Hemiboreal Forest Mapping with Interferometric Synthetic Aperture Radar
Date
2016-10-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Käesolev doktoritöö uurib tehisavaradari (SAR) kasutusvõimalusi metsa kõrguse hindamiseks hemiboreaalsete metsade vööndis. Uurimistöö viidi läbi Tartu Üli¬kooli, Tartu Observatooriumi, Aalto Ülikooli, Euroopa Kosmoseagentuuri (ESA) kaugseire keskuse ESRIN ja Reach-U koostöös. Uurimistöös kasutatud satelliidi¬andmed on pärit Saksa Kosmosekeskuse (DLR) kõrglahutusega bistaatilise
X-laineala tehisavaradari TanDEM-X satelliidipaarilt.
Sagedasti uuenevad satelliidiandmed, nende globaalne katvus ja kõrge ruumi¬line lahutus võimaldavad tehisavaradari abil kaardistada metsi ning nendes toimu¬vaid muutusi suurtel maa-aladel. Radari abil on võimalik saada kõrge lahutusvõimega pilte, mis on tundlikud taimestikule, maapinna karedusele ja dielektrilistele omadustele. Sünkroonis lendava radaripaari samaaegselt tehtud pildid elimineerivad võimalikud ajalised muutused taimestikus ning tänu sellele on radariandmetest võimalik tuletada metsade vertikaalset struktuuri ja kõrgust.
Uurimistöös käsitletakse tehisavaradari interferomeetrilise koherentsuse tund¬likkust metsa kõrguse suhtes ning analüüsitakse, millised keskkonna ja klimaati¬lised tingimused ning satelliidi orbiidiga seotud parameetrid mõjutavad radari¬piltidelt erinevate puuliikide kõrguse hindamise täpsust. Lisaks keskendub väitekiri interferomeetrilisele koherentsusele tuginevate mudelite analüüsi¬misele ning nende täpsuse hindamisele operatiivse metsa kõrguse kaardistamise raken-duseks. Vaatluse alla on võetud kolm testala, mis asuvad Soomaa rahvuspargis, Võrtsjärve idakaldal Rannus ja Peipsiveere looduskaitsealal ning katavad kokku 2291 hektarit metsa. 23 TanDEM-X satelliidipildi koherentsuspilte võrreldakse samadel testaladel aerolaserskaneerimise (LiDAR) abil mõõdetud puistute kõrgu¬sega, mis on omakorda jagatud kolme rühma (kuused, männid ja laia¬lehised segametsad).
RVoG (Random Volume over Ground) taimekatte mudel ning sellest tule¬tatud lihtsamad pooleempiirilised mudelid sobituvad olemasolevate TanDEM-X koherentsuse ning LiDARi metsa puistute kõrgusandmetega hästi. Töö tule¬mused kinnitavad, et tulevikus on suurte ja erinevatest metsatüüpidest koosne¬vate metsade kõrguse kosmosest kaardistamisel otstarbekas kasutusele võtta esmalt just soovitatud lihtsamad ja universaalsemad mudelid.
This thesis presents research in the field of radar remote sensing and contributes to the forest monitoring application development using space-borne synthetic aperture radar (SAR). Satellite data is particularly useful for large-scale forestry applications making high revisit monitoring of the state of forests worldwide possible. The sensitivity of SAR to the dielectric and geometrical properties of the targets, penetration capacity and coherent imaging properties make it a unique tool for mapping and monitoring forest biomes. SAR satellites are also capable of retrieving additional information about the structure of the forest, tree height and biomass estimates as an essential input for monitoring the changes in the carbon stocks. Interferometric SAR (InSAR) is an advanced SAR imaging technique that allows the retrieval of forest parameters while working in nearly all weather conditions, independently of daylight and cloud cover. This research concen¬trates on assessing the impact of different variables affecting hemiboreal forest height estimation from space-borne X-band interferometric SAR coherence data. In particular, the research analyses the changes in coherence dynamics related to seasonal conditions, tree species and imaging properties using a large collection of interferometric SAR images from different seasons over a four-year period. The study is carried out over three test sites in Estonia using the extensive multi-temporal dataset of 23 TanDEM-X images, covering 2291 hectares of forests to describe the relation between the interferometric SAR coherence mag¬nitude and forest parameters. The work demonstrates how the correlation of interferometric coherence and Airborne LiDAR Scanning (ALS)-derived forest height varies for pine and deciduous tree species, for summer (leaf-on) and winter (leaf-off) conditions and for flooded forest floor. A simple semi-empirical modelling approach is proposed as being suitable for wide area forest mapping with limited a priori information under a range of seasonal and environ¬¬mental conditions. A Random Volume over Ground (RVoG) model and three semi-empirical models are compared and validated against a large dataset of coherence magnitude and ALS-measured data over hemiboreal forests in Estonia. The results show that all proposed models perform well in describing the relationship between hemiboreal forest height and interferometric coherence, allowing in future to derive forest stand height with an accuracy suitable for a wide range of applications.
This thesis presents research in the field of radar remote sensing and contributes to the forest monitoring application development using space-borne synthetic aperture radar (SAR). Satellite data is particularly useful for large-scale forestry applications making high revisit monitoring of the state of forests worldwide possible. The sensitivity of SAR to the dielectric and geometrical properties of the targets, penetration capacity and coherent imaging properties make it a unique tool for mapping and monitoring forest biomes. SAR satellites are also capable of retrieving additional information about the structure of the forest, tree height and biomass estimates as an essential input for monitoring the changes in the carbon stocks. Interferometric SAR (InSAR) is an advanced SAR imaging technique that allows the retrieval of forest parameters while working in nearly all weather conditions, independently of daylight and cloud cover. This research concen¬trates on assessing the impact of different variables affecting hemiboreal forest height estimation from space-borne X-band interferometric SAR coherence data. In particular, the research analyses the changes in coherence dynamics related to seasonal conditions, tree species and imaging properties using a large collection of interferometric SAR images from different seasons over a four-year period. The study is carried out over three test sites in Estonia using the extensive multi-temporal dataset of 23 TanDEM-X images, covering 2291 hectares of forests to describe the relation between the interferometric SAR coherence mag¬nitude and forest parameters. The work demonstrates how the correlation of interferometric coherence and Airborne LiDAR Scanning (ALS)-derived forest height varies for pine and deciduous tree species, for summer (leaf-on) and winter (leaf-off) conditions and for flooded forest floor. A simple semi-empirical modelling approach is proposed as being suitable for wide area forest mapping with limited a priori information under a range of seasonal and environ¬¬mental conditions. A Random Volume over Ground (RVoG) model and three semi-empirical models are compared and validated against a large dataset of coherence magnitude and ALS-measured data over hemiboreal forests in Estonia. The results show that all proposed models perform well in describing the relationship between hemiboreal forest height and interferometric coherence, allowing in future to derive forest stand height with an accuracy suitable for a wide range of applications.
Description
Väitekirja elektrooniline versioon ei sisalda publikatsioone.
Keywords
radarid, satelliidid (tehn.), segametsad, metsad, kõrgus, radarkaardistus, radars, artificial satellites, mixed forests, forests, height, radar mapping