Show simple item record

dc.contributor.advisorHancock, Steven
dc.contributor.authorMitchell, Euan C.
dc.date.accessioned2022-05-03T14:05:42Z
dc.date.available2022-05-03T14:05:42Z
dc.date.issued2021
dc.identifier.urihttps://hdl.handle.net/1842/38935
dc.identifier.urihttp://dx.doi.org/10.7488/era/2187
dc.description.abstractNASA’s GEDI instrument is a full-waveform lidar, installed on the International Space Station since late 2018, with the primary mission goal of measuring the 3D structure of the world’s temperate and tropical forests to enable more accurate estimation of global aboveground biomass density (AGBD). Accurately identifying the ground elevation of each GEDI footprint is an essential first step in determining forest structure, relative height metrics, and ultimately biomass. The accuracy of GEDI’s Version 2.0 ground elevation estimates has been investigated across three study sites by comparison with airborne laser scanning (ALS) data. GEDI performs well over less dense temperate forests in the United States, where mean canopy cover is approximately 80%, with ground elevation bias of ~ -1 m. In contrast, in denser tropical rainforest in Costa Rica, where mean canopy cover is greater than 90%, GEDI performs less well, failing to identify low amplitude ground signals in many cases, with a bias of 4 m. A Random Forest machine learning model applied to the Costa Rica data shows limited ability to predict canopy height from Sentinel-2 imagery, with issues of overfitting and spatial autocorrelation in the canopy height data. Regardless, at local scales the model can identify waveforms where GEDI has overestimated ground elevation, thus underestimating canopy height and biomass, allowing for correction of ground elevation estimates and a reduction in bias from 4 m to 1.4 m, resulting in more accurate biomass estimates.en
dc.language.isoenen
dc.publisherThe University of Edinburghen
dc.subjectGEDI, lidar, forest structure, biomass, tropics, rainforest, remote sensingen
dc.titleAccuracy of GEDI Lidar Ground Elevation Estimates and an Assessment of the Potential of Machine Learning to Improve Ground and Biomass Estimatesen
dc.typeThesis or Dissertationen
dc.type.qualificationlevelMastersen
dc.type.qualificationnameMSc Master of Scienceen


Files in this item

This item appears in the following Collection(s)

Show simple item record