Coupled Eco-hydrological Modelling for Assessing Environmental Flow Releases

Abstract

Environmental flow management is a key measure for vegetation restoration and maintenance, and has yielded a variety of positive ecological outcomes. Remote sensing can be used to monitor and improve our understanding of vegetation condition in response to hydro-climatic factors by providing high spatial and temporal resolution data over long periods. This is particularly useful in data-scarce or remote regions. However, using remote sensing data to predict vegetation response to flooding and drought and to inform environmental water management remains challenging in semi-arid floodplain wetlands, where vegetation condition exhibits high heterogeneity during inundation. This research focuses on a semi-arid floodplain-lakes system in the northern Murray-Darling Basin, Australia - the Narran Lakes floodplain. Through four interrelated chapters, this thesis analyses vegetation response to hydro-climatic factors and leverages this understanding for the development and application of an eco-hydrological model to support environmental flow delivery in this region. Chapter 2 analyses the effect of hydro-climatic factors on vegetation condition during 2000-2021 using correlation analysis and generalised additive models. The analysis considers five key variables: precipitation, inflow, soil moisture, temperature, and solar exposure. Results show that soil moisture is the primary factor influencing NDVI (Normalized Difference Vegetation Index). Water availability factors including soil moisture, inflow, and precipitation interact in a complex manner to affect NDVI. Chapter 3 develops a coupled eco-hydrological model that simulates vegetation condition from soil moisture deficit in response to hydro-climatic inputs. The model integrates three components: an inundation module to simulate inundation extent, a soil moisture accounting module adapted from IHACRES-CMD, and an ecological response module that estimates vegetation condition using Leaf Area Index (LAI). Results show that the simulation of inundation extent achieves high accuracy, and the model captures general trends for LAI values in the Narran Lakes. Chapter 4 assesses vegetation condition in response to different environmental flow scenarios, incorporating predictive uncertainty using Bayesian inference. Results show no clear long-term decline in vegetation condition across annual and 5-year scales, though periods of drought stress and greater variability have increased since the 2000s; During the growth period, an LAI below 0.4-0.6, accounting for predictive uncertainty in threshold estimates, indicates that the vegetation is in a critical state; Environmental water delivered during the growth period can improve vegetation condition, with improvements plateauing around 50,000 ML. Chapter 5 proposes a framework for reflecting on modelling decisions through a fitness-for-purpose lens and pathway perspective consisting of four steps: a) Modelling decision boundaries; b) Modelling pathway and reasoning; c) Evaluation of modelling decisions; and d) Iterative refinement. Applied to the eco-hydrological model, this framework comprehensively investigates and rates modelling decisions made in the modelling cycle. The application of this framework provides practical guidance for improving model reliability and usability. In summary, this thesis investigates vegetation dynamics in response to hydro-climatic drivers in the Narran Lakes and develops a coupled modelling approach to inform environmental flow management. It demonstrates the dominant role of soil moisture in influencing vegetation condition in the Narran Lakes, introduces a novel coupled eco-hydrological model to predict vegetation response, incorporates predictive uncertainty in scenario analysis, and provides a framework to reflect on modelling decisions. These contributions offer technical support for vegetation monitoring and conservation in the context of environmental flow management in semi-arid wetlands. Show more