• Login
• Admin

Home, Philip (2022) Quantifying urban forest structure in Greater Manchester with open-access remote sensing datasets. MRes thesis, University of Nottingham.

Preview

PDF (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (2MB) | Preview

Abstract

A growing body of evidence links the adverse impacts of expanding urbanism including increased air pollution, and exposure to heat stress with the removal of vegetation within cities. As the global population is estimated to reach 10 billion by 2050, urban trees and extended green infrastructure are advocated as a remedy to the effects of increasing urbanisation through delivering a multitude of ecosystem services including pollution abatement, reduction of urban heat islands and social benefits. To accurately quantify the services afforded by urban forests, it is vital to measure the extent and structure of urban forests, including through time, in addition for assessing the success of policy to maintain and promote green infrastructure assets. Current ground fieldwork methods rely on plot networks to measure a range of metrics across the tree population; these methods are locally comprehensive however do not fully describe the spatial heterogeneity of the urban fabric, given the limited sampling and often laborious data collection. The increasing availability and

access to remote sensing/earth observation datasets provide an opportunity to collate synoptic measurements across large regions. Direct measurements though active sensors,

particularly LiDAR, have seen wide adoption when measuring forest structure, however surveys can be expensive, and coverage limited. Fusing LiDAR with satellite imagery though

machine learning methods such as Random Forests can drastically increase coverage through capturing complex non linear relationships. A framework is presented to estimate forest structure using open access data and software across Greater Manchester. This workflow estimates three forest structure metrics, canopy cover, canopy height and tree

number/density. Random forest models were trained with airborne Environment Agency LiDAR, and predictor variables derived from Sentinel 2 and ancillary climatic and topographic datasets. Results indicate estimates in 2018, mean canopy cover of 14.9% (RMSE = 13.75), mean canopy height of 14.83m (RMSE = 6.14m) and home to ~2.6 million trees (RMSE = 0.95 per pixel). Results appear to illustrate higher canopy cover than i-Tree ground data but lower tree density and canopy heights. Altering input resolution was found to change structure estimations, attributed to methodological issues. Forest structure estimates were found to change from 2018 to 2021 indicating net decreases in canopy cover and number of trees, while average canopy height was found to increase, although change distribution of metrics across boroughs is not equal. Presented methods can augment traditional inventory methods and can assist urban forest/land managers to produce consistent monitoring information to support the sustainability of urban forests worldwide.

Item Type:	Thesis (University of Nottingham only) (MRes)
Supervisors:	Boyd, Doreen
Keywords:	Urban forestry, Manchester (England); Remote sensing; Data sets
Subjects:	G Geography. Anthropology. Recreation > G Geography (General) S Agriculture > SB Plant culture
Faculties/Schools:	UK Campuses > Faculty of Engineering
Item ID:	71866
Depositing User:	Home, Philip
Date Deposited:	16 Feb 2023 09:11
Last Modified:	16 Feb 2023 09:11
URI:	https://eprints.nottingham.ac.uk/id/eprint/71866

Actions (Archive Staff Only)

Edit View