5:15 PM - 7:15 PM
[SGD03-P08] Quality investigation of GNSS Kinematic Analysis at Izu-Oshima
Keywords:GNSS, Kinematic, GPS
Introduction
Crustal deformation is an important measure of volcanic activities as it reflects charging of magma reservoirs, shallow hydrothermal activities, and/or pre-eruptive ascent of fluids. Among various tools of crustal deformation observation, GNSS, which can detect three-dimensional surface displacement, plays an important role in deformation monitoring. For this purpose, JMA carries out continuous GNSS observation at almost 150 stations around volcanoes in Japan. Observed GNSS data are analyzed by daily static analysis using precise ephemeris. Because of this analysis strategy, crustal deformation is obtained as one-sample-per-day data with delay of 2 weeks (final product) or 3 days (rapid product). This implies that the GNSS is used for mid- to long-term evaluation of volcanic activity. Crustal deformations of shorter timescale are monitored by tiltmeters.
However, the number of tiltmeters around the volcanoes is limited. Utilizing GNSS observations for such displacements can contribute to the early detection of the anomalous volcanic activities, and thus finer volcano monitoring. To accomplish this goal, we investigated the quality of GNSS kinematic analysis.
GNSS Kinematic Analysis
MRI have been carrying out GNSS continuous observation at 15 sites in Izu-Oshima for daily static analysis. The sampling rate of the GNSS data is 30 seconds and we use GPS signals only. GNSS data are sent to analysis server every hour.
We concatenated latest 5 hours of these hourly data and applied PPK (Post-Processing Kinematic) analysis using broadcast ephemeris. Analysis tool is rtklib (Takasu et al., 2007).
We then investigated the precision of the obtained coordinate values of Jan. 2024. The standard deviation of the baseline length is generally less than 1 cm. However, we found that stations whose baseline is longer than 4km exhibit larger standard deviation (typically 2 cm) and lower ambiguity fix ratio. We also found that the standard deviation changes along time in a day. This change cycle shifts -2 hours a month. Based on this fact, we concluded that this is due to the satellite constellation.
Improvements
To improve the precision, we introduced GLONASS and QZSS. The number of tracked satellites increased from a minimum of 6 to 12 by introducing GLONASS, and 14 by introducing QZSS as well. As the minimum number of tracked satellites is greatly improved, the ratio of outliers (standard deviation exceeding 1 cm) decreased from 8.2 % to 2~3 %. We also found the standard deviation of non-outliers improved slightly.
We also tested the effect of precise ephemeris using ultra-rapid product analyzed by MADOCA (Kawate et al., 2023). However, no significant improvement is observed. Since we evaluated the precision by standard deviation of baseline length every hour, this may indicate that the baseline error caused by the ephemeris error has longer time constant than an hour.
Summary
We investigated the quality of GNSS kinematic analysis using our GNSS stations in Izu-Oshima. With GPS only, the number of tracked satellites directly affects the precision, however this is compensated by introducing GLONASS and QZSS. The effect of precise ephemeris to the standard deviation of one hour baseline length change is limited.
Some of our GNSS stations record 1 Hz data and send them every 15 minutes. We are planning to apply kinematic analysis to those for further improvement of the latency.
Crustal deformation is an important measure of volcanic activities as it reflects charging of magma reservoirs, shallow hydrothermal activities, and/or pre-eruptive ascent of fluids. Among various tools of crustal deformation observation, GNSS, which can detect three-dimensional surface displacement, plays an important role in deformation monitoring. For this purpose, JMA carries out continuous GNSS observation at almost 150 stations around volcanoes in Japan. Observed GNSS data are analyzed by daily static analysis using precise ephemeris. Because of this analysis strategy, crustal deformation is obtained as one-sample-per-day data with delay of 2 weeks (final product) or 3 days (rapid product). This implies that the GNSS is used for mid- to long-term evaluation of volcanic activity. Crustal deformations of shorter timescale are monitored by tiltmeters.
However, the number of tiltmeters around the volcanoes is limited. Utilizing GNSS observations for such displacements can contribute to the early detection of the anomalous volcanic activities, and thus finer volcano monitoring. To accomplish this goal, we investigated the quality of GNSS kinematic analysis.
GNSS Kinematic Analysis
MRI have been carrying out GNSS continuous observation at 15 sites in Izu-Oshima for daily static analysis. The sampling rate of the GNSS data is 30 seconds and we use GPS signals only. GNSS data are sent to analysis server every hour.
We concatenated latest 5 hours of these hourly data and applied PPK (Post-Processing Kinematic) analysis using broadcast ephemeris. Analysis tool is rtklib (Takasu et al., 2007).
We then investigated the precision of the obtained coordinate values of Jan. 2024. The standard deviation of the baseline length is generally less than 1 cm. However, we found that stations whose baseline is longer than 4km exhibit larger standard deviation (typically 2 cm) and lower ambiguity fix ratio. We also found that the standard deviation changes along time in a day. This change cycle shifts -2 hours a month. Based on this fact, we concluded that this is due to the satellite constellation.
Improvements
To improve the precision, we introduced GLONASS and QZSS. The number of tracked satellites increased from a minimum of 6 to 12 by introducing GLONASS, and 14 by introducing QZSS as well. As the minimum number of tracked satellites is greatly improved, the ratio of outliers (standard deviation exceeding 1 cm) decreased from 8.2 % to 2~3 %. We also found the standard deviation of non-outliers improved slightly.
We also tested the effect of precise ephemeris using ultra-rapid product analyzed by MADOCA (Kawate et al., 2023). However, no significant improvement is observed. Since we evaluated the precision by standard deviation of baseline length every hour, this may indicate that the baseline error caused by the ephemeris error has longer time constant than an hour.
Summary
We investigated the quality of GNSS kinematic analysis using our GNSS stations in Izu-Oshima. With GPS only, the number of tracked satellites directly affects the precision, however this is compensated by introducing GLONASS and QZSS. The effect of precise ephemeris to the standard deviation of one hour baseline length change is limited.
Some of our GNSS stations record 1 Hz data and send them every 15 minutes. We are planning to apply kinematic analysis to those for further improvement of the latency.