Incorporating fire severity mapping into fuel accumulation curves

Models of vegetation and fuel accumulation following fire are important for carbon accounting, species conservation and fire prediction. In terms of fire behaviour, models of fuel accumulation are used to predict in situ fuel loads, which are required to predict fire spread and fire-line intensity. However, attributes of the historical fire regime, such as severity, are rarely considered in these models. In Australian forests, a negative exponential model of fuel accumulation (i.e. the Olson model) is used in fire management operations to predict fuel loads through time as a function of time-since-fire. The Olson model takes as inputs postfire fuel load ('Initial'), fuel decay rate ('k') and steady-state fuel load ('Limit'). Here, we use empirical data from 150 sites across forests of south-east Australia to understand how fire severity - low representing cooler understory fire, high representing hotter tree canopy fire - impacts the Initial parameter for ground-lying fine fuels, which are a key driver of fire behaviour. We then compare different Olson model fuel load predictions accounting for fire severity effects. Postfire fine fuel load was lower at sites previously burned by high rather than low severity fire across three of the four eucalypt dominated forest types assessed, with mean predicted differences of between 1.0 and 2.3 t ha^-1. These differences translated to higher Olson model fuel load predictions for 12-20 years at sites burned by low rather than high severity fire, with the magnitude of these differences diminishing through time. Collectively, our results suggest that operational fuel prediction models may underestimate fuel loads in areas burnt at low severity. Accounting for such effects will increase the accuracy of fuel load and fire hazard prediction.