Plants in river channels and on floodplains are involved in complex and reciprocal interactions with the hydrologic regime and the geomorphological processes that configure the fluvial landscape. Drag forces exerted on vegetation by floodwater are the core driver for these interactions, influencing the hydraulic regime by absorbing flow momentum and, in turn, altering sediment transport processes. Some tree species recruited to fluvial environments (termed rheophytes) have an adapted leaf, stem and root architecture that reduces drag forces and the tree’s propensity for uprooting or rupture. In large floods, where drag forces exceed the anchorage resistance of a tree’s root system or cause excessive tensile stresses in the main stem, forest disturbances due to uprooting or stem rupture can occur. Such disturbances can lead to a cascade of morphological responses as the underlying substrate is suddenly exposed to swift-flowing water and woody debris is liberated into the river system. This thesis advances our understanding and ability to simulate/predict two aspects of these interactions: flow resistance through and disturbances to riparian forests during floods. These processes are investigated in the context of subtropical rivers, which are characterised by forested bars and floodplain pockets that flood deeply due to the variation in seasonal runoff. While the various model parameters that are developed may be specific to subtropical rivers and riparian trees, the underlying physical processes are relevant for all climate regions and tree species where floodwaters flow through forest. Thus, the methods used for experiments, data analysis, and model development are applicable in all regions. The geographic relevance of this work is also broadened by the fact that the subtropical species that have been studied have been introduced to other regions. Flow resistance through forests is generally simulated in hydraulic models using roughness parameters like Manning’s n. These parameters are commonly assigned based on published look-up tables or from experience, and the fact that Manning’s n is dependent on flow conditions when this parameter is used to simulate forest flow resistance is often overlooked. However, advances have been made on this topic using drag-force models to formulate Manning’s n in forests. These drag force-based approaches are burdened by the additional data collection and analysis needed to parameterise the models. However, advances in the technology used to collect and analyse environmental data are reducing this burden, making the application of drag-force based methods more feasible. Some of the drag force model parameters are species-specific, and available data is limited to a few broad-leaf tree species distributed in temperate regions. This thesis aims to advance the drag-force based approach to simulate flow resistance through riparian forests in a subtropical region, where differences in foliage morphology may manifest differences in drag force behaviour compared to temperate species. The thesis commences with the question of disturbance thresholds for riparian forests during floods. Thresholds for vegetation disturbances during floods, especially for mature trees, are lacking in the literature, and an extreme flood in January 2011 on the lower North Pine River in south-east Queensland, Australia, provided an opportunity to investigate forest disturbance thresholds. The flood disturbed patches of forest and grass growing on bars in the compound channel of this subtropical river. The hydraulic conditions during the flood were simulated using a depth-averaged 2D hydraulic model. A range of hydraulic metrics, including bed shear stress and stream power, were estimated and tested to determine if they could be used as thresholds to distinguish disturbed from undisturbed patches of vegetation. Several of the metrics could be used to establish disturbance thresholds in the forest with a reasonable accuracy of 65-70% relative to the area of the disturbed forest. For example, a stream power threshold of 834 W/m2 predicted the extent of disturbed forest with an accuracy of 65%. Unit flow was also capable of being used as a disturbance threshold in grassy areas, predicting the extent of stripped grass with an accuracy of 23% using a threshold value of 26 m2/s. The initial methods used to investigate disturbance thresholds in forests did not consider the driving (i.e. drag forces on trees) and resisting (i.e. anchorage resistance of the root-soil plate) forces in detail. To analyse tree stability at such detail would first require the development of a drag force model to estimate drag forces on trees during floods. Such a model was also needed to develop the drag force-based method to simulate flow resis tance in forests. Therefore, experiments were undertaken to measure drag forces on fully submerged, freshly cut trees towed through still water on a reservoir with a boat. Two subtropical, riparian species were tested: sheoak (Casuarina spp.) and river tea tree (Melaleuca bracteata). The results were used to parameterise and test two existing tree drag models. For the sheoak trees, the existing models were unable to predict the transition between a rigid regime at lower velocities (.0.5 m/s), where stem bending is minimal, and a reconfiguration regime at higher velocities where stems become more streamlined as velocities increase. Drag coefficients are uniform in the rigid regime and reduce with increasing velocity in the reconfiguration regime. An improved tree drag model was developed that successfully predicted regime transition for the sheoak trees, which is important for simulating flow resistance through forests as flow resistance would otherwise be over estimated when drag forces are weak compared to tree rigidity. Measurements for the tea trees were within the reconfiguration regime, and slower velocities would be needed to confirm the location of regime transition. The tree drag model developed in this thesis predicted d ag forces with an average accuracy of 85% for the sheoak trees and 90% for the tea trees. The drag force model was used to develop a flow resistance model for single-storey, monospecific forests, using Manning’s n to simulate flow resistance through forests. The results indicate that Manning’s n values in the forest increase with flow depth. Manning’s n values were independent of velocity in the rigid regime and decreased with increasing velocity in the reconfiguration regime. The drag force-based model was applied by simulating the January 2011 flood through a sheoak forest on the North Pine River using a depth-averaged 2D hydraulic model. Data obtained from an airborne laser scanning (ALS) survey were used to estimate the height, density, and position of trees in the forest. Data collected from terrestrial laser scanning (TLS) surveys were used to develop allometric relationships for tree frontal area, enabling parameterisation of the drag force-based model. The hydraulic model results were in reasonable agreement with flood levels from wrack marks collected after the January 2011 flood, with average flood levels being 0.16 m lower than recorded (the average flood depth is 5 m in the forest). Floods with a range of return periods were simulated in the hydraulic model using both the drag force-based model and a fixed Manning’s n for the forest. The two methods were compared to investigate improvements in accuracy gained from the drag force-based model developed in this thesis. Differences in flood levels between the two methods were 0.02-0.03 m at the flood peak and up to 0.16 m in the lead up to the flood peak on the rising limb of the flood hydrograph. The differences at the flood peak are marginal compared to overall model accuracy. Nevertheless, the drag force-based model developed in this thesis has the potential to improve the accuracy of flood simulations by providing a means to compute flow condition-dependent Manning’s n values in forests based on measures of the forest structure, rather than estimating a uniform Manning’s n from experience or look-up tables. This is particularly relevant in catchments where flow resistance through forests controls flow rates down the river system and where calibration data is limited. The model results also provide an example of how forests, by attenuating floodwater, can offer a flood mitigation service by reducing flood levels downstream. Finally, the research on forest disturbance thresholds during floods was revisited, investigating the driving and resisting forces in greater detail. Observations of canopy patches in the sheoak forest that were lost due to uprooting or rupture during the January 2011 flood were identified from aerial photographs and used to classify stable and unstable trees in the forest during the flood. Results from the drag experiments were used to develop a model predicting the moment of drag forces on sheoak trees, which was then applied to the sheoak forest using the results of the hydraulic model to estimate the moment of drag forces on the trees during the flood. Factors that affect the anchorage resistance were also computed, including the depth to water table (a surrogate for root depth), scour depths around the trunks (similar to bridge piers), and stream competence (a measure of erosion propensity). These factors were combined in several combinations to produce a tree stability model that could predict the observed stability. The adopted model had an accuracy of 71% for the January 2011 flood, and was used to simulate the stability of trees for a range of design floods with return periods from 2 to 10,000 years. The results suggest that no more than a third of the forest may be lost in a single flood across the range of floods tested. Results from this work are highly relevant to studies seeking to rehabilitate riparian forests and to studies investigating the hydraulic, geomorphic, and ecological effects of nature-based solutions for river and flood management. For such studies, this work can be used to improve the accuracy of flood simulations and to provide an indication of forest stability during floods.