Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali

Tropical rainforest in Malaysia experienced logging activities either legal or illegal since 1960s. Natural regeneration become slower if the permanent production of forest is takes place in inappropriate manner until today. Remote Sensing technology is widely used in forestry application to preserv...

Full description

Saved in:

Bibliographic Details
Main Author:	Rozali, Syaza
Format:	Thesis
Language:	English
Published:	2021
Subjects:	Remote Sensing Environmental aspects of forestry
Online Access:	https://ir.uitm.edu.my/id/eprint/60955/1/60955.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!

id	my-uitm-ir.60955
record_format	uketd_dc
spelling	my-uitm-ir.609552022-06-03T04:27:29Z Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali 2021-07 Rozali, Syaza Remote Sensing Environmental aspects of forestry Tropical rainforest in Malaysia experienced logging activities either legal or illegal since 1960s. Natural regeneration become slower if the permanent production of forest is takes place in inappropriate manner until today. Remote Sensing technology is widely used in forestry application to preserve the natural resources. Airborne LiDAR sensor (ALS) is one of the technologies that can be applied together with satellite imagery available such as Landsat and has been proved as an efficient sensor in managing and analyzing deforestation and forest degradation in tropical rainforest. This research aim is to quantify forest disturbance using multi-sensor Remote Sensing in tropical rainforest. The research objectives are to determine forest and non-forest cover from Landsat time series in 2007 until 2018 and Airborne LiDAR sensor data by using vegetation indices (NDVI) and canopy height model (CHM); to identify the forest disturbance detection from Landsat time series by measuring the history period and structural change using BFAST model, and disturbance index (DI); and to quantify the forest disturbance in tropical rainforest using fusion image of spectral and height data by resampling Landsat (30m) into (1m, 15m) resolutions, object-based segmentation technique and classification method using Nearest Neighbor, Random Forest and Support Vector Machine. Two fusion data are tested; 1) SpectralLandsat and 2) SpectralLandsat + HeightALS by Random Forest and Support Vector Machine classification algorithm. The result shows second fusion data having 1 meter Landsat resolution and Airborne LiDAR performed better classification using object-based segmentation and Random Forest classification, about 96% of the overall accuracy with 0.91 kappa index of aggreement. Using the scale of interpretation for kappa statistic shows the very good agreement. Mcnemar‘s test (p-value <0.05) for SpectralLandsat + HeightALS using Random Forest classifier is 0.03. All objectives were achieved successfully and the findings shows that; 1) the higher the resolution of the fusion image, the higher the number of the scale parameter will be used in multi-resolution segmentation; 2) the accuracy of classification was improved when combining LiDAR and Landsat image and 3) quantifying the forest disturbance can be performed using NDVI, CHM and DI information. 2021-07 Thesis https://ir.uitm.edu.my/id/eprint/60955/ https://ir.uitm.edu.my/id/eprint/60955/1/60955.pdf text en public phd doctoral Universiti Teknologi MARA Faculty of Architecture, Planning and Surveying Abd Latif, Zulkiflee (Prof. Sr. Dr.)
institution	Universiti Teknologi MARA
collection	UiTM Institutional Repository
language	English
advisor	Abd Latif, Zulkiflee (Prof. Sr. Dr.)
topic	Remote Sensing Environmental aspects of forestry
spellingShingle	Remote Sensing Environmental aspects of forestry Rozali, Syaza Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali
description	Tropical rainforest in Malaysia experienced logging activities either legal or illegal since 1960s. Natural regeneration become slower if the permanent production of forest is takes place in inappropriate manner until today. Remote Sensing technology is widely used in forestry application to preserve the natural resources. Airborne LiDAR sensor (ALS) is one of the technologies that can be applied together with satellite imagery available such as Landsat and has been proved as an efficient sensor in managing and analyzing deforestation and forest degradation in tropical rainforest. This research aim is to quantify forest disturbance using multi-sensor Remote Sensing in tropical rainforest. The research objectives are to determine forest and non-forest cover from Landsat time series in 2007 until 2018 and Airborne LiDAR sensor data by using vegetation indices (NDVI) and canopy height model (CHM); to identify the forest disturbance detection from Landsat time series by measuring the history period and structural change using BFAST model, and disturbance index (DI); and to quantify the forest disturbance in tropical rainforest using fusion image of spectral and height data by resampling Landsat (30m) into (1m, 15m) resolutions, object-based segmentation technique and classification method using Nearest Neighbor, Random Forest and Support Vector Machine. Two fusion data are tested; 1) SpectralLandsat and 2) SpectralLandsat + HeightALS by Random Forest and Support Vector Machine classification algorithm. The result shows second fusion data having 1 meter Landsat resolution and Airborne LiDAR performed better classification using object-based segmentation and Random Forest classification, about 96% of the overall accuracy with 0.91 kappa index of aggreement. Using the scale of interpretation for kappa statistic shows the very good agreement. Mcnemar‘s test (p-value <0.05) for SpectralLandsat + HeightALS using Random Forest classifier is 0.03. All objectives were achieved successfully and the findings shows that; 1) the higher the resolution of the fusion image, the higher the number of the scale parameter will be used in multi-resolution segmentation; 2) the accuracy of classification was improved when combining LiDAR and Landsat image and 3) quantifying the forest disturbance can be performed using NDVI, CHM and DI information.
format	Thesis
qualification_name	Doctor of Philosophy (PhD.)
qualification_level	Doctorate
author	Rozali, Syaza
author_facet	Rozali, Syaza
author_sort	Rozali, Syaza
title	Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali
title_short	Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali
title_full	Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali
title_fullStr	Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali
title_full_unstemmed	Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali
title_sort	quantifying forest disturbance using lidar data and time series landsat images / syaza rozali
granting_institution	Universiti Teknologi MARA
granting_department	Faculty of Architecture, Planning and Surveying
publishDate	2021
url	https://ir.uitm.edu.my/id/eprint/60955/1/60955.pdf
_version_	1783735191890558976

Quantifying forest disturbance using LiDAR data and time series Landsat images / Syaza Rozali

Similar Items