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

FindBook

Google Book

Amazon

博客來

Application of Machine Learning Methods to Imager Cloud Property Estimation and the Feasibility of Their Use in Operations and Climate Data Records.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Application of Machine Learning Methods to Imager Cloud Property Estimation and the Feasibility of Their Use in Operations and Climate Data Records./
作者:	White, Charles H.
面頁冊數:	1 online resource (182 pages)
附註:	Source: Dissertations Abstracts International, Volume: 83-09, Section: B.
Contained By:	Dissertations Abstracts International83-09B.
標題:	Atmospheric sciences. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29060398click for full text (PQDT)
ISBN:	9798209900924

Application of Machine Learning Methods to Imager Cloud Property Estimation and the Feasibility of Their Use in Operations and Climate Data Records.
White, Charles H.

Application of Machine Learning Methods to Imager Cloud Property Estimation and the Feasibility of Their Use in Operations and Climate Data Records. - 1 online resource (182 pages)

Source: Dissertations Abstracts International, Volume: 83-09, Section: B.

Thesis (Ph.D.)--The University of Wisconsin - Madison, 2022.

Includes bibliographical references

Estimates of cloud properties are critical to our understanding of weather and climate variability, but their estimation from satellite imagers is a nontrivial task. Machine learning(ML) approaches have recently gained popularity in earth science and remote sensing. This work explores the use of a kind of ML model, a neural network, for cloud detection and cloud-top pressure estimation from the Visible Infrared Imaging Radiometer Suite (VIIRS),Advanced Baseline Imager (ABI), and Moderate Resolution Imaging Spectroradiometer(MODIS). Several comparisons illustrate large improvement over current operational products which rely on more conventional statistical or physically-based approaches.This increase in performance merits study into the interpretability of neural network cloud property models. A comparison of several modern interpretability frameworks for neural networks shows mixed results and implies that current tools may be insufficient for explaining neural network output in remote sensing tasks with multicollinear predictors. Nonetheless, we find some agreement on the importance of particular spectral features, spatial metrics, and numerical weather prediction output that could inform future algorithm development.A key challenge in transitioning algorithms to satellite climate records is ensuring intersensor consistency. If this is not considered, then long-term analyses of clouds risk being affected by changes in observation platform which can be frequent in our longest satellite records. A method is proposed that simultaneously minimizes differences between imager predictions for matching observations and predictions with respect to a reference instrument. These results offer one pathway for ensuring the appropriateness of ML algorithms in the analysis of satellite climate records.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2023

Mode of access: World Wide Web

ISBN: 9798209900924Subjects--Topical Terms:

3168354
Atmospheric sciences.
Subjects--Index Terms:

Cloud propertiesIndex Terms--Genre/Form:

542853
Electronic books.

Application of Machine Learning Methods to Imager Cloud Property Estimation and the Feasibility of Their Use in Operations and Climate Data Records.
LDR:03154nmm a2200349K 4500 001 2354139
005 20230324111209.5
006 m o d
007 cr mn ---uuuuu
008 241011s2022 xx obm 000 0 eng d
020 $a 9798209900924
035 $a (MiAaPQ)AAI29060398
035 $a AAI29060398
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a White, Charles H. $3 3694485
245 1 0 $a Application of Machine Learning Methods to Imager Cloud Property Estimation and the Feasibility of Their Use in Operations and Climate Data Records.
264 0 $c 2022
300 $a 1 online resource (182 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 83-09, Section: B.
500 $a Advisor: Heidinger, Andrew K.
502 $a Thesis (Ph.D.)--The University of Wisconsin - Madison, 2022.
504 $a Includes bibliographical references
520 $a Estimates of cloud properties are critical to our understanding of weather and climate variability, but their estimation from satellite imagers is a nontrivial task. Machine learning(ML) approaches have recently gained popularity in earth science and remote sensing. This work explores the use of a kind of ML model, a neural network, for cloud detection and cloud-top pressure estimation from the Visible Infrared Imaging Radiometer Suite (VIIRS),Advanced Baseline Imager (ABI), and Moderate Resolution Imaging Spectroradiometer(MODIS). Several comparisons illustrate large improvement over current operational products which rely on more conventional statistical or physically-based approaches.This increase in performance merits study into the interpretability of neural network cloud property models. A comparison of several modern interpretability frameworks for neural networks shows mixed results and implies that current tools may be insufficient for explaining neural network output in remote sensing tasks with multicollinear predictors. Nonetheless, we find some agreement on the importance of particular spectral features, spatial metrics, and numerical weather prediction output that could inform future algorithm development.A key challenge in transitioning algorithms to satellite climate records is ensuring intersensor consistency. If this is not considered, then long-term analyses of clouds risk being affected by changes in observation platform which can be frequent in our longest satellite records. A method is proposed that simultaneously minimizes differences between imager predictions for matching observations and predictions with respect to a reference instrument. These results offer one pathway for ensuring the appropriateness of ML algorithms in the analysis of satellite climate records.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2023
538 $a Mode of access: World Wide Web
650 4 $a Atmospheric sciences. $3 3168354
650 4 $a Artificial intelligence. $3 516317
650 4 $a Remote sensing. $3 535394
653 $a Cloud properties
655 7 $a Electronic books. $2 lcsh $3 542853
690 $a 0725
690 $a 0800
690 $a 0799
710 2 $a ProQuest Information and Learning Co. $3 783688
710 2 $a The University of Wisconsin - Madison. $b Atmospheric & Oceanic Sciences. $3 3350321
773 0 $t Dissertations Abstracts International $g 83-09B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29060398 $z click for full text (PQDT)