著者

山内 寛之

出版者

麻布大学

巻号頁・発行日

2014

---

Many countries, including Japan, have suffered from extensive damage due to earthquakes (EQs). In Japan, there is the possibility that more large EQs, such as the Tonankai and Nankai EQs, may occur in the near future. The Headquarters for Earthquake Research Promotion assumes that these EQs would occur with a magnitude (M) of 8 to 9. If the date, location and the M of such EQs could be predicted, then it would reduce the human loss and material damages. There are reports of anomalies related to various physical and/or chemical phenomena occurring from a few months to the last minutes before EQs. Many researchers have studied the use of such unusual phenomena in making short-term EQ predictions. However, the application of these predictions is regarded as being extremely difficult at the present time. Precursors of EQs include not only changes or abnormalities in physical and chemical phenomena but also changes in various animal behaviors. These unusual animal behaviors prior to EQs are considered the results of stress or emotional responses to some physical or chemical variation. If we could identify what phenomena trigger the unusual animal behaviors prior to EQs, then we could measure and observe these phenomena by machines to predict EQs. However, clear mechanisms behind the unusual animal behaviors have remained unidentified. The object of this study was to verify whether animal behavioral changes prior to EQs are useful for predicting EQs. To achieve our objective, we surveyed and observed these behavioral changes using various methods. Chapter 1.The cross investigation into unusual animal behaviors preceding the 2011 earthquake off the Pacific Coast of Tohoku, JapanOn 11 March 2011, a megathrust EQ with moment magnitude (Mw) of 9.0 occurred in the northwestern Pacific Ocean at a shallow depth of 24 km, which is formally named “The 2011 off the Pacific coast of Tohoku Earthquake” (Tohoku EQ). Using a web questionnaire, we surveyed whether any unusual animal behaviors were noted prior to this large EQ, and, if so, what kinds of behaviors. The survey was limited to the owners of domestic dogs and cats. In the results, 236 of 1,259 dog owners (18.7%) and 115 of 703 cat owners (16.4%) observed behavioral changes. Most of the unusual animal behaviors occurred within one day of the EQ. In the dogs and cats, 23.3% and 25.0%, respectively, of the total number of behavioral changes appeared one or more days before the EQ. A “restless behavior” was the most common change noted in both animal species (22.5% in the dogs and 23.7% in the cats), suggesting that it is important to quantify this behavior. Chapter 2.The longitudinal survey on the relationship between activities in animals and EQsBased on the results of Chapter 1, we verified the relationship between the “restless behavior” and various EQs by longitudinal observation. We measured the daily activities of two cats, one horse and one turtle as an index of restless behavior using infrared sensors for one year. In the results, the daily activities of each animal increased significantly in the week before some EQs with Ms greater than 6.0. Specifically, the activity level of one cat increased significantly within the week prior to three out of four EQs. The results of time series’ analyses indicated that the activity levels in the cats and the turtle increased significantly in the week prior to seismically active dates with effective magnitudes (Meffs) greater than 5.5, as theorized by Maekawa et al. (2006). Chapter 3.The longitudinal survey on the relationship between the productivity of farm animals and EQsSection 1.The changes in dairy cows’ milk yields as a precursor of EQsWe focused on dairy cows’ milk yields. If dairy cows responded to stimuli emerging before EQs, then the milk yields might change, because the milk yields decrease when dairy cows are stressed by various stimuli. To clarify whether the milk yields decrease prior to EQs, we verified the relationship between various EQs and the individual daily milk yields measured at three institutes of animal industry located in different regions. The observation period was approximately two years. Because the milk yields of cows vary depending on the number of days after calving and environmental conditions, such as temperature and humidity, we removed these factors affecting the milk yields before verification of the relevance to EQs. In the results, a milking facility closest to the epicenter (340 km) showed significantly lower milk yields within three to six days before the Tohoku EQ (P ≤ 0.001). The relationship between the milk yields and various EQs showed that the daily mean values of the milk yields in all of the institutes decreased significantly within 15 days prior to seismically active dates with Meffs greater than 5.0 (M = 6.0 ± 0.1). In addition, the decrease in the milk yields correlated with increasing Ms. We concluded that the dairy cow milk yields, which were measured every day in wide regions, could contribute to the ability to predict EQs. Section 2.The changes in laying hens egg production as a precursor of EQsIn addition to the milk yields of dairy cows, the egg production of laying hens are also recorded every day in poultry farms and decrease in response to stress stimuli. If the daily egg production changes prior to EQs, then they might also contribute to the ability to predict EQs. In this section, using the flocks’ egg production rate data for three years from four poultry farms located in different regions, we verified the egg production rates’ relevance to many EQs. Prior to performing the analysis, we removed known factors that affect egg production, including the hens’ age and environmental factors, such as temperature and humidity, to verify the relevance to EQs. The results of the investigation as to whether the egg production rates changed prior to the Tohoku EQ revealed that the rate on the farm closest to the epicenter (390 km) decreased significantly seven and nine days before the EQ. The results with respect to many EQs indicated that the daily mean egg production rate in all of the farms decreased significantly within 27 days prior to seismically active dates with Meffs greater than 5.0 (M = 6.2 ± 0.1). Furthermore, the decrease in the egg production rates positively correlated with the distances from the epicenters. Based on these results, we devised some prototype monitoring systems to predict EQs using animal behaviors. First, to allow for the unified collection of the production data in farm animals, we devised a collection system using cloud computing. Second, we focused on an activity meter that could automatically upload data on the activities of companion animals to the server. We are currently testing the applications for animals in households. Precursors of EQs include various phenomena other than animal behaviors, such as electromagnetic anomalies and changes in radon concentrations. Some of these phenomena are being continuously monitored for EQ predictions. This study indicated that the unusual animal behaviors had a certain degree of statistical relevance to EQs; however, predictions based only on animals might result in many false alarms because there were noted behavioral changes that were irrelevant to the EQs. Therefore, the accuracy of EQ predictions might be increased by the use of varied information. We devised the following methods for EQ predictions based on the results of this study and earlier research: I. Forecast impending EQs with a certain degree of probability using the milk yields, the egg production rates, and the physical and chemical phenomena. II. Predict detailed probabilities and dates of EQs by paying more attention to precursors, such as the behavioral changes in the companion animals, when anomalies have emerged in part I.We concluded that this system is the best possible method for predicting impending EQs utilizing the animal behaviors. Furthermore, it could forecast the EQs location based on the presence or absence of the behavioral anomalies at widely dispersed observation points, and the M of the EQ from the degree of the behavioral anomaly at each observation point. To realize this system in the future, interdisciplinary interactions between various special agencies will be necessary.