1:45 PM - 3:15 PM
[SCG59-P01] Principal component analysis and its geological implication from geo chemical data set of the hot springs in Kyushu, Japan
Keywords:hot springs, big data, multivariate analysis
1. Introduction
The Japanese archipelago is located in active subduction zone. Monitoring hot and cold spring water is useful to understand crustal activity. Although there are many hot spring sources in Japan, no comprehensive database exists. The classification of hot spring is usually based on the combinations of major dissolved ions (Na+K, Ca, Mg, Cl-, SO4-, and HCO3). We collected and compiled the analytical reports of chemical component data for the hot springs in Kyushu and performed big data analysis using QGIS and statistical analysis methods to clarify the regional characteristics and proposed a new classification method.
2.Method
We collected the analytical report of hot spring through our fieldwork and through requests for information disclosure to prefectures and cities in Kyushu and Okinawa regions. We classified the springs according to temperature, liquid properties, and chemical composition, and the results were plotted on a 1:200,000 seamless geological map and a white map provided by the National Institute of Advanced Industrial Science and Technology (AIST) using QGIS. We also drew a trilinear diagram out of the dataset.
Principal component analysis was performed using the R language.
3.Results and Discussion
(1) Database construction
1,742 paper-based hot spring analysis reports were obtained. A database of 72 data items such as source name, source location, analysis date, spring temperature, pH, cations, anions, etc. was created and digitized. Latitude and longitude were determined from the source location using Geocoding.jp API version 1.2.
(2) Regional analysis based on database analysis
The spring temperatures tended to be higher around active volcanoes and Quaternary volcanoes. In Fukuoka, Saga and Okinawa prefectures, there were no active volcanoes, but hot springs with high spring temperatures were observed. The relationship with the temperature gradient is considered, but it is unclear because there is no information on the depth of drilling in the hot spring analysis report.
Unzen (Nagasaki Prefecture), Kuju, Yufudake, Tsurumidake and Kyan-dake (Oita Prefecture), Kirishima, Yonemaru, Sumiyoshi Pond, Sakurajima, Ikeda and Yamakawa (Kagoshima Prefecture). Acidic springs in other areas tended to have low temperatures. Neutral and alkaline springs were found in various areas of Kyushu and Okinawa.
The Na-Cl type springs were mostly found along the coast, while inland they were concentrated in sedimentary rock formations. Na-SO4, Ca-SO4, Mg-SO4, Ca-HCO3, and Mg-HCO3 types were found around volcanoes. These are considered to be hot springs affected by volcanoes. In Kyushu, Na-HCO3 type springs were the most abundant and widely distributed in both sedimentary and igneous formations, while Ca-Cl type springs were abundant around the Kego Fault in Fukuoka City and are thought to be the result of ion exchange between Na and Ca in seawater passing through the Kego Fault.
(3) Proposal of a new classification method based on principal component analysis for database construction
The most predominant principal components PC1 shows the concentration of major dissolved ions and Mn, PC2 shows the concentration of Al, Fe, and Fe, PC3 shows the conflicting nature of free carbon dioxide gas and dissolved carbonate ions, and PC4 shows the concentration of S2O3- and HS- and free gas. The results were similar to those of PC1 and PC2. The contribution of PC1 to PC4 was 21.93%, 10.62%, 8.329%, and 6.82%, in that order. The results of the principal component analysis enabled a new classification.
4.Conclusion
In this study, a database was constructed using hot spring analysis sheets from the Kyushu region, and analyzed as big data of geospatial information. The regional analysis revealed certain trends in spring quality and geology. Differences related to crustal activities such as faults and volcanoes were also found. Furthermore, principal component analysis has made it possible to classify hot springs using the components that contribute substantially to the classification. Unlike conventional classification using major ions, this method is considered effective for geological interpretation.
The Japanese archipelago is located in active subduction zone. Monitoring hot and cold spring water is useful to understand crustal activity. Although there are many hot spring sources in Japan, no comprehensive database exists. The classification of hot spring is usually based on the combinations of major dissolved ions (Na+K, Ca, Mg, Cl-, SO4-, and HCO3). We collected and compiled the analytical reports of chemical component data for the hot springs in Kyushu and performed big data analysis using QGIS and statistical analysis methods to clarify the regional characteristics and proposed a new classification method.
2.Method
We collected the analytical report of hot spring through our fieldwork and through requests for information disclosure to prefectures and cities in Kyushu and Okinawa regions. We classified the springs according to temperature, liquid properties, and chemical composition, and the results were plotted on a 1:200,000 seamless geological map and a white map provided by the National Institute of Advanced Industrial Science and Technology (AIST) using QGIS. We also drew a trilinear diagram out of the dataset.
Principal component analysis was performed using the R language.
3.Results and Discussion
(1) Database construction
1,742 paper-based hot spring analysis reports were obtained. A database of 72 data items such as source name, source location, analysis date, spring temperature, pH, cations, anions, etc. was created and digitized. Latitude and longitude were determined from the source location using Geocoding.jp API version 1.2.
(2) Regional analysis based on database analysis
The spring temperatures tended to be higher around active volcanoes and Quaternary volcanoes. In Fukuoka, Saga and Okinawa prefectures, there were no active volcanoes, but hot springs with high spring temperatures were observed. The relationship with the temperature gradient is considered, but it is unclear because there is no information on the depth of drilling in the hot spring analysis report.
Unzen (Nagasaki Prefecture), Kuju, Yufudake, Tsurumidake and Kyan-dake (Oita Prefecture), Kirishima, Yonemaru, Sumiyoshi Pond, Sakurajima, Ikeda and Yamakawa (Kagoshima Prefecture). Acidic springs in other areas tended to have low temperatures. Neutral and alkaline springs were found in various areas of Kyushu and Okinawa.
The Na-Cl type springs were mostly found along the coast, while inland they were concentrated in sedimentary rock formations. Na-SO4, Ca-SO4, Mg-SO4, Ca-HCO3, and Mg-HCO3 types were found around volcanoes. These are considered to be hot springs affected by volcanoes. In Kyushu, Na-HCO3 type springs were the most abundant and widely distributed in both sedimentary and igneous formations, while Ca-Cl type springs were abundant around the Kego Fault in Fukuoka City and are thought to be the result of ion exchange between Na and Ca in seawater passing through the Kego Fault.
(3) Proposal of a new classification method based on principal component analysis for database construction
The most predominant principal components PC1 shows the concentration of major dissolved ions and Mn, PC2 shows the concentration of Al, Fe, and Fe, PC3 shows the conflicting nature of free carbon dioxide gas and dissolved carbonate ions, and PC4 shows the concentration of S2O3- and HS- and free gas. The results were similar to those of PC1 and PC2. The contribution of PC1 to PC4 was 21.93%, 10.62%, 8.329%, and 6.82%, in that order. The results of the principal component analysis enabled a new classification.
4.Conclusion
In this study, a database was constructed using hot spring analysis sheets from the Kyushu region, and analyzed as big data of geospatial information. The regional analysis revealed certain trends in spring quality and geology. Differences related to crustal activities such as faults and volcanoes were also found. Furthermore, principal component analysis has made it possible to classify hot springs using the components that contribute substantially to the classification. Unlike conventional classification using major ions, this method is considered effective for geological interpretation.