Model adaptation to post-fire hydrological impacts

• Autor(a) principal: Basso, Marta
• Data de Publicação: 2023
• Idioma: eng
• Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Texto Completo: http://hdl.handle.net/10773/39739
• topic: Hydrological models; Wildfires; Catchment hydrology; Erosion; Nutrients

Wildfires can have significant impacts on the natural water cycle in forested watersheds. These events can cause changes in vegetation and soil properties, affecting the hydrological and erosive response of the area. Post-fire mobilization of ash, sediments, nutrients, and contaminants can lead to negative effects on downstream waterbodies, which can threaten global water security. The scientific community has thoroughly addressed the impacts of wildfires through field assessments and laboratory experiments but, to a lesser extent, through the use of hydrological models. Hydrological models can be valuable tools to predict the potential negative impacts of wildfires at different scales. Models allow different climate and land use scenarios to be tested, providing a base for post-fire land management decisions. However, most of the existing studies have focused on predicting hydrological and erosive responses at large spatiotemporal scales, often focusing only on the first year after the fire. An exhaustive literature review has shown a lack of studies simulating the propagation of fire impacts to downstream waterbodies, particularly the transport of nutrients and contaminants. The aim of this thesis was to advance the state of the art in post-fire hydrological modeling by addressing research gaps and advancing our understanding of the complex interactions between wildfires, the water cycle, and downstream water quality. A review of scientific literature identified the principal research gaps and the adaptations made to existing models to simulate the specificities of post-fire conditions. The propagation of wildfire impacts beyond the catchment outlet was studied by coupling a watershed model (The Soil and Water Assessment Tool - SWAT) with a reservoir model (CE-QUAL-W2). A simple methodology was implemented to assess the effects of the wildfire on a drinking water supply inlet, using the outputs of main streams as inputs to reservoir branches. The results showed that nutrient concentrations depended strongly on the volume of the waterbody, so dilution effects largely determine the threat to water quality. This study also highlighted the importance of the use of integrated modeling frameworks for anticipating the magnitude of off-site wildfire impacts. Wildfires can cause substantial transformations in affected regions, and management practices implemented after a fire can also significantly affect the catchment response. However, the relationship between post-fire impacts and management interventions remains uncertain. To further investigate the long-term effects of wildfires and the role of different management choices, simulations based on three different management scenarios were conducted. Specifically, post-fire impacts on water quantity and quality were compared considering no intervention (spontaneous ecosystem recovery), implementation of an emergency stabilization measure to mitigate erosion risk (mulching), and reforestation for commercial purposes (terracing + eucalyptus plantation). The three post-fire interventions were implemented in the hydrological model SWAT by parameterizing distinct spatiotemporal patterns in post-fire vegetation and soil recovery, and then calibrating SWAT at the slope scale (mulching scenario) and catchment scale (spontaneous and terracing scenarios). Without any intervention, the wildfire impacts on water quality persisted for up to 8 years after the fire and were largest in terms of sediment yield and nitrogen exports. The terraces revealed smaller impacts on sediment yield and nitrogen export but, at the same time, larger impacts on phosphorous export. The mulching scenario showed the measure was not only highly effective to mitigate post-fire soil and soil fertility losses at the hillslope scale (as is well-established) but also at the catchment scale. Short-term hydrological impacts of wildfires are typically more visible during the first intense rainfall events. However, hydrological models running with a daily or larger time-step are inadequate to predict flash floods that have often been observed in recently burned areas and are typically associated with high sediment loads. These sediment loads consist predominantly of fire ash, thereby containing elevated levels of nutrients and contaminants. For evaluating these impacts, two hydrological models, namely OpenLISEM and MOHID Land, were implemented for a small experimental catchment at event scale. Model parameterization and calibration were addressed through spatial patterns in burn severity, derived from satellite imagery, and the antecedent weather conditions of the individual rainfall-runoff events, especially in terms of initial topsoil moisture conditions. The results showed that the parametrization of the model through initial conditions of individual events provides a simple methodology for obtaining coherent parameter sets across events and, thereby, reducing the uncertainties that typically result from event-wise calibration. In addition, the comparison of the two models was instrumental to highlight the importance of subsurface flow in the rapid hydrological response at catchment scale. The robustness of the sediment yield predictions was ultimately compromised by the large number of relevant input parameters, commonly referred to as the equifinality problem. Establishing a unique parametrization for hydrological models in post-fire conditions is challenging due to the extreme heterogeneity of burned areas. Integration of burn severity and ecosystem recovery is crucial for accurately estimating wildfire impacts. Together with vegetation depletion, modelers must adapt infiltration rate parameters to account for increased surface runoff after wildfires. The selection of parameters to simulate the repellency effect varies based on the governing equations used. Simulating erosion response necessitates the inclusion of factors such as increased soil erodibility and the presence of mobile ash layers. Nutrient and contaminant impacts are often limited in post-fire prediction models, as nutrient and contaminant cycling is commonly omitted. However, the studies revealed strong correlations between sediment and nutrients, emphasizing the role of erosion in nutrient transport in recently burned areas. The adaptation of hydrological parameters in different case studies has highlighted the link between these parameters and the spatiotemporal scales at which they are calibrated. The averaging scale processes together with the increase of processes that influence the hydrological response can explain the difference in parameter values between studies. Simulations of post-fire conditions at higher spatiotemporal resolution provide detailed results for post-fire peak flow prediction. However, high-resolution simulations require high detail and are sensitive to small parameter variation, potentially leading to decreased model performance. While event-based predictions are suitable for assessing profound wildfire impacts on peak flow and sediment transport, they are not ideal for long-term simulations due to data volume and computation time. Models with larger time steps are useful for predicting continuous wildfire impacts on the hydrological cycle. By utilizing continuous models, it becomes possible to implement different landuse and climate scenarios, allowing to study the spatiotemporal variability in the responses of burned areas.