Comparison of Regression Methods to Compute Atmospheric Pressure and Earth Tidal Coefficients in Water Level Associated with Wenchuan Earthquake of 12 May 2008

Abstract

The earth tide, atmospheric pressure, precipitation and earthquake fluctuations, especially earthquake greatly impacts water well levels, thus anomalous co-seismic changes in ground water levels have been observed. In this paper, we have used four different models, simple linear regression (SLR), multiple linear regression (MLR), principal component analysis (PCA) and partial least squares (PLS) to compute the atmospheric pressure and earth tidal effects on water level. Furthermore, we have used the Akaike information criterion (AIC) to study the performance of various models. Based on the lowest AIC and sum of squares for error values, the best estimate of the effects of atmospheric pressure and earth tide on water level is found using the MLR model. However, MLR model does not provide multicollinearity between inputs, as a result the atmospheric pressure and earth tidal response coefficients fail to reflect the mechanisms associated with the groundwater level fluctuations. On the premise of solving serious multicollinearity of inputs, PLS model shows the minimum AIC value. The atmospheric pressure and earth tidal response coefficients show close response with the observation using PLS model. The atmospheric pressure and the earth tidal response coefficients are found to be sensitive to the stress-strain state using the observed data for the period 1 April-8 June 2008 of Chuan 03# well. The transient enhancement of porosity of rock mass around Chuan 03# well associated with the Wenchuan earthquake (Mw = 7.9 of 12 May 2008) that has taken its original pre-seismic level after 13 days indicates that the co-seismic sharp rise of water well could be induced by static stress change, rather than development of new fractures.

Hydrogeological Characterization of Coastal Aquifer on the Basis of Observed Sea Level and Groundwater Level Fluctuations: Neretva Valley Aquifer, Croatia

“…If the external stress represented by barometric pressure is transmitted to the aquifer, reaction stress can be divided between the sediments that the aquifer is made of and water contained within the aquifer pore volume. Because BE represents the percentage of stress that is accommodated by the sediment matrix [43,44,50], a zero BE value corresponds to no response case, whereas a unity BE value respresents a fully efficient response to barometric changes. In case of a zero BE value, total barometric pressure-induced stress is accommodated by the pore water, whereas for a unity BE value, total stress is accommodated purely by sediment matrix.…”

“…Because some of the tidal methods imply the knowledge of loading efficiency [9,14,26], barometric efficiency, BE, has to be determined to characterize tidally induced changes inland from the shoreline. The topic of excluding the effects of barometric pressure from observed groundwater level was of interest in several publications [41][42][43][44][45][46][47][48][49][50][51], whereas BE determination can be approached by time domain or frequency domain analysis [50].…”

Groundwater level response to the Wenchuan earthquake of May 2008

“…The water level dynamic can often be expressed as (Matsumoto 1992;Matsumoto et al 2003;He et al 2016):…”

“…Since then, digital and automatic instrumental records of water level changes during earthquakes have become available, and significant achievements have been obtained in terms of quantitative analyses (Cooper et al 1965;Roeloffs 1998;Wang and Manga 2010). With an abundance of observational data, the co-and post-seismic changes in the water levels of wells have been reported and are among the most widely documented earthquake-induced hydrologic phenomena (Wang and Manga 2010); for example, the changes in groundwater and the mechanisms associated with earthquakes have been studied in the U.S. (Rojstaczer and Wolf 1992;Peltzer et al 1996), Japan (Wakita 1975;Akita and Matsumoto 2004;Itaba et al 2008), India (Ramana et al 2007;Radhakrishnan et al 2010), Europe (Grecksch et al 1999), and China (Lee et al 2002(Lee et al , 2004Zheng et al 2012;Shi et al 2013;Sun et al 2015;He et al 2016He et al , 2017aHe et al , 2017b. There are four possible causal mechanisms that we have summarized from existing studies.…”

Comparing Methods of Barometric Efficiency Characterization for Specific Storage Estimation

“…Butler et al (2011) described the physical processes involved in the transmission of barometric pressure loading from the ground surface (1) and (2) to an unconfined aquifer. The same concepts were depicted visually by He et al (2016). The descriptions provided by Butler et al (2011) are paraphrased as follows.…”