語系:
繁體中文
English
說明(常見問題)
登入
回首頁
切換:
標籤
|
MARC模式
|
ISBD
Machine Learning Approach to Burned Area Mapping for Southern California.
紀錄類型:
書目-語言資料,手稿 : Monograph/item
正題名/作者:
Machine Learning Approach to Burned Area Mapping for Southern California./
作者:
Ross, Chandler.
面頁冊數:
1 online resource (51 pages)
附註:
Source: Masters Abstracts International, Volume: 84-10.
Contained By:
Masters Abstracts International84-10.
標題:
Remote sensing. -
電子資源:
click for full text (PQDT)
ISBN:
9798379426668
Machine Learning Approach to Burned Area Mapping for Southern California.
Ross, Chandler.
Machine Learning Approach to Burned Area Mapping for Southern California.
- 1 online resource (51 pages)
Source: Masters Abstracts International, Volume: 84-10.
Thesis (M.S.)--San Diego State University, 2023.
Includes bibliographical references
Accurate representation of the location and amount of burned areas is vital to the understanding of spatial and temporal patterns of fires, and to assessing their environmental impacts. Extant burned area maps currently have high commission errors, which lead to an overrepresentation of burned area. The primary research objective of this thesis was to assess whether localized training data used for image-based machine learning routines improve accuracy of burned area products for western San Diego County. I used localized training data derived from fine scale aerial imagery to create and compare three training sets, with a Landsat scale pixel burn threshold of 20%, 50%, or 80%. These training data were input into a gradient-boosted regression model with the same 64 spectral vegetation indices (SVI) inputs as the US Geological Survey's Landsat Burned Area (LBA) product to classify burned and not burned lands. I compared the burned area product from the localized gradient boosted model (L-GBRM) to the three other products: Monitoring Trends in Burn Severity (MTBS), Fire and Resource Assessment Program (FRAP), and LBA. I found 20% to be the burn pixel threshold for the training data that yielded the most accurate classification. I used a 50% pixel burn threshold for the reference data since the burned area associated with it is closely aligned with the fine-scale burn area. I conducted an accuracy assessment for each of the products by randomly sampling 300 points from the validation dataset. The L-GBRM was the most accurate product while also mapping the smallest area burned, suggesting that the extant products have high commission errors, often through omitting interior unburned patches. Using local training data achieved a higher accuracy than nationwide training data.
Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024
Mode of access: World Wide Web
ISBN: 9798379426668Subjects--Topical Terms:
557272
Remote sensing.
Subjects--Index Terms:
Machine learningIndex Terms--Genre/Form:
554714
Electronic books.
Machine Learning Approach to Burned Area Mapping for Southern California.
LDR
:03089ntm a22003737 4500
001
1144809
005
20240611104924.5
006
m o d
007
cr mn ---uuuuu
008
250605s2023 xx obm 000 0 eng d
020
$a
9798379426668
035
$a
(MiAaPQ)AAI30424291
035
$a
AAI30424291
040
$a
MiAaPQ
$b
eng
$c
MiAaPQ
$d
NTU
100
1
$a
Ross, Chandler.
$3
1469960
245
1 0
$a
Machine Learning Approach to Burned Area Mapping for Southern California.
264
0
$c
2023
300
$a
1 online resource (51 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: Masters Abstracts International, Volume: 84-10.
500
$a
Advisor: Stow, Douglas.
502
$a
Thesis (M.S.)--San Diego State University, 2023.
504
$a
Includes bibliographical references
520
$a
Accurate representation of the location and amount of burned areas is vital to the understanding of spatial and temporal patterns of fires, and to assessing their environmental impacts. Extant burned area maps currently have high commission errors, which lead to an overrepresentation of burned area. The primary research objective of this thesis was to assess whether localized training data used for image-based machine learning routines improve accuracy of burned area products for western San Diego County. I used localized training data derived from fine scale aerial imagery to create and compare three training sets, with a Landsat scale pixel burn threshold of 20%, 50%, or 80%. These training data were input into a gradient-boosted regression model with the same 64 spectral vegetation indices (SVI) inputs as the US Geological Survey's Landsat Burned Area (LBA) product to classify burned and not burned lands. I compared the burned area product from the localized gradient boosted model (L-GBRM) to the three other products: Monitoring Trends in Burn Severity (MTBS), Fire and Resource Assessment Program (FRAP), and LBA. I found 20% to be the burn pixel threshold for the training data that yielded the most accurate classification. I used a 50% pixel burn threshold for the reference data since the burned area associated with it is closely aligned with the fine-scale burn area. I conducted an accuracy assessment for each of the products by randomly sampling 300 points from the validation dataset. The L-GBRM was the most accurate product while also mapping the smallest area burned, suggesting that the extant products have high commission errors, often through omitting interior unburned patches. Using local training data achieved a higher accuracy than nationwide training data.
533
$a
Electronic reproduction.
$b
Ann Arbor, Mich. :
$c
ProQuest,
$d
2024
538
$a
Mode of access: World Wide Web
650
4
$a
Remote sensing.
$3
557272
650
4
$a
Geographic information science.
$3
1372545
653
$a
Machine learning
653
$a
Area mapping
653
$a
Environmental impacts
653
$a
Burned lands
655
7
$a
Electronic books.
$2
local
$3
554714
690
$a
0370
690
$a
0799
710
2
$a
ProQuest Information and Learning Co.
$3
1178819
710
2
$a
San Diego State University.
$b
Geography.
$3
1180991
773
0
$t
Masters Abstracts International
$g
84-10.
856
4 0
$u
http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30424291
$z
click for full text (PQDT)
筆 0 讀者評論
多媒體
評論
新增評論
分享你的心得
Export
取書館別
處理中
...
變更密碼[密碼必須為2種組合(英文和數字)及長度為10碼以上]
登入