A spatially explicit model of tree leaf litter accumulation in fire maintained longleaf pine forests of the southeastern US

The continuity and depth of litter fuelbeds are major drivers of fire spread and fuel consumption. However, no established approach is available for the spatially explicit prediction of litter loads over large areas. Local fuel heterogeneity introduces large uncertainties on estimates derived from field-based models based on the extrapolation of sparse data samples. In fire-maintained pine forests of the southeastern US, litter accumulation and its distribution over the forest floor are mainly driven by vegetation productivity and by the number of years since the last fire (YSF). Some ecological models that simulate fire effects allow for a time-dependent estimation of litter by accounting for the opposing rates of litter deposition and decomposition as a function of YSF at the landscape level, but they do not account for spatial heterogeneity. We developed a conceptually simple approach for wall-to-wall estimation of tree leaf litter loads at high spatial resolution (5 m). The approach involved, first, estimating spatial patterns of tree annual litterfall. We mapped individual tree crowns through segmentation of airborne laser scanning (ALS) data, and we estimated crown foliage biomass using tree inventory data and ALS derived tree crown attributes. Tree annual litterfall was calculated as a fraction of the crown foliage biomass based on leaf turnover rates. We then quantified tree leaf litter accumulation through a spatially explicit implementation of the established Olson (1963) accumulation and negative decay model. We tested and validated our model in several management and research units at Eglin Air Force Base (Florida), Pebble Hill Plantation (Georgia), and Osceola National Forest (Florida), where managers maintain predominantly longleaf pine forests using frequent fire. Pixel-level RMSD and BIAS between tree leaf litter biomass estimated by the proposed model and reference field measurements were 0.21 and 0.01 kg m−2, and area-level RMSD and BIAS were 0.09 and 0.01 kg m−2. We concluded that linking patterns of litterfall and tree leaf litter accumulation to tree crown objects provides a means to characterizing the discontinuity of the litter layer, accounting for spatial heterogeneity largely traceable to tree crown foliage inputs.