Remote sensing for forest water stress monitoring: The case of the Northern Jarrah Forest in southwestern Australia

Thai Son Le

doi:10.60867/00000161

Climate change is increasingly altering global environmental conditions through rising temperatures and shifts in the hydrological cycle. These changes have intensified droughts, heat waves, and water shortages, placing growing pressure on ecosystems. In plants, prolonged water deficit causes water stress, disrupting processes such as photosynthesis, transpiration, nutrient uptake, and growth, leading to reduced productivity, dieback, and mortality. As climate change accelerates, monitoring plant water stress is essential for understanding ecosystem responses and supporting environmental management. Mediterranean-type forests are particularly vulnerable to climate change because of their long, hot, dry summers and frequent droughts. The Northern Jarrah Forest in southwestern Australia is an important case study, dominated by Eucalyptus marginata (jarrah), a species resilient to seasonal drought. Despite these adaptations, declining rainfall and rising temperatures have intensified drought impacts, leading to groundwater decline, crown dieback, and tree mortality. Monitoring water stress across this large and complex forest is difficult using field methods alone, underscoring the need for remote sensing. Advances in remote sensing and vegetation indices offer potential for regional assessment of forest water stress, though their application in jarrah forests remains limited. This thesis investigates the applications of remote sensing for monitoring forest water stress and drought impacts through a multifaceted methodological framework. Specifically, the study develops and applies a remote sensing-based approach to quantify forest water stress dynamics and assess drought-induced canopy vulnerability in the Northern Jarrah Forest (NJF), with the broader aim of supporting adaptive forest management under climate change. The research was conducted at the regional scale across the Northern Jarrah Forest over nearly four decades (1987–2024). The study commenced with a comprehensive review of remote sensing applications for monitoring forest water stress, followed by the development of a novel remote sensing index specifically designed to detect canopy water stress. Subsequent applications of the index included an examination of fire impacts on canopy water status and a long-term time-series analysis to characterize the spatiotemporal dynamics of water stress across jarrah forests. Finally, a habitat modelling framework integrating the newly developed index with surrounding environmental variables was employed to identify areas at elevated risk of drought-induced canopy loss and to determine the primary drivers. The literature review (Chapter 2) indicated that traditional field-based methods for assessing water stress are often impractical for large forested landscapes, leading to the increasing adoption of remote sensing technologies, especially satellite imagery, over the past two decades. Although groundbased approaches remain essential for calibration and validation, satellite remote sensing provides the spatial and temporal coverage necessary for regional-scale monitoring. Existing applications have largely relied on visible and near-infrared spectral bands and common vegetation indices to assess vegetation condition and drought impacts. However, recent advances in remote sensing platforms and higher-resolution spectral datasets have facilitated the development of more physiologically sensitive indicators of water stress. Emerging technologies, including UAV-based observations, hyperspectral imagery, and solar-induced chlorophyll fluorescence, show strong potential for future applications. Nevertheless, widely accessible satellite systems such as MODIS, Landsat, and Sentinel-2 remain fundamental due to their historical continuity, accessibility, and operational scalability. The review further highlighted the importance of integrating remote sensing observations with environmental datasets to improve risk assessment and support sustainable forest management under climate change. Chapter 3 introduced the development of the Infrared Canopy Dryness Index (ICDI), a new remote sensing index based on spectral responses in the near-infrared and shortwave infrared regions derived from Landsat imagery. The index integrates the Normalized Difference Infrared Index (NDII) with canopy structural conditions represented by the Normalized Difference Vegetation Index (NDVI). The ICDI framework was established through the construction of an NDII–NDVI feature space based on a conceptual trapezoidal model. Validation using historical drought-affected and control plots within the study area demonstrated that the ICDI effectively captured variations in canopy dryness and was capable of detecting incipient water stress several months prior to visible drought-induced damage. The use of freely available Landsat imagery further highlights the potential for broad operational application in forest drought monitoring. Chapter 4 investigated the effects of fire on canopy water stress using Landsat time-series imagery and the ICDI to evaluate long-term changes in canopy dryness across jarrah-dominated forests. By examining five prescribed burns and five wildfire events, the study showed that canopy structure, represented by NDVI, recovered relatively rapidly following fire, typically within one to three years depending on fire severity. In contrast, ICDI-derived canopy water stress remained reduced for substantially longer periods, with stress alleviation persisting for approximately 7–8 years following low-severity prescribed burns and more than 20 years following high-severity wildfires. These findings suggest that prescribed burning may function as a strategic tool for mitigating canopy water stress in drought-prone forests by reducing stand-level water demand. The chapter also demonstrated the broader utility of the ICDI for monitoring post-disturbance ecohydrological responses. Chapter 5 characterized the multi-decadal dynamics of canopy water stress using ICDI derived from a 36-year multispectral satellite archive (1988–2024). Analysis of the 1988–2024 period revealed a clear trend of increasing forest water stress, associated with declining climatic water balance, and a systematic expansion of canopy dryness across the landscape. Hotspot analysis indicated a transition from localized and peripheral drought stress to widespread, chronic landscape-scale water limitation. In addition, remote sensing indicators were integrated with GIS-based environmental datasets within a MaxEnt modelling framework to identify the key drivers of drought-induced canopy collapse and delineate high-risk areas under increasing climatic stress. The MaxEnt model achieved high predictive performance (AUC = 0.952). Regolith depth emerged as the primary determinant of canopy collapse risk, followed by ICDI, NDVI, and slope. Importantly, high-biomass stands exhibited disproportionately greater vulnerability, revealing a density-dependent response in which productive forests are approaching critical hydraulic thresholds. In contrast, lower-stature forests in the eastern portion of the study area appeared more resilient, likely due to lower evapotranspirative demand. Collectively, these findings provide strong spatial evidence supporting a transition from reactive monitoring to proactive forest management. Overall, this thesis provides important insights into forest water stress dynamics in the Northern Jarrah Forest and demonstrates broader relevance for Mediterranean-type forest ecosystems globally. The research contributes both a novel remote sensing index and a predictive habitat modelling framework capable of supporting evidence-based decision-making for forest management. The findings further indicate that targeted interventions, including ecological thinning and prescribed burning in identified high-risk areas, are essential for reducing vulnerability and maintaining the structural and functional integrity of forest ecosystems under a drying climate. As forests worldwide face increasing pressure from climate change, the methodologies and findings presented in this thesis provide valuable baseline information to support sustainable forest management and climate adaptation strategies.

Remote sensing for forest water stress monitoring: The case of the Northern Jarrah Forest in southwestern Australia

Files and links (1)

Abstract

Details

Metrics