A spatially explicit model of litter accumulation in fire maintained longleaf pine forest ecosystems of the Southeastern USA

Abstract

The continuity and depth of the surface fuel layer (i.e., litter and duff) are major drivers of fire spread and fuel consumption. Nevertheless, its spatial explicit quantification over relatively large areas remains unresolved: local fuel heterogeneity introduces large uncertainties in estimates derived from field-based models and sparse data samples. Besides that, the sensitivity of remote sensors to surface litter loads is limited, particularly under canopy cover. In fire-maintained pine forests of the Southeastern US, surface fuel accumulation and its distribution over the forest floor are mainly driven by vegetation productivity, decomposition, and time since fire (TSF). Traditional ecological and stand-level models provide a means to equilibrate between opposing rates of deposition and decomposition as a function of TSF at the landscape level but don’t account for spatial heterogeneity. We developed a top-down, object-based approach for wall-to-wall estimation of surface litter loads using TSF records, the ecological-based Olson model, and tree crown objects derived from airborne laser scanning (ALS) data. The approach involves, first, the spatially explicit estimation of litter production through a tree crown production model, driven by tree crown attributes extracted from the ALS point clouds, and informed by tree inventory data and allometric equations, including vegetation leaf turnover rates. Second, litter accumulation is estimated using the fire-driven Olson equation, which models accumulation progressively with time until decomposition balances deposition and a steady state of accumulation is reached. The methodology is demonstrated at several fire-maintained longleaf pine forest locations in southeastern USA, where tree inventory data, surface litter loads, prescribed fire records, and ALS data are available for testing and validation of the methodology. Comparison between preliminary modeled estimates and observed litter loads shows a relatively good agreement (RMSE=0.21 [kg m-2]; BIAS 0.07 [kg m-2]). This suggests that the proposed approach to indirectly map patterns of litter production and litter accumulation can provide a realistic means to map the continuity of the litter layer, thus overcoming the limitation of traditional ecological landscape models to account for spatial heterogeneity. This high-resolution map of litter loads will be further valuable as input to physics-based fire behavior and spread models and to improve the spatially explicit characterization of the duff layer.