東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Earth's Radiation Belts: From Machin...

Ma, Donglai.

FindBook

Google Book

Amazon

博客來

Earth's Radiation Belts: From Machine Learning to Physical Understanding.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Earth's Radiation Belts: From Machine Learning to Physical Understanding./
作者:	Ma, Donglai.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2023,
面頁冊數:	165 p.
附註:	Source: Dissertations Abstracts International, Volume: 85-08, Section: B.
Contained By:	Dissertations Abstracts International85-08B.
標題:	Geophysics. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30993822
ISBN:	9798381736458

Earth's Radiation Belts: From Machine Learning to Physical Understanding.
Ma, Donglai.

Earth's Radiation Belts: From Machine Learning to Physical Understanding. - Ann Arbor : ProQuest Dissertations & Theses, 2023 - 165 p.

Source: Dissertations Abstracts International, Volume: 85-08, Section: B.

Thesis (Ph.D.)--University of California, Los Angeles, 2023.

This dissertation explores how machine learning can be used to study the Earth's radiation belts and the physical conclusions we derive from machine learning. To be concrete, the Earth's radiation belts contain many high-energy electrons, with their energies ranging from kilo-electron volts (keV) to several Mega-electron volts (MeV). This radiation environment, exhibiting rich dynamical variations, is known to be particularly hazardous to spacecraft and is difficult to predict, particularly because of the delicate balance between acceleration, transport, and loss, combined with the many different physical processes that produce these effects. With high-quality data from the Van Allen Probes mission, we present a set of machine-learning-based models of electron fluxes ranging from 50 keV to several MeV using a neural network approach in the Earth's outer radiation belt. The Outer RadIation belt Electron Neural neT model (ORIENT) uses only a few days of the history of solar wind conditions and geomagnetic indices as input. The models show great performance ($R.

ISBN: 9798381736458Subjects--Topical Terms:

535228
Geophysics.
Subjects--Index Terms:

Machine learning

Earth's Radiation Belts: From Machine Learning to Physical Understanding.
LDR:04516nmm a2200373 4500 001 2398358
005 20240812064620.5
006 m o d
007 cr#unu||||||||
008 251215s2023 ||||||||||||||||| ||eng d
020 $a 9798381736458
035 $a (MiAaPQ)AAI30993822
035 $a AAI30993822
040 $a MiAaPQ $c MiAaPQ
100 1 $a Ma, Donglai. $3 3768267
245 1 0 $a Earth's Radiation Belts: From Machine Learning to Physical Understanding.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2023
300 $a 165 p.
500 $a Source: Dissertations Abstracts International, Volume: 85-08, Section: B.
500 $a Advisor: Bortnik, Jacob.
502 $a Thesis (Ph.D.)--University of California, Los Angeles, 2023.
520 $a This dissertation explores how machine learning can be used to study the Earth's radiation belts and the physical conclusions we derive from machine learning. To be concrete, the Earth's radiation belts contain many high-energy electrons, with their energies ranging from kilo-electron volts (keV) to several Mega-electron volts (MeV). This radiation environment, exhibiting rich dynamical variations, is known to be particularly hazardous to spacecraft and is difficult to predict, particularly because of the delicate balance between acceleration, transport, and loss, combined with the many different physical processes that produce these effects. With high-quality data from the Van Allen Probes mission, we present a set of machine-learning-based models of electron fluxes ranging from 50 keV to several MeV using a neural network approach in the Earth's outer radiation belt. The Outer RadIation belt Electron Neural neT model (ORIENT) uses only a few days of the history of solar wind conditions and geomagnetic indices as input. The models show great performance ($R.
520 $a 2 \\sim 0.7-0.9$) on the out-of-sample dataset and are able to capture electron dynamics such as intensifications, decays, dropouts, and the Magnetic Local Time dependence of the lower energy ($\\sim < 100$ keV) electron fluxes during storms. Motivated by the great performance of the machine learning model, we realize that the trained model contain the information of the repeated magnetospheric dynamics driven by solar activity. Thus, we utilize a state-of-the-art feature attribution method called DeepSHAP, which was based on Shapley values in game theory, to explain the behavior of the ORIENT model at a representative electron energy of $\\sim 1$ MeV during a storm time event and a non-storm time event. The results show that the feature importance calculated from the purely data-driven ORIENT model identifies physically meaningful behaviors such as magnetopause shadowing, substorm-driven acceleration, and Dst effect. We then combine this method with superposed epoch analysis to identify the long-debated question: What causes the radiation belt electrons to have two different responses, namely `enhancement' and `depletion' to storms? Our feature attribution results indicate that the depletion events can be thought of essentially as "non-acceleration" events that occur when substorm activity following the pressure maximum is not sufficient to accelerate the fluxes above its pre-storm level. The results show that average AL over storm-time period and recovery phase has a significant correlation with the resulting flux levels suggesting that it is important to incorporate the AL index history more directly into the radiation belt modeling. We then turn back to physics and build the statistical model of waves and density related to the AL index to create a Fokker-Planck simulation driven by time-varying geomagnetic activity. The result reproduces the enhancement of electrons at the ultra-relativistic range very well. The observations and simulations of 186 events illustrate a clear threshold on integral AL increasing with energy and demonstrate this is due to the influence of substorm activity on wave intensity and the density of background electrons. We demonstrate that the continuously elevated substorm activity is the determining feature needed for ultra-relativistic electron acceleration.
590 $a School code: 0031.
650 4 $a Geophysics. $3 535228
650 4 $a Plasma physics. $3 3175417
653 $a Machine learning
653 $a Radiation belts
653 $a Electron fluxes
690 $a 0373
690 $a 0759
690 $a 0467
710 2 $a University of California, Los Angeles. $b Atmospheric & Oceanic Sciences 002E. $3 3280774
773 0 $t Dissertations Abstracts International $g 85-08B.
790 $a 0031
791 $a Ph.D.
792 $a 2023
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30993822