본문

서브메뉴

Statistical Approaches for Understanding Atmospheric Dynamics From Noisy Data
Statistical Approaches for Understanding Atmospheric Dynamics From Noisy Data
Statistical Approaches for Understanding Atmospheric Dynamics From Noisy Data

상세정보

자료유형  
 학위논문 서양
최종처리일시  
20260202104959
ISBN  
9798293841486
DDC  
551.5
저자명  
Vishny, David N.
서명/저자  
Statistical Approaches for Understanding Atmospheric Dynamics From Noisy Data
발행사항  
[Sl] : University of California, San Diego, 2025
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2025
형태사항  
145 p
주기사항  
Source: Dissertations Abstracts International, Volume: 87-03, Section: B.
주기사항  
Advisor: Lutsko, Nicholas J.
학위논문주기  
Thesis (Ph.D.)--University of California, San Diego, 2025.
초록/해제  
요약The chaotic nature of the atmosphere limits its predictability, while the sparsity and measurement uncertainty intrinsic to atmospheric observations limits our knowledge of the current atmospheric state. Our limited knowledge of current and future atmospheric states can be expressed through frameworks that combine probabilistic theories with statistical methods for learning from data. This thesis focuses on probabilistic and statistical frameworks for understanding atmospheric predictability and for data assimilation. Such frameworks should be built on accurate assumptions, though interpretability and computational efficiency are also important criteria. For example, if the framework's purpose is to understand mechanisms underlying the predictability of jet stream variability, then a complex deep learning model is probably too opaque, and a simple linear-Gaussian model might be preferable. In this thesis, I promote methods of probabilistic and statistical inference that relax overly-restrictive assumptions while retaining conceptual simplicity, computational speed, and ease of implementation. I apply such methods toward improving our understanding of midlatitude atmospheric predictability in Chapters 1 and 2, and toward improving the robustness of ensemble data assimilation in Chapter 3. In Chapter 1, I introduce a nonparametric technique for relating the autocorrelation of jet shifts to the momentum budget; in Chapter 2, I use autoregressive linear-Gaussian models to quantify lag-dependent feedbacks between eddies and the background flow, and to understand the importance of these feedbacks for midlatitude predictability. Both of these approaches avoid conventional oversimplifications, such as the notion that eddy-mean flow feedbacks are exclusively positive. In Chapter 3, I introduce a technique for improving covariance matrix estimates from limited samples that does not require assumptions about the matrix structure, allowing the technique to be used in situations where conventional covariance localization fails. The minimal assumptions, simplicity and computational efficiency of these statistical methods makes them attractive for a wide variety of applications.
일반주제명  
Atmospheric sciences
일반주제명  
Applied mathematics
일반주제명  
Statistics
키워드  
Nature
키워드  
Atmospheric observations
키워드  
Midlatitude predictability
기타저자  
University of California, San Diego Scripps Institution of Oceanography
기본자료저록  
Dissertations Abstracts International. 87-03B.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008260126s2025        us                              c    eng  d
■001000017359271
■00520260202104959
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798293841486
■035    ▼a(MiAaPQ)AAI32119231
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a551.5
■1001  ▼aVishny,  David  N.
■24510▼aStatistical  Approaches  for  Understanding  Atmospheric  Dynamics  From  Noisy  Data
■260    ▼a[Sl]▼bUniversity  of  California,  San  Diego▼c2025
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2025
■300    ▼a145  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  87-03,  Section:  B.
■500    ▼aAdvisor:  Lutsko,  Nicholas  J.
■5021  ▼aThesis  (Ph.D.)--University  of  California,  San  Diego,  2025.
■520    ▼aThe  chaotic  nature  of  the  atmosphere  limits  its  predictability,  while  the  sparsity  and  measurement  uncertainty  intrinsic  to  atmospheric  observations  limits  our  knowledge  of  the  current  atmospheric  state.  Our  limited  knowledge  of  current  and  future  atmospheric  states  can  be  expressed  through  frameworks  that  combine  probabilistic  theories  with  statistical  methods  for  learning  from  data.  This  thesis  focuses  on  probabilistic  and  statistical  frameworks  for  understanding  atmospheric  predictability  and  for  data  assimilation.  Such  frameworks  should  be  built  on  accurate  assumptions,  though  interpretability  and  computational  efficiency  are  also  important  criteria.  For  example,  if  the  framework's  purpose  is  to  understand  mechanisms  underlying  the  predictability  of  jet  stream  variability,  then  a  complex  deep  learning  model  is  probably  too  opaque,  and  a  simple  linear-Gaussian  model  might  be  preferable. In  this  thesis,  I  promote  methods  of  probabilistic  and  statistical  inference  that  relax  overly-restrictive  assumptions  while  retaining  conceptual  simplicity,  computational  speed,  and  ease  of  implementation.  I  apply  such  methods  toward  improving  our  understanding  of  midlatitude  atmospheric  predictability  in  Chapters  1  and  2,  and  toward  improving  the  robustness  of  ensemble  data  assimilation  in  Chapter  3.  In  Chapter  1,  I  introduce  a  nonparametric  technique  for  relating  the  autocorrelation  of  jet  shifts  to  the  momentum  budget;  in  Chapter  2,  I  use  autoregressive  linear-Gaussian  models  to  quantify  lag-dependent  feedbacks  between  eddies  and  the  background  flow,  and  to  understand  the  importance  of  these  feedbacks  for  midlatitude  predictability.  Both  of  these  approaches  avoid  conventional  oversimplifications,  such  as  the  notion  that  eddy-mean  flow  feedbacks  are  exclusively  positive.  In  Chapter  3,  I  introduce  a  technique  for  improving  covariance  matrix  estimates  from  limited  samples  that  does  not  require  assumptions  about  the  matrix  structure,  allowing  the  technique  to  be  used  in  situations  where  conventional  covariance  localization  fails.  The  minimal  assumptions,  simplicity  and  computational  efficiency  of  these  statistical  methods  makes  them  attractive  for  a  wide  variety  of  applications.
■590    ▼aSchool  code:  0033.
■650  4▼aAtmospheric  sciences
■650  4▼aApplied  mathematics
■650  4▼aStatistics
■653    ▼aNature
■653    ▼aAtmospheric  observations
■653    ▼aMidlatitude  predictability
■690    ▼a0725
■690    ▼a0364
■690    ▼a0463
■71020▼aUniversity  of  California,  San  Diego▼bScripps  Institution  of  Oceanography.
■7730  ▼tDissertations  Abstracts  International▼g87-03B.
■790    ▼a0033
■791    ▼aPh.D.
■792    ▼a2025
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17359271▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

미리보기

내보내기

chatGPT토론

Ai 추천 관련 도서


    신착도서 더보기
    최근 3년간 통계입니다.

    소장정보

    • 예약
    • 소재불명신고
    • 나의폴더
    • 우선정리요청
    • 비도서대출신청
    • 야간 도서대출신청
    소장자료
    등록번호 청구기호 소장처 대출가능여부 대출정보
    TF15079 전자도서 대출가능 마이폴더 부재도서신고 비도서대출신청 야간 도서대출신청

    * 대출중인 자료에 한하여 예약이 가능합니다. 예약을 원하시면 예약버튼을 클릭하십시오.

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.