In order to develop new tools or techniques to detect short-term slow slip events (S-SSEs) along subduction zones, we attempted to detect S-SSEs by conducting groundwater pressure observations. At ANO station, which is a groundwater observation station operated by the Geological Survey of Japan, the National Institute of Advanced Industrial Science and Technology, for earthquake prediction research, groundwater pressures changed due to six S-SSEs that occurred near ANO from June 2011 to April in 2013. The fault models of these S-SSEs, which were estimated mainly by observing the crustal strains and tilts, explained the changes in the groundwater pressures. If the strain sensitivity of the observed groundwater pressure or level is larger than 1 mm/nstrain and the noise level is smaller than 50 mm/day, it is possible to detect S-SSEs that occur in southwest Japan by conducting groundwater pressure or level observations.
 Nonvolcanic deep low-frequency (DLF) tremors are detected on plate boundaries along many subduction zones around the world [Obara, 2002; Ide, 2012]. Short-term slow slip events (S-SSEs), which cause small crustal deformation with no usual seismic waves, are also detected in subduction zones [Rogers and Dragert, 2003; Schwartz and Rokosky, 2007; Beroza and Ide, 2011]. There is a close spatial and temporal correlation between DLF tremors and S-SSEs. However, S-SSEs do not always occur in areas where DLF tremors occur and vice versa [Schwartz and Rokosky, 2007; Sekine et al., 2010]. Therefore, it is important to clarify detailed spatial and temporal correlations in order to know what occurs on the plate boundaries along subduction zones. This clarification will also contribute to forecasting large earthquakes in the subduction zones. In general, detecting S-SSEs via crustal deformation is more difficult than with DLF tremors when using a seismograph. One major reason for this is that the decay of crustal deformation by distance is much larger than that of seismic waves. Therefore, it is necessary to develop new tools or techniques to detect S-SSEs.
 S-SSEs are generally detected using a strainmeter or a tilt meter [e.g., Kobayashi et al., 2006; Sekine et al., 2010], and large S-SSEs also can be detected via Global Navigation Satellite System observations [Dragert et al., 2001; Nishimura et al., 2013]. However, the observation tools that are used to monitor crustal deformation are generally expensive. Therefore, they are not always popular in the areas and countries near the subduction zones. In contrast, groundwater observations are more popular in such areas and countries. If S-SSEs can be detected via groundwater observations, more knowledge will be obtained regarding plate boundary information. According to poroelastic theory [e.g., Roeloffs, 1996; Wang, 2000], the groundwater pressure or level is coupled with strain. It is known that groundwater pressure change is proportional to volumetric strain change under the undrained condition. In other words, we can estimate volumetric strain changes by observing the groundwater pressure or level. In this paper, we will show that groundwater pressure or level can be used to detect S-SSEs. We will also discuss the conditions for detecting S-SSEs when conducting groundwater pressure or level observations.
 The Geological Survey of Japan (GSJ), the National Institute of Advanced Industrial Science and Technology (AIST), has a network composed of approximately 50 groundwater observation stations in and around the Tokai, Kinki, and Shikoku regions of Japan for earthquake prediction research (Figure 1) [Matsumoto et al., 2007; Koizumi, 2013]. At these stations, groundwater levels or pressures are continuously monitored. Seventeen of these stations (N1–N16 and TYE in Figure 1), which are located in the areas facing the Nankai trough, are also equipped with borehole strainmeters. We have been monitoring S-SSEs using borehole strainmeter data since 2007 [Itaba et al., 2010; Itaba and Ando, 2011]. Observations at ANO station (N14 in Figure 1) started in February 2010.
 In the following analysis, we mainly examined and analyzed the observation data at ANO, which has three observation wells (Figure 2 and Table 1). All of the three boreholes are cased. Two of them (borehole-1 and borehole-2) are artesian wells and the heads of the waters are higher than the surface. Therefore, we sealed the two wells and monitor the groundwater pressures in the wells. Borehole-1 is equipped with a borehole strainmeter at the depth of 590 m.
3.1 Noise Level of the Groundwater Pressure Observations at ANO Station
 Figure 3 shows the observation results from February 2010 to August 2013. First, the groundwater pressures at borehole-1 and borehole-2 gradually increased since both boreholes were sealed when the observations started. In July 2010, the pressures became almost stable. It should be noted that we sometimes opened borehole-1 between January and March 2011 to check and repair the temperature meters and strainmeters. Therefore, the pressure during this period was not stable. Borehole-3 was observed to have shallow unconfined groundwater. Therefore, the groundwater level at borehole-3 was greatly affected by rainfall. During the first month of observations, the strain data were not stable due to maintenance to the borehole strainmeters. After the first month, the strain data changed exponentially following the installation of the strainmeter in borehole-1. As a result, the groundwater pressure/level data and strain data have been usable for detailed analyses since June 2011.
 The groundwater pressures at borehole-1 and borehole-2 show clear tidal changes caused by Earth tides, which indicates that they are sensitive to volumetric strain changes [e.g. Roeloffs, 1996; Wang, 2000]. There are almost no phase differences between the pressure changes and volumetric strain changes caused by the Earth tides [Kitagawa et al., 2011]. It means that groundwater pressure observation at borehole-1 and borehole-2 are under the undrained condition. The strain sensitivities of the groundwater pressures, which were estimated from the tidal groundwater pressure changes and theoretical volumetric strain changes caused by the Earth tides, are 3.1 mm/nstrain at borehole-1 and 3.7 mm/nstrain at borehole-2, where “nstrain” represents 10−9 strain. The effects of rainfall on groundwater pressure at borehole-1 and borehole-2 were eliminated using the program created by Matsumoto . As a result, the noise level of the groundwater pressure observations is very low at ANO station. The standard deviations of the 24 h difference of the groundwater pressure in 2012 are 2.9 mm/day at borehole-1 and 3.8 mm/day at borehole-2. If we regard twice the standard deviation as the noise level of the groundwater pressure observation at ANO, the noise levels are 5.8 mm/day at borehole-1 and 7.6 mm/day at borehole-2. Using the strain sensitivities mentioned above, the strain-converted noise level was estimated to be 2 nstrain/day at borehole-1 and borehole-2 at ANO station.
3.2 Groundwater Pressure Changes Related to S-SSEs
 Active DLF tremors occurred near ANO on 30 September 2012 and lasted for several days (Figures 4 and 5). The crustal strains and groundwater pressures changed at ANO from 2 October to 4 October in 2012 (Figure 4). During this same period, strains and tilts also changed at the stations near ANO. The fault model for the S-SSE (Table 2) was estimated from the strain and tilt changes at ANO and the surrounding observation stations (Figure 5) [Itaba et al., 2013a]. Using the fault model, we calculated the volumetric strain change on the surface at ANO by a program of MICAP-G [Naito and Yoshikawa, 1999], which can estimate deformation due to faults in a homogeneous elastic half space using a program of Okada . The calculated volumetric strain change is −10.8 nstrain at ANO, in which a negative value indicates that a contraction occurred. On the other hand, the volumetric strain changes calculated from the observed groundwater pressures at borehole-1 and borehole-2 for clarifying the observed value are −11.1 nstrain and −8.5 nstrain, respectively (Table 3). Therefore, the fault model explained 79–103 % of the volumetric strain changes estimated from the groundwater pressures at ANO.
Table 2. Fault Models of the S-SSEs That Occurred Near ANO From June 2011 to April 2013
Op1, Observed groundwater pressure changes at borehole-1.
Op2, Observed groundwater pressure changes at borehole-2.
Vp1, Volumetric strain changes calculated from Op1.
Vp2, Volumetric strain changes calculated from Op2.
V, Volumetric strain changes calculated from the fault model of the S-SSE.
 From June 2011 to April 2013, 12 S-SSEs were detected near ANO (Table 2) [Itaba et al., 2011, 2012, 2013a, 2013b; Kitagawa et al., 2012] and the groundwater pressure data for ANO were analyzed when the S-SSEs occurred. The results are shown in Table 2. In Table 2 “Start” means the time when change in strain and groundwater pressure at ANO started and “End” means the time when it ended. Actually these Start and End were not always clearly recognized. Therefore, there can be a difference of a few hours. The model for no.6 and no.12 (12′) were estimated from groundwater pressure data at ANO as well as the data of tilts and strains at ANO and its neighboring stations. The other models were estimated only from the data of strains and tilts. The fault models for six of the 12 S-SSEs (Nos. 1, 3, 5, 6, 10 and 12 (12′) in Table 3) explained the volumetric strain changes that were converted from the observed groundwater pressures at borehole-1 and borehole-2 in ANO (Table 3). However, the remaining six models do not explain the converted volumetric strain changes. For these cases, the average daily volumetric strain changes (V/P in Table 3) were almost equal to or smaller than 2 nstrain/day, which was the noise level mentioned above. Therefore, it is possible that the estimated groundwater pressure changes during the S-SSEs are not accurate. In other words, the groundwater pressure at ANO can be used to detect S-SSEs if S-SSEs are large enough or close enough to ANO to cause volumetric strain changes that are larger than the noise level at ANO.
 The volumetric strain changes caused by the S-SSEs in southwest Japan were 10–20 nstrain/day at most (Table 3) [Kobayashi et al., 2006]. Therefore, the strain-converted noise level should be 5 nstrain/day or smaller to detect the S-SSEs via groundwater observations. Strain sensitivity of the groundwater level or pressure is usually smaller than 10 mm/nstrain [Roeloffs, 1988, 1996; Itaba et al., 2008]. Thus, the noise level of the groundwater pressure/level observations should be smaller than 50 mm/day for the 24 h differences of the data used to detect the S-SSEs. On the other hand, the noise level of the groundwater pressure/level was usually larger than 5 mm/day for the 24 h differences. Therefore, the strain sensitivity should be larger than 1 mm/nstrain. The noise level and strain sensitivities at borehole-1 and borehole-2 in ANO satisfy these conditions even though they are generally difficult conditions to satisfy for groundwater pressure/level observations.
 Appropriate observation of the strain by borehole strainmeter in soft rock and sediment is usually difficult because borehole strainmeter is made of metal and harder than the soft rock and sediment. It means that the borehole strainmeter can hardly detect the strains of deformed soft rock and sediment. However, by observing the groundwater pressure or level, we can easily estimate volumetric strain changes in soft rock and sediment. Large areas of the land and ocean floor are covered with soft rock and sediment. Therefore, investigating the ability to detect S-SSEs via groundwater pressure/level observations is also important to magnify the observable strain area for detecting and analyzing S-SSEs throughout the world. Taking the conditions mentioned above into consideration, we will investigate other wells to detect S-SSEs in future studies.
 At the ANO observation station, the groundwater pressure changed due to the six S-SSEs that occurred near ANO from June 2011 to April 2013. The fault models of those S-SSEs, which were estimated mainly by observing the crustal strains and tilts, explained the changes in groundwater pressure. If the strain sensitivity of the observed groundwater pressure or level is larger than 1 mm/nstrain and the noise level is smaller than 50 mm/day, it is possible to detect S-SSEs in southwest Japan by conducting groundwater pressure/level observations.
 We thank S. Itaba for providing the fault models and figures of the S-SSEs and giving insightful suggestions.
 The Editor thanks Hitoshi Hirose and an anonymous reviewer for their assistance in evaluating this paper.