{"id":{"repo_id":"gmu","oai_identifier":"oai:MARS:1920/14734"},"canonical_url":"https://search.dev.ndltd.org/etd/gmu/oai:MARS:1920/14734","repository":{"repo_id":"gmu","name":"George Mason University","base_url":"https://mars.gmu.edu/server/oai/request"},"display":{"title":"INVESTIGATION OF SPATIOTEMPORAL PATTERNS OF VOLCANIC ERUPTIONS ON GLOBAL AND REGIONAL SCALES","abstract":"Throughout the study of Earth’s volcanism, scientists have utilized a combination of both in person data collection methods and remotely sensed data to analyze volcanic eruptive patterns. These eruptive patterns have concentrated predominantly on the study of individual volcanoes or on smaller regions of volcanic activity defined by their similar tectonic backgrounds to deduce potential eruptive activity. From a planetary science perspective, Earth as well as other planetary bodies can be viewed as a single system, allowing the examination and analysis of potential volcanic patterns at global and regional scales. As a mechanism to study global and regional patterns in volcanic activity, data science provides the ability to deduce such patterns via basic statistical and time series analysis methods. These techniques give insight into the potential seasonality, clustering, and correlations associated with each variable that can be used to construct behavioral models of activity. This study examines three variables in the context of a time series spanning from 1958-2019 to determine if seasonality, clustering, and correlations exist between the variables. The Volcanic Explosivity Index (VEI), which measures the magnitude of eruption, is the first of these variables and reflects the intensity of a volcanic eruption. The second metric examines the number of elapsed days occurring between successive start dates of each eruption, and the third metric is the Haversine distance between two successive eruptive events. While geologically and physically speaking, the relationships between these variables depend upon the tectonic setting and magma composition associated with individual or small groups of volcanoes, the data science perspective considers each eruptive event as a stochastic occurrence unbound by tectonic and physical constraints. The advantage of this perspective is that it can illuminate possible patterns within the time series dataset that can be integrated into physical modeling efforts to improve potential volcanic eruption forecasting. Given the sporadic nature of volcanic eruptions, this study also compares the traditionally based modeling technique of multiple linear regression against a non-linear random forest method, a geostatistical Empirical Bayesian Kriging method and the Teunter-Syntetos-Babai (TSB) Method used with intermittent time series analysis to determine which method is most appropriate for use globally and with fourteen regional areas. The most notable findings of this study are non-linear random forest testing and clustering techniques best modeled eruptive behavior at global and regional scales, and Pearson correlation revealed that cyclicity of eruptive behavior exists, particularly at regional levels. TSB modeling did prove successful in regions where one volcano dominated the dataset and there were less than 70 eruptions.","abstract_html":"Throughout the study of Earth’s volcanism, scientists have utilized a combination of both in person data collection methods and remotely sensed data to analyze volcanic eruptive patterns. These eruptive patterns have concentrated predominantly on the study of individual volcanoes or on smaller regions of volcanic activity defined by their similar tectonic backgrounds to deduce potential eruptive activity. From a planetary science perspective, Earth as well as other planetary bodies can be viewed as a single system, allowing the examination and analysis of potential volcanic patterns at global and regional scales. As a mechanism to study global and regional patterns in volcanic activity, data science provides the ability to deduce such patterns via basic statistical and time series analysis methods. These techniques give insight into the potential seasonality, clustering, and correlations associated with each variable that can be used to construct behavioral models of activity. This study examines three variables in the context of a time series spanning from 1958-2019 to determine if seasonality, clustering, and correlations exist between the variables. The Volcanic Explosivity Index (VEI), which measures the magnitude of eruption, is the first of these variables and reflects the intensity of a volcanic eruption. The second metric examines the number of elapsed days occurring between successive start dates of each eruption, and the third metric is the Haversine distance between two successive eruptive events. While geologically and physically speaking, the relationships between these variables depend upon the tectonic setting and magma composition associated with individual or small groups of volcanoes, the data science perspective considers each eruptive event as a stochastic occurrence unbound by tectonic and physical constraints. The advantage of this perspective is that it can illuminate possible patterns within the time series dataset that can be integrated into physical modeling efforts to improve potential volcanic eruption forecasting. Given the sporadic nature of volcanic eruptions, this study also compares the traditionally based modeling technique of multiple linear regression against a non-linear random forest method, a geostatistical Empirical Bayesian Kriging method and the Teunter-Syntetos-Babai (TSB) Method used with intermittent time series analysis to determine which method is most appropriate for use globally and with fourteen regional areas. The most notable findings of this study are non-linear random forest testing and clustering techniques best modeled eruptive behavior at global and regional scales, and Pearson correlation revealed that cyclicity of eruptive behavior exists, particularly at regional levels. TSB modeling did prove successful in regions where one volcano dominated the dataset and there were less than 70 eruptions.","abstract_has_math":false,"creators":["Mandable, Lori"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-07-27T19:52:12Z","subjects":["Clustering Analysis","Eruptive Patterns","Spatiotemporal Analysis","Volcanic Eruptions","Volcanic Regions"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["hdl:1920/14734"],"render_values":[{"text":"hdl:1920/14734","href":null,"code":true}]}]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null},"metadata_groups":[{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:type","label":"Dc Type","values":["Dissertation"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Clustering Analysis","Eruptive Patterns","Spatiotemporal Analysis","Volcanic Eruptions","Volcanic Regions"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["hdl:1920/14734"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.other","label":"Dc Description Other","values":["Throughout the study of Earth’s volcanism, scientists have utilized a combination of both in person data collection methods and remotely sensed data to analyze volcanic eruptive patterns. These eruptive patterns have concentrated predominantly on the study of individual volcanoes or on smaller regions of volcanic activity defined by their similar tectonic backgrounds to deduce potential eruptive activity. From a planetary science perspective, Earth as well as other planetary bodies can be viewed as a single system, allowing the examination and analysis of potential volcanic patterns at global and regional scales. As a mechanism to study global and regional patterns in volcanic activity, data science provides the ability to deduce such patterns via basic statistical and time series analysis methods. These techniques give insight into the potential seasonality, clustering, and correlations associated with each variable that can be used to construct behavioral models of activity. This study examines three variables in the context of a time series spanning from 1958-2019 to determine if seasonality, clustering, and correlations exist between the variables. The Volcanic Explosivity Index (VEI), which measures the magnitude of eruption, is the first of these variables and reflects the intensity of a volcanic eruption. The second metric examines the number of elapsed days occurring between successive start dates of each eruption, and the third metric is the Haversine distance between two successive eruptive events. While geologically and physically speaking, the relationships between these variables depend upon the tectonic setting and magma composition associated with individual or small groups of volcanoes, the data science perspective considers each eruptive event as a stochastic occurrence unbound by tectonic and physical constraints. The advantage of this perspective is that it can illuminate possible patterns within the time series dataset that can be integrated into physical modeling efforts to improve potential volcanic eruption forecasting. Given the sporadic nature of volcanic eruptions, this study also compares the traditionally based modeling technique of multiple linear regression against a non-linear random forest method, a geostatistical Empirical Bayesian Kriging method and the Teunter-Syntetos-Babai (TSB) Method used with intermittent time series analysis to determine which method is most appropriate for use globally and with fourteen regional areas. The most notable findings of this study are non-linear random forest testing and clustering techniques best modeled eruptive behavior at global and regional scales, and Pearson correlation revealed that cyclicity of eruptive behavior exists, particularly at regional levels. TSB modeling did prove successful in regions where one volcano dominated the dataset and there were less than 70 eruptions."]},{"key":"dc:title","label":"Title","values":["INVESTIGATION OF SPATIOTEMPORAL PATTERNS OF VOLCANIC ERUPTIONS ON GLOBAL AND REGIONAL SCALES"]}]}],"canonical_facts":{"dc:date.issued":["2025"],"dc:description.other":["Throughout the study of Earth’s volcanism, scientists have utilized a combination of both in person data collection methods and remotely sensed data to analyze volcanic eruptive patterns. These eruptive patterns have concentrated predominantly on the study of individual volcanoes or on smaller regions of volcanic activity defined by their similar tectonic backgrounds to deduce potential eruptive activity. From a planetary science perspective, Earth as well as other planetary bodies can be viewed as a single system, allowing the examination and analysis of potential volcanic patterns at global and regional scales. As a mechanism to study global and regional patterns in volcanic activity, data science provides the ability to deduce such patterns via basic statistical and time series analysis methods. These techniques give insight into the potential seasonality, clustering, and correlations associated with each variable that can be used to construct behavioral models of activity. This study examines three variables in the context of a time series spanning from 1958-2019 to determine if seasonality, clustering, and correlations exist between the variables. The Volcanic Explosivity Index (VEI), which measures the magnitude of eruption, is the first of these variables and reflects the intensity of a volcanic eruption. The second metric examines the number of elapsed days occurring between successive start dates of each eruption, and the third metric is the Haversine distance between two successive eruptive events. While geologically and physically speaking, the relationships between these variables depend upon the tectonic setting and magma composition associated with individual or small groups of volcanoes, the data science perspective considers each eruptive event as a stochastic occurrence unbound by tectonic and physical constraints. The advantage of this perspective is that it can illuminate possible patterns within the time series dataset that can be integrated into physical modeling efforts to improve potential volcanic eruption forecasting. Given the sporadic nature of volcanic eruptions, this study also compares the traditionally based modeling technique of multiple linear regression against a non-linear random forest method, a geostatistical Empirical Bayesian Kriging method and the Teunter-Syntetos-Babai (TSB) Method used with intermittent time series analysis to determine which method is most appropriate for use globally and with fourteen regional areas. The most notable findings of this study are non-linear random forest testing and clustering techniques best modeled eruptive behavior at global and regional scales, and Pearson correlation revealed that cyclicity of eruptive behavior exists, particularly at regional levels. TSB modeling did prove successful in regions where one volcano dominated the dataset and there were less than 70 eruptions."],"dc:identifier":["hdl:1920/14734"],"dc:subject":["Clustering Analysis","Eruptive Patterns","Spatiotemporal Analysis","Volcanic Eruptions","Volcanic Regions"],"dc:title":["INVESTIGATION OF SPATIOTEMPORAL PATTERNS OF VOLCANIC ERUPTIONS ON GLOBAL AND REGIONAL SCALES"],"dc:type":["Dissertation"]},"updated_at":"2026-07-27T19:52:12Z"}