서브메뉴
검색
Detection and Characterization of Ionospheric Sporadic-E: A Machine Learning Approach
Detection and Characterization of Ionospheric Sporadic-E: A Machine Learning Approach
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105556
- ISBN
- 9798265403247
- DDC
- 001
- 저자명
- Ellis, Joseph.
- 서명/저자
- Detection and Characterization of Ionospheric Sporadic-E: A Machine Learning Approach
- 발행사항
- [Sl] : Georgia Institute of Technology, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 149 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: Cohen, Morris.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2024.
- 초록/해제
- 요약Sporadic-E (Es) manifests as regions of enhanced ionization, primarily occurring 90- 130 km above Earth's surface. These irregularly ionized layers can reflect or degrade radio waves propagating through the ionosphere and impact applications such as satellite and high frequency (HF) communications, Global Navigation Satellite System (GNSS) navigation and positioning, and over-the-horizon radar. In order to effectively operate in these complex electromagnetic environments, a global understanding and accurate characterization of Es is critical.To better model Es, advanced signal processing and machine learning (ML) techniques were used to develop models that are able to characterize the ionospheric phenomena in terms of occurrence, intensity, and location. Models were developed for the cases where in-situ radio occultation (RO) measurements are available, only global features are available, and a combination where input features consist of both RO measurements and global features. While RO based models perform well, addition of global features was shown to improve performance. These in-situ based models also generalize well globally to locations where training data is not available. Models featuring only global features as inputs also perform admirably on their own, even boasting the best performance for height predictions. However, these models did require implementation of dataset augmentation techniques in order to be useful globally, as there was a tendency to overfit to training locations.Additionally, a Es climatology study was carried out using the developed models which agrees with the known physics and observations, giving higher confidence in the models. Diurnal variations are present, with high intensities occurring during the day and low intensity at night. Seasonal variations are also present, with the highest intensities occurring during the summer months and lowest intensities occurring during the winter months of the respective hemisphere. Es is also much weaker in the vicinity of the magnetic equator and at high latitudes, as expected due to the unique physics in those locations.
- 일반주제명
- Software
- 일반주제명
- Meteors & meteorites
- 일반주제명
- Climate science
- 일반주제명
- Climate change
- 기본자료저록
- Dissertations Abstracts International. 87-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2024 us c eng d■001000017360617
■00520260202105556
■006m o d
■007cr#unu||||||||
■020 ▼a9798265403247
■035 ▼a(MiAaPQ)AAI32315888
■035 ▼a(MiAaPQ)GeorgiaTech76859
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a001
■1001 ▼aEllis, Joseph.
■24510▼aDetection and Characterization of Ionospheric Sporadic-E: A Machine Learning Approach
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a149 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: Cohen, Morris.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2024.
■520 ▼aSporadic-E (Es) manifests as regions of enhanced ionization, primarily occurring 90- 130 km above Earth's surface. These irregularly ionized layers can reflect or degrade radio waves propagating through the ionosphere and impact applications such as satellite and high frequency (HF) communications, Global Navigation Satellite System (GNSS) navigation and positioning, and over-the-horizon radar. In order to effectively operate in these complex electromagnetic environments, a global understanding and accurate characterization of Es is critical.To better model Es, advanced signal processing and machine learning (ML) techniques were used to develop models that are able to characterize the ionospheric phenomena in terms of occurrence, intensity, and location. Models were developed for the cases where in-situ radio occultation (RO) measurements are available, only global features are available, and a combination where input features consist of both RO measurements and global features. While RO based models perform well, addition of global features was shown to improve performance. These in-situ based models also generalize well globally to locations where training data is not available. Models featuring only global features as inputs also perform admirably on their own, even boasting the best performance for height predictions. However, these models did require implementation of dataset augmentation techniques in order to be useful globally, as there was a tendency to overfit to training locations.Additionally, a Es climatology study was carried out using the developed models which agrees with the known physics and observations, giving higher confidence in the models. Diurnal variations are present, with high intensities occurring during the day and low intensity at night. Seasonal variations are also present, with the highest intensities occurring during the summer months and lowest intensities occurring during the winter months of the respective hemisphere. Es is also much weaker in the vicinity of the magnetic equator and at high latitudes, as expected due to the unique physics in those locations.
■590 ▼aSchool code: 0078.
■650 4▼aSoftware
■650 4▼aMeteors & meteorites
■650 4▼aClimate science
■650 4▼aClimate change
■690 ▼a0800
■690 ▼a0404
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05B.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360617▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
