東華大學圖書館 |

Advancing Near Real-Time Deforestation Monitoring Via Multi-Source Remote Sensing Data, Landscape Processes, and End-User Contributions.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Advancing Near Real-Time Deforestation Monitoring Via Multi-Source Remote Sensing Data, Landscape Processes, and End-User Contributions./
作者:	McGregor, Ian Ramsey.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2023,
面頁冊數:	158 p.
附註:	Source: Dissertations Abstracts International, Volume: 85-05, Section: B.
Contained By:	Dissertations Abstracts International85-05B.
標題:	Remote sensing. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30673686
ISBN:	9798380714242

Advancing Near Real-Time Deforestation Monitoring Via Multi-Source Remote Sensing Data, Landscape Processes, and End-User Contributions.
McGregor, Ian Ramsey.

Advancing Near Real-Time Deforestation Monitoring Via Multi-Source Remote Sensing Data, Landscape Processes, and End-User Contributions. - Ann Arbor : ProQuest Dissertations & Theses, 2023 - 158 p.

Source: Dissertations Abstracts International, Volume: 85-05, Section: B.

Thesis (Ph.D.)--North Carolina State University, 2023.

This item must not be sold to any third party vendors.

Preserving forests and mitigating deforestation have been primary goals of tropical remote sensing for decades. From landscape change estimates to recent near real-time (NRT) monitoring alerts, methods have continuously evolved with the help of an increasing amount of satellite imagery, the shift from static to time series approaches, and the near-instantaneous availability of satellite data. However, as it has developed, NRT monitoring with remote sensing has encountered challenges in two main areas: technical improvements, and bridging the gap between model development and clear impacts on the ground. The former includes issues with detecting small-scale change over periods with extensive cloud cover, and the use of short training periods due to either data availability or computational intensity. In addition, previous methods have indicated the presence of trade-offs may preclude an optimization of both detection accuracy and latency, yet there have been few dedicated discussions about the implications for users. Meanwhile, the ability of remote sensing analyses to directly influence policy is hampered by the exclusion of external spatial data describing landscape processes, as well as its inability to incorporate data users as part of the system.For the first chapter, we proposed a novel, multi-source, NRT monitoring algorithm that harmonizes data from three sensors, two optical and one synthetic aperture radar. We chose our study region to be Myanmar, as despite high levels of forest cover and biodiversity, deforestation rates are high in the region, with spatial scales of forest removal ranging from shifting cultivation to the expansion of commercial agriculture. In creating the algorithm, we generated training data and also focused on whether or not we could optimize detection latency and accuracy. Trade-offs between the two were characterized for different parameterizations to understand the impacts on decision-making by users, such as forest managers deciding to prioritize fast or accurate detections. We then demonstrated the application of the method at landscape scale by creating daily probability maps at 10 m resolution for two sub-regions and assessed the outputs by validating positive detections within them.In the second chapter we developed an extension to Chapter 1 for which we incorporated external spatial data from landscape processes into the probability estimates. We first identified proximate drivers of deforestation in the study region and modeled their influence on forest disturbance while explicitly accounting for spatial covariance. These estimated effects were then combined with the remote sensing probabilities from the first chapter using a Bayesian methodology, resulting in a more representative and confident detection estimate. We conducted another sensitivity analysis to understand how the trade-offs from the first chapter changed, and subsequently created new daily probability estimates for the sub-regions to determine if incorporating the external data improved the NRT detection metrics.Finally, the third chapter introduced a new method for incorporating users into the workflow. We proposed an iterative approach in which field validation data was directly used by the algorithm to update the probability equation, allowing for continuous improvement over time. We explored the impact of varying the different inputs, and discussed the implications of them for users in an operational setting.

ISBN: 9798380714242Subjects--Topical Terms:

535394
Remote sensing.

Advancing Near Real-Time Deforestation Monitoring Via Multi-Source Remote Sensing Data, Landscape Processes, and End-User Contributions.
LDR:04676nmm a2200361 4500 001 2396771
005 20240611104936.5
006 m o d
007 cr#unu||||||||
008 251215s2023 ||||||||||||||||| ||eng d
020 $a 9798380714242
035 $a (MiAaPQ)AAI30673686
035 $a (MiAaPQ)NCState_Univ18402041090
035 $a AAI30673686
040 $a MiAaPQ $c MiAaPQ
100 1 $a McGregor, Ian Ramsey. $3 3766512
245 1 0 $a Advancing Near Real-Time Deforestation Monitoring Via Multi-Source Remote Sensing Data, Landscape Processes, and End-User Contributions.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2023
300 $a 158 p.
500 $a Source: Dissertations Abstracts International, Volume: 85-05, Section: B.
500 $a Advisor: Gray, Joshua M.
502 $a Thesis (Ph.D.)--North Carolina State University, 2023.
506 $a This item must not be sold to any third party vendors.
520 $a Preserving forests and mitigating deforestation have been primary goals of tropical remote sensing for decades. From landscape change estimates to recent near real-time (NRT) monitoring alerts, methods have continuously evolved with the help of an increasing amount of satellite imagery, the shift from static to time series approaches, and the near-instantaneous availability of satellite data. However, as it has developed, NRT monitoring with remote sensing has encountered challenges in two main areas: technical improvements, and bridging the gap between model development and clear impacts on the ground. The former includes issues with detecting small-scale change over periods with extensive cloud cover, and the use of short training periods due to either data availability or computational intensity. In addition, previous methods have indicated the presence of trade-offs may preclude an optimization of both detection accuracy and latency, yet there have been few dedicated discussions about the implications for users. Meanwhile, the ability of remote sensing analyses to directly influence policy is hampered by the exclusion of external spatial data describing landscape processes, as well as its inability to incorporate data users as part of the system.For the first chapter, we proposed a novel, multi-source, NRT monitoring algorithm that harmonizes data from three sensors, two optical and one synthetic aperture radar. We chose our study region to be Myanmar, as despite high levels of forest cover and biodiversity, deforestation rates are high in the region, with spatial scales of forest removal ranging from shifting cultivation to the expansion of commercial agriculture. In creating the algorithm, we generated training data and also focused on whether or not we could optimize detection latency and accuracy. Trade-offs between the two were characterized for different parameterizations to understand the impacts on decision-making by users, such as forest managers deciding to prioritize fast or accurate detections. We then demonstrated the application of the method at landscape scale by creating daily probability maps at 10 m resolution for two sub-regions and assessed the outputs by validating positive detections within them.In the second chapter we developed an extension to Chapter 1 for which we incorporated external spatial data from landscape processes into the probability estimates. We first identified proximate drivers of deforestation in the study region and modeled their influence on forest disturbance while explicitly accounting for spatial covariance. These estimated effects were then combined with the remote sensing probabilities from the first chapter using a Bayesian methodology, resulting in a more representative and confident detection estimate. We conducted another sensitivity analysis to understand how the trade-offs from the first chapter changed, and subsequently created new daily probability estimates for the sub-regions to determine if incorporating the external data improved the NRT detection metrics.Finally, the third chapter introduced a new method for incorporating users into the workflow. We proposed an iterative approach in which field validation data was directly used by the algorithm to update the probability equation, allowing for continuous improvement over time. We explored the impact of varying the different inputs, and discussed the implications of them for users in an operational setting.
590 $a School code: 0155.
650 4 $a Remote sensing. $3 535394
650 4 $a Partial differential equations. $3 2180177
650 4 $a Sensitivity analysis. $3 3560752
650 4 $a Funding. $3 3557916
650 4 $a Wildlife sanctuaries. $3 3693523
650 4 $a Sensors. $3 3549539
650 4 $a Probability. $3 518898
650 4 $a Time series. $3 3561811
650 4 $a Forests. $3 3560105
650 4 $a Environmental policy. $3 518940
650 4 $a Deforestation. $3 600701
650 4 $a Forestry. $3 895157
650 4 $a Wildlife conservation. $2 fast $3 542165
690 $a 0799
690 $a 0474
690 $a 0478
690 $a 0284
710 2 $a North Carolina State University. $3 1018772
773 0 $t Dissertations Abstracts International $g 85-05B.
790 $a 0155
791 $a Ph.D.
792 $a 2023
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30673686