We use cookies to ensure that we give you the best experience on our website. By continuing to browse this repository, you give consent for essential cookies to be used. You can read more about our Privacy and Cookie Policy.

Durham Research Online
You are in:

Quantifying structural diversity to better estimate change at mountain forest margins.

Morley, Peter J. and Donoghue, Daniel N.M. and Chen, Jan-Chang and Jump, Alistair S. (2019) 'Quantifying structural diversity to better estimate change at mountain forest margins.', Remote sensing of environment., 223 . pp. 291-306.


Global environmental changes are driving shifts in forest distribution across the globe with significant implications for biodiversity and ecosystem function. At the upper elevational limit of forest distribution, patterns of forest advance and stasis can be highly spatially variable. Reliable estimations of forest distribution shifts require assessments of forest change to account for variation in treeline advance across entire mountain ranges. Multispectral satellite remote sensing is well suited to this purpose and is particularly valuable in regions where the scope of field campaigns is restricted. However, there is little understanding of how much information about forest structure at the mountain treeline can be derived from multispectral remote sensing data. Here we combine field data from a structurally diverse treeline ecotone in the Central Mountain Range, Taiwan, with data from four multispectral satellite sensors (GeoEye, SPOT-7, Sentinel-2 and Landsat-8) to identify spectral features that best explain variation in vegetation structure at the mountain treeline and the effect of sensor spatial resolution on the characterisation of structural variation. The green, red and short-wave infrared spectral bands and vegetation indices based on green and short-wave infrared bands offer the best characterisation of forest structure with R2 values reported up to 0.723. There is very little quantitative difference in the ability of the sensors tested here to discriminate between discrete descriptors of vegetation structure (difference of R2MF within 0.09). While Landsat-8 is less well suited to defining above-ground woody biomass (R2 0.12–0.29 lower than the alternative sensors), there is little difference between the relationships defined for GeoEye, SPOT-7 and Sentinel-2 data (difference in R2 < 0.03). Discrete classifications are best suited to the identification of forest structures indicative of treeline advance or stasis, using a simplified class designation to separate areas of old growth forest, forest advance and grassland habitats. Consequently, our results present a major opportunity to improve quantification of forest range shifts across mountain systems and to estimate the impacts of forest advance on biodiversity and ecosystem function.

Item Type:Article
Full text:(VoR) Version of Record
Available under License - Creative Commons Attribution.
Download PDF
Publisher Web site:
Publisher statement:© 2019 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (
Date accepted:21 January 2019
Date deposited:13 February 2019
Date of first online publication:01 February 2019
Date first made open access:13 February 2019

Save or Share this output

Look up in GoogleScholar