellation had 21 operational satellites at that time. The 95 percentile ENU accuracy recorded was 3 m, 4.7 m and 2.8 m for Galileo SF, and 4.5 m, 2.8 m and 6.5 m for GPS SF. For DF, the 95 percentile ENU accuracy was 3 m, 4.6 m, 3 m for Galileo, and 2 m, 4.3 m, 5.4 m for GPS. In terms of dilution of precision, the results for Galileo were slightly higher than the ones for GPS, but both constellations had an increase in the vertical DOP in the second part of the data collection campaign. A significant spike could be observed in the Up-direction error analysis, corresponding to the moment of time of the jump in the vertical DOP. The PVT solutions for the East and North directions provided lower standard deviations for GPS, with maximum 1.4 m for L1 and 1.6 m for L2, while in the Up direction the results were better for Galileo, with maximum 1.3 m for E1 and 1.4 m for E1 E5a E5b. Comparing single to dual frequency, the performances were very similar. For Galileo, the choke ring antenna provided lower standard deviations in the North and Up directions, and higher values for the East direction, as compared to the marine antenna. For GPS, the choke ring antenna improved the PVT solutions in all directions. The fourth data collection campaign was performed in August 2019, during two hours which started when Cpt. Cdor. Catuneanu was leaving its docking position in the Port of Constanta. The analysed PVT solutions were Galileo E1 (SF) and E1E5a (DF), GPS L1 and L1 L2, Galileo E1 + GPS L1 and GALILEO + GPS with all available frequencies (E1/E5a/E5b, L1/L2/L5). The results were compared based on a reference positioning solution computed using RTK corrections. The total number of samples used in the PVT computation was 7197. There were 7-10 GPS satellites and 5-7 Galileo satellites used in PVT processing. The Galileo constellation had 22 operational satellites at the time of this campaign. The dilution of precision analysis revealed better results for GPS, with maximum values lower than 2.5. For Galileo, the maximum position DOP reaches 3.4. The increase in the DOP values for GALILEO corresponds to an epoch when the number of used satellites decreases from seven to five. Also, a large spike in the Up-direction error analysis can be correlated to the high values of the DOP. For the combination between GPS and Galileo, there were 13-16 used satellites and the DOP values reach a maximum of 1.5. The 95 percentile ENU accuracy recorded was 1.2 m, 2.4 m and 1.7 m for Galileo SF; 1.25 m, 1.5 m and 0.9 m for GPS SF; and 1.5 m, 1.5 m, 1 m for GPS/Galileo combination SF. For DF, the 95 percentile ENU accuracy was 0.9 m, 2.9 m, 2.6 m for Galileo; 1.2 m, 1.5 m, 0.9 m for GPS; and 1.5 m, 1.5 m, 1 m for GPS/Galileo combination L1/L2/L5/E1/E5a/E5b. Overall, the PVT solutions obtained using Galileo only provided the highest standard deviation values, with a maximum of 0.9 m in the Up direction. The results of GPS only and Galileo + GPS combination were very similar, reaching a maximum value of 0.5 m, and the addition of multiple frequencies did not significantly improve the results. The choke ring antenna provided better standard deviation values for all constellation combinations, except for GALILEO only multi-frequency, for which the marine antenna provided significantly improved results. In the fifth data collection campaign, performed in October 2019, Catuneanu navigated in the Black Sea, near the Romanian coastline. The analysed PVT solutions followed the same constellations and frequency combinations as in the fourth campaign. Similarly, the reference solution was computed using RTK corrections from a reference station located in the Port of Constanta. There were 6519 samples used in the computation of the PVT solutions. Regarding the amount of satellites used, there were 8-11 GPS satellites, 6-8 GALILEO satellites and 14-19 satellites used in the multi-constellation case. The Galileo constellation had 22 operational satellite at the time of this campaign. The dilution of precision of GPS reached a maximum of 2.2, while for Galileo a large jump that reached 7 could be observed in the analysis of the vertical DOP. For the GPS + Galileo combination, the DOP values were much lower, reaching a maximum of 1.5. The 95 percentile ENU accuracy recorded was 1.3 m, 1.2 m and 7.7 m for Galileo SF; 1.3 m, 1.2 m and 5.5 m for GPS SF; and 1.5 m, 0.7 m, 5.1 m for GPS/Galileo combination SF. For DF, the 95 percentile ENU accuracy was 1 m, 1.8 m, 13.8 m for Galileo; 1.3 m, 1.2 m, 5.5 m for GPS; and 1.5 m, 0.7 m, 5.2 m for GPS/Galileo combination L1/L2/L5/E1/E5a/E5b. Compared to the East and North directions, the error values in the Up direction for the Galileo solutions were very large and they could be correlated with the increase of the DOP and the low number of used satellites. This tendency was also observed in the analysis of the standard deviation values, which reached 4.5 m for GALILEO in the Up direction, while the other two constellation combinations had a maximum standard deviation of 1.1 m in the same direction. In the East and North directions, the maximum standard deviation was 0.8 m and it was observed that, in most of the cases, the performance was not improved when multiple frequencies were added. In terms of antenna configurations, it was observed that the choke ring antenna provided better results in the Up direction, while the marine antenna had better East and North results. Comparing the results obtained from these five data collection campaigns, it was observed that the variations in performance could be explained by the satellite geometry and the number of satellites used in PVT computations. Also, the source of the RTK corrections was considered to be an influencing factor, as both GPS and Galileo results obtained with the data provided by the Constanta reference station had improved performance as compared to the ones obtained using RTK corrections from the Mangalia reference station. In conclusion, although the Galileo constellation is not yet complete, its positioning accuracy performance in a typical maritime environment is shown to be at the same level of the performance observed with a GPS only solution.