Characterising the structure of forest disturbances and identifying wind-related damage at the global scale

Download Statistics

Downloads

Downloads per month over past year

Acil, Nezha ORCID: https://orcid.org/0000-0002-4043-0414 (2023). Characterising the structure of forest disturbances and identifying wind-related damage at the global scale. University of Birmingham. Ph.D.

Acil2023PhD_Redacted.pdf
Text - Accepted Version
Restricted to Repository staff only
Available under License All rights reserved.
Download (24MB) | Request a copy

Abstract

Stand-replacing disturbances are periodic events that damage a group of trees, generating heterogeneity in forest structure, composition and functions. They vary in their size, shape and frequency across regions and depending on the causal agent. However, their structural and spatiotemporal characteristics are not yet quantified at the global scale and by disturbance agent. In particular, the distribution, frequency and structural patterns of wind-related disturbances are still unknown in many regions. In the context of climate change, these quantifications are needed to assess the contribution of storm damage to overall disturbance regimes and to build a baseline for projecting future forest dynamics.
The aims of this thesis are to: 1) explore the structural patterns of forest disturbances at the patch level and quantify their typology and distribution across the world, 2) assess the differences between natural and human-influenced disturbance regimes, 3) assess how the characteristics of wind-related disturbances vary among regions and differ from those of other disturbances, and 4) develop a methodology to classify wind-related disturbances based on identified structural and environmental characteristics.
A framework leveraging both high-performance and cloud computing systems was designed and implemented. The purpose was to delineate disturbance patches individually at the global level, measure their structural metrics and link them with a selection of ancillary information sourced from different global databases. The result was a global database of 413 million raw patches, covering the period 2001-2018, further corrected into 345 million patches. Disturbance structure, landscape context and environmental characteristics were then described and analysed using more than 100 covariates. Patches attributable to fire and large-scale forest conversion were excluded to narrow down to only disturbances that can potentially be caused by storms.
Key outputs of this thesis include: 1) maps illustrating the global distribution of disturbance patterns and highlighting four types of structures: small-isolated, complex, clustered and large-multi-year patches ; 2) quantifications of human effects on natural disturbance regimes across the spectrum of patch sizes, showing that human influence is homogenising disturbance sizes across biomes and asymmetrically amplifying disturbance rates; 3) a characterisation of the structure and environmental affinities of wind-related disturbances, as opposed to other non-fire disturbances across regions, showing that wind-related patches tend to be more temporally-clustered than the other agents across ecobiomes (i.e. subdivisions of biomes by ecological realms); 4) a preliminary global model for the attribution of wind-related disturbances, generated using machine learning and achieving a satisfactory performance accuracy.
This thesis capitalised on open Earth Observation data and state-of–the-art computing systems to produce a globally consistent database of forest disturbance information established at the patch level. This database enables multiple applications, notably those involving the interplay between, disturbance structure and the environment. The framework developed provide a solid progress toward attributing disturbances agents at the global scale. The identification of wind-induced disturbance patches will enable the characterisation of their regime metrics, the quantification of their carbon emissions, and the projection of their influence on future forest dynamics.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):