{"id":{"repo_id":"sdstate","oai_identifier":"oai:openprairie.sdstate.edu:etd2-2346"},"canonical_url":"https://search.ndltd.org/etd/sdstate/oai:openprairie.sdstate.edu:etd2-2346","repository":{"repo_id":"sdstate","name":"South Dakota State University","base_url":"https://openprairie.sdstate.edu/do/oai/"},"display":{"title":"Lidar Remote Sensing and the Monitoring of Brazilian Amazon Forest Structure: Tackling Issues and Finding New Possibilities","abstract":"The Amazon is the largest tropical forest in the world and around 60% of it is in Brazil. The amount of carbon stored in the region uncertain, the impacts of land use and climate changes are unknown, and what drives the dynamic of the forest structure is still under a heated debate. Optical remote sensing has been used for a long time to assist the monitoring of the Brazilian Amazon. However, optical remote sensing only allows a comprehensive study from the top of the forest canopy. In contrast, lidar remote sensing of forests can produce robust information regarding the canopy structure and terrain morphology. However, it has disadvantages that cannot be disregarded and, on the contrary, must be well addressed to provide reliable information of the forest structure. This study aims to investigate common problems and propose new insight to monitor the Brazilian Amazon forest structure using lidar data. Specifically, in Chapter 2, canopy height prediction accuracy from random forest model were evaluated, focusing on three Landsat 8 reflectance inputs—Top-of-Atmosphere (TOA) reflectance, surface reflectance, and Nadir BRDF Adjusted reflectance (NBAR)—along with sample sizes, split strategies (pixel based 80%-20% random split vs. transect based 16-4 spatial split), and inclusion of geographic coordinates. Results supported the use of NBAR data over TOA or surface reflectance products for modeling canopy heights in the Brazilian Amazon as NBAR was not affected by spatial autocorrelation of view zenith angle. The 80%-20% random split consistently showed lower errors due to spatial autocorrelation between training and test data, while the 16-4 spatial split had higher errors and biases when geographic coordinates were included. In Chapter 3, the effects of beam sensitivity and terrain slope on Global Ecosystem Dynamics Investigation (GEDI) L2A relative heights are evaluated by comparing those heights with GEDI simulated ones. Then, the effects of beam sensitivity and terrain slope on GEDI L2A were assessed by accounting the geolocation uncertainty through bootstrapping new GEDI footprints center coordinates and creating multiple Monte Carlo-simulated waveforms. Lastly, the use of geolocation-adjusted footprints was evaluated. Results suggested that, in the Brazilian Amazon, beam sensitivity primarily affects GEDI L2A relative heights (RH), while terrain slope generally does not. In Chapter 4, tropical forest phenology in the Brazilian Amazon using GEDI data was examined. Time-series of GEDI’s Plant Area Index (PAI) were used to evaluate the spatial distribution of PAI in the Brazilian Amazon forest and to identify the forests’ phenological timing. The results revealed lower PAI in the centralwest Brazilian Amazon and higher PAI in surrounding areas, with decreases near neighboring biomes. The phenological timing in the Brazilian Amazon varied both spatially and temporally, with upper canopy and understory cycles sometimes widening or narrowing occasionally resulting in two cycles per year in certain regions. Chapter 5 presents a summary of the research as well as recommendations. In short, this dissertation provides pathways to better use GEDI data and to improve the modelling of canopy heights, as well as demonstrates efficacy of determining phenological timing of the Brazilian Amazon forest.","abstract_html":"The Amazon is the largest tropical forest in the world and around 60% of it is in Brazil. The amount of carbon stored in the region uncertain, the impacts of land use and climate changes are unknown, and what drives the dynamic of the forest structure is still under a heated debate. Optical remote sensing has been used for a long time to assist the monitoring of the Brazilian Amazon. However, optical remote sensing only allows a comprehensive study from the top of the forest canopy. In contrast, lidar remote sensing of forests can produce robust information regarding the canopy structure and terrain morphology. However, it has disadvantages that cannot be disregarded and, on the contrary, must be well addressed to provide reliable information of the forest structure. This study aims to investigate common problems and propose new insight to monitor the Brazilian Amazon forest structure using lidar data. Specifically, in Chapter 2, canopy height prediction accuracy from random forest model were evaluated, focusing on three Landsat 8 reflectance inputs—Top-of-Atmosphere (TOA) reflectance, surface reflectance, and Nadir BRDF Adjusted reflectance (NBAR)—along with sample sizes, split strategies (pixel based 80%-20% random split vs. transect based 16-4 spatial split), and inclusion of geographic coordinates. Results supported the use of NBAR data over TOA or surface reflectance products for modeling canopy heights in the Brazilian Amazon as NBAR was not affected by spatial autocorrelation of view zenith angle. The 80%-20% random split consistently showed lower errors due to spatial autocorrelation between training and test data, while the 16-4 spatial split had higher errors and biases when geographic coordinates were included. In Chapter 3, the effects of beam sensitivity and terrain slope on Global Ecosystem Dynamics Investigation (GEDI) L2A relative heights are evaluated by comparing those heights with GEDI simulated ones. Then, the effects of beam sensitivity and terrain slope on GEDI L2A were assessed by accounting the geolocation uncertainty through bootstrapping new GEDI footprints center coordinates and creating multiple Monte Carlo-simulated waveforms. Lastly, the use of geolocation-adjusted footprints was evaluated. Results suggested that, in the Brazilian Amazon, beam sensitivity primarily affects GEDI L2A relative heights (RH), while terrain slope generally does not. In Chapter 4, tropical forest phenology in the Brazilian Amazon using GEDI data was examined. Time-series of GEDI’s Plant Area Index (PAI) were used to evaluate the spatial distribution of PAI in the Brazilian Amazon forest and to identify the forests’ phenological timing. The results revealed lower PAI in the centralwest Brazilian Amazon and higher PAI in surrounding areas, with decreases near neighboring biomes. The phenological timing in the Brazilian Amazon varied both spatially and temporally, with upper canopy and understory cycles sometimes widening or narrowing occasionally resulting in two cycles per year in certain regions. Chapter 5 presents a summary of the research as well as recommendations. In short, this dissertation provides pathways to better use GEDI data and to improve the modelling of canopy heights, as well as demonstrates efficacy of determining phenological timing of the Brazilian Amazon forest.","abstract_has_math":false,"creators":["Valle de Carvalho e Oliveira, Pedro"],"institution":null,"degree_name":"Doctor of Philosophy (PhD)","degree_level":"Dissertation - Open Access","degree_discipline":"Geospatial Science and Engineering","degree_department":null,"school":null,"contributors":["Xiaoyang Zhang"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T09:05:24Z","subjects":["Geographic Information Sciences","Physical and Environmental Geography","Remote Sensing"],"languages":[],"rights":[],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://openprairie.sdstate.edu/etd2/1342","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://openprairie.sdstate.edu/etd2/1342","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://openprairie.sdstate.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Aopenprairie.sdstate.edu%3Aetd2-2346","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Xiaoyang Zhang"]},{"key":"dc:creator","label":"Author","values":["Valle de Carvalho e Oliveira, Pedro"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2024-12-16T08:00:00Z"]},{"key":"dc:date.created","label":"Dc Date Created","values":["2024-01-01T08:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Geospatial Science and Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation - Open Access"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Geographic Information Sciences","Physical and Environmental Geography","Remote Sensing"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://openprairie.sdstate.edu/etd2/1342"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The Amazon is the largest tropical forest in the world and around 60% of it is in Brazil. The amount of carbon stored in the region uncertain, the impacts of land use and climate changes are unknown, and what drives the dynamic of the forest structure is still under a heated debate. Optical remote sensing has been used for a long time to assist the monitoring of the Brazilian Amazon. However, optical remote sensing only allows a comprehensive study from the top of the forest canopy. In contrast, lidar remote sensing of forests can produce robust information regarding the canopy structure and terrain morphology. However, it has disadvantages that cannot be disregarded and, on the contrary, must be well addressed to provide reliable information of the forest structure. This study aims to investigate common problems and propose new insight to monitor the Brazilian Amazon forest structure using lidar data. Specifically, in Chapter 2, canopy height prediction accuracy from random forest model were evaluated, focusing on three Landsat 8 reflectance inputs—Top-of-Atmosphere (TOA) reflectance, surface reflectance, and Nadir BRDF Adjusted reflectance (NBAR)—along with sample sizes, split strategies (pixel based 80%-20% random split vs. transect based 16-4 spatial split), and inclusion of geographic coordinates. Results supported the use of NBAR data over TOA or surface reflectance products for modeling canopy heights in the Brazilian Amazon as NBAR was not affected by spatial autocorrelation of view zenith angle. The 80%-20% random split consistently showed lower errors due to spatial autocorrelation between training and test data, while the 16-4 spatial split had higher errors and biases when geographic coordinates were included. In Chapter 3, the effects of beam sensitivity and terrain slope on Global Ecosystem Dynamics Investigation (GEDI) L2A relative heights are evaluated by comparing those heights with GEDI simulated ones. Then, the effects of beam sensitivity and terrain slope on GEDI L2A were assessed by accounting the geolocation uncertainty through bootstrapping new GEDI footprints center coordinates and creating multiple Monte Carlo-simulated waveforms. Lastly, the use of geolocation-adjusted footprints was evaluated. Results suggested that, in the Brazilian Amazon, beam sensitivity primarily affects GEDI L2A relative heights (RH), while terrain slope generally does not. In Chapter 4, tropical forest phenology in the Brazilian Amazon using GEDI data was examined. Time-series of GEDI’s Plant Area Index (PAI) were used to evaluate the spatial distribution of PAI in the Brazilian Amazon forest and to identify the forests’ phenological timing. The results revealed lower PAI in the centralwest Brazilian Amazon and higher PAI in surrounding areas, with decreases near neighboring biomes. The phenological timing in the Brazilian Amazon varied both spatially and temporally, with upper canopy and understory cycles sometimes widening or narrowing occasionally resulting in two cycles per year in certain regions. Chapter 5 presents a summary of the research as well as recommendations. In short, this dissertation provides pathways to better use GEDI data and to improve the modelling of canopy heights, as well as demonstrates efficacy of determining phenological timing of the Brazilian Amazon forest."]},{"key":"dc:title","label":"Title","values":["Lidar Remote Sensing and the Monitoring of Brazilian Amazon Forest Structure: Tackling Issues and Finding New Possibilities"]}]}],"canonical_facts":{"dc:contributor":["Xiaoyang Zhang"],"dc:creator":["Valle de Carvalho e Oliveira, Pedro"],"dc:date.available":["2024-12-16T08:00:00Z"],"dc:date.created":["2024-01-01T08:00:00Z"],"dc:description.abstract":["The Amazon is the largest tropical forest in the world and around 60% of it is in Brazil. The amount of carbon stored in the region uncertain, the impacts of land use and climate changes are unknown, and what drives the dynamic of the forest structure is still under a heated debate. Optical remote sensing has been used for a long time to assist the monitoring of the Brazilian Amazon. However, optical remote sensing only allows a comprehensive study from the top of the forest canopy. In contrast, lidar remote sensing of forests can produce robust information regarding the canopy structure and terrain morphology. However, it has disadvantages that cannot be disregarded and, on the contrary, must be well addressed to provide reliable information of the forest structure. This study aims to investigate common problems and propose new insight to monitor the Brazilian Amazon forest structure using lidar data. Specifically, in Chapter 2, canopy height prediction accuracy from random forest model were evaluated, focusing on three Landsat 8 reflectance inputs—Top-of-Atmosphere (TOA) reflectance, surface reflectance, and Nadir BRDF Adjusted reflectance (NBAR)—along with sample sizes, split strategies (pixel based 80%-20% random split vs. transect based 16-4 spatial split), and inclusion of geographic coordinates. Results supported the use of NBAR data over TOA or surface reflectance products for modeling canopy heights in the Brazilian Amazon as NBAR was not affected by spatial autocorrelation of view zenith angle. The 80%-20% random split consistently showed lower errors due to spatial autocorrelation between training and test data, while the 16-4 spatial split had higher errors and biases when geographic coordinates were included. In Chapter 3, the effects of beam sensitivity and terrain slope on Global Ecosystem Dynamics Investigation (GEDI) L2A relative heights are evaluated by comparing those heights with GEDI simulated ones. Then, the effects of beam sensitivity and terrain slope on GEDI L2A were assessed by accounting the geolocation uncertainty through bootstrapping new GEDI footprints center coordinates and creating multiple Monte Carlo-simulated waveforms. Lastly, the use of geolocation-adjusted footprints was evaluated. Results suggested that, in the Brazilian Amazon, beam sensitivity primarily affects GEDI L2A relative heights (RH), while terrain slope generally does not. In Chapter 4, tropical forest phenology in the Brazilian Amazon using GEDI data was examined. Time-series of GEDI’s Plant Area Index (PAI) were used to evaluate the spatial distribution of PAI in the Brazilian Amazon forest and to identify the forests’ phenological timing. The results revealed lower PAI in the centralwest Brazilian Amazon and higher PAI in surrounding areas, with decreases near neighboring biomes. The phenological timing in the Brazilian Amazon varied both spatially and temporally, with upper canopy and understory cycles sometimes widening or narrowing occasionally resulting in two cycles per year in certain regions. Chapter 5 presents a summary of the research as well as recommendations. In short, this dissertation provides pathways to better use GEDI data and to improve the modelling of canopy heights, as well as demonstrates efficacy of determining phenological timing of the Brazilian Amazon forest."],"dc:identifier":["https://openprairie.sdstate.edu/etd2/1342"],"dc:rights":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:subject":["Geographic Information Sciences","Physical and Environmental Geography","Remote Sensing"],"dc:title":["Lidar Remote Sensing and the Monitoring of Brazilian Amazon Forest Structure: Tackling Issues and Finding New Possibilities"],"thesis:degree_discipline":["Geospatial Science and Engineering"],"thesis:degree_level":["Dissertation - Open Access"],"thesis:degree_name":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-11-19T09:05:24Z"}