Assessing seasonal and interannual water storage variations in Taiwan using geodetic and hydrological data

Synchronized and asynchronous modulation of seismicity by hydrological loading: A case study in Taiwan

“…Deep earthquakes show peak occurrence in March to April, between the time periods with the lowest GNSS-EWT and groundwater level ( Table 1 ). The GNSS-EWT may better reflect the temporal variation of total water storage, because peak groundwater storage likely lags behind the peak land water storage by about 2 months ( 14 ).…”

“…Taiwan is located on the convergent boundary where the Philippine Sea plate is colliding with the Eurasian plate at a rate of 85 to 90 mm/year ( 11 , 12 ). This region has both frequent damaging earthquakes, with about one magnitude 6+ earthquake each year ( 13 ) and heavy seasonal precipitation of more than 2000 mm/year on average ( 14 ). The largest annual precipitation is close to 4 m, and the spatiotemporal distribution of rainfall is uneven ( Fig.…”

“…About 70% of the annual precipitation occurs from May to September and is primarily contributed from monsoons and typhoons. The seasonal water cycle causes annual fluctuations of 5 to 15 m in groundwater levels and 5 to 20 mm in Global Navigation Satellite System (GNSS) vertical displacement time series, reflecting the Earth’s elastic response to the seasonal water loading ( 14 ). Taking advantage of active seismicity ( Fig.…”

“… ( A ) Difference of rainfall between the wet (month-month) and dry (month-month) seasons modified from figure 1 in ( 14 ). Red lines denote the boundaries of six physiographic regions in Taiwan: From west to east, these are the Coastal Plain (CP), the Western Foothills (WF), the Hsueshan Range (HR), the Central Range (CR), the Longitudinal Valley (LV), and the Coastal Range (CoR).…”

“…Principal components analysis (PCA) is used to extract common hydrological signals in monthly time series of 40 groundwater level stations and 40 GNSS stations with continuous data in 2002–2018. The GNSS vertical position time series are converted to the effective water thickness (GNSS-EWT) using the method described by Hsu et al ( 14 ). We next use empirical mode decomposition (EMD) to reconstruct groundwater level, GNSS vertical position, and seismicity rate time series into annual and interannual variations according to the frequency bandwidth of each intrinsic mode function (IMF).…”