Boreal forest classification and monitoring using Sentinel-1 coherence information

Many works in the world of Earth observation focus on forest monitoring and classification to analyze the evolution and impact on the planet and society, notably for climate change. In all those studies, using Synthetic Aperture Radar in different frequency bands is one of the most common approaches...

Descripción completa

Detalles Bibliográficos
Autor: Herrera Gimenez, Marc
Tipo de recurso: tesis de maestría
Fecha de publicación:2022
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/378129
Acceso en línea:https://hdl.handle.net/2117/378129
Access Level:acceso abierto
Palabra clave:Synthetic aperture radar
Forest and forestry - Mensuration
Taigas
Synthetic Aperture Radar (SAR)
Sentinel-1
Coherence
Forest Monitoring
Height Estimation
Coherencia
Monitorización Bosques
Estimación Altura
Radar d'obertura sintètica
Dendrometria
Taigàs
Àrees temàtiques de la UPC::Enginyeria de la telecomunicació::Radiocomunicació i exploració electromagnètica::Radar
Descripción
Sumario:Many works in the world of Earth observation focus on forest monitoring and classification to analyze the evolution and impact on the planet and society, notably for climate change. In all those studies, using Synthetic Aperture Radar in different frequency bands is one of the most common approaches. This work aims to use the Sentinel-1 system at C-Band to analyze boreal forest regions. Specifically, it will be studied if, with this system, it is possible to extract height information and biomass information. Additionally, it is tried to obtain the conditions in which that is possible. Two areas of interest in Finland are studied through the analysis of the interferometric coherence, meteorological data and ground truth. First, a pixel-based study is conducted. Then, an area-based. Three types of regions are studied: tall trees, low trees and flat areas (without trees). All the results are contrasted with the different theoretical volume and temporal decorrelation models. Finally, a reflection on the physical phenom behind this decorrelation effect in the data is made. It has been discovered that it is possible to determine qualitatively areas with taller trees and areas with lower ones under particular conditions in winter when there is a lot of snow, which is frozen due to the low temperatures. Also, it can be appreciated how quantitative analysis is very complex because all the external factors involved in C-band together with a certain aleatory component.