Modelling hydrological change due to wildfires

Wildfires are an important hydrological disturbance altering runoff by the fire induced changes (from heat and smoke) to vegetation and soils. The damage to both vegetation and soils creates immediate local changes to water portioning and associated runoff which can accumulate to catchment scale changes. Such changes are dynamic depend on the subsequent recovery of soils and regrowth/recovery of vegetation. As the prevalence of wildfires increases due to a warming and drying climate, such disturbances are a necessary consideration in the management and associated modelling of water resources. The impact of wildfires on water resources decisions comes not only in the immediate aftermath of the fires, but throughout the recovery phase and over extended periods with multiple cycles of wildfire and growth. Associated with these three timescales are pertinent hydrological issues, i.e. flood risk, water quality, water allocations and planning of reservoir operations. Modelling the wildfire induced hydrological impacts at each of these timescales requires a focus on the dominant processes; hydroclimate/meteorological, vegetation, hydrological, fire spread and effects; the latter here is only of significance in consideration of future fires. Focusing on different levels of abstractions of the dominant processes has resulted in a diverse set of approaches across existing models. Here, we explore some of the existing models in the literature that have been applied in assessing wildfire induced changes to runoff. For the purposes of comparison we broadly categorise these models into one of three categories: data-driven, conceptual and physically-based (eco-)hydrological models. We consider their demonstrated applications (assessing changes to streamflow and baseflow, historical analysis of pre- and post-fire periods of streamflow, predicting long term changes to yield), process representations (implicit vs explicit, lumped vs distributed) and spatiotemporal scales (from plot scales over days to watersheds over decades). Based on these characteristics and considering the computational requirements, data type breadth needed and the ability to predict wildfire impacts on runoff for different fires in the future, we describe the key limitations of each model category. With the significant changes to hydrological functioning that are possible after wildfire, physically-based models that utilize physical and biological principals are likely to receive increased attention with their perceived ability to extrapolate outside of the historical record. However, the computational and data costs of physically-based models are limiting the ability to completely support water resources planning. We argue that overcoming such limitations, while leveraging strengths, is possible through adoption of a hybrid modelling approach which combines computationally efficient conceptual models with reduced-order models. Such a hybrid approach would enable the requisite simulations of wildfire-induced changes to runoff for critical water resources planning scenarios.