An Alternative Approach to Using LiDAR Remote Sensing Data to Predict Stem Diameter Distributions across a Temperate Forest Landscape

Abstract

Abstract:We apply a spatially-implicit, allometry-based modelling approach to predict stem diameter distributions (SDDs) from low density airborne LiDAR data in a heterogeneous, temperate forest in Ontario, Canada. Using a recently published algorithm that relates the density, size, and species of individual trees to the height distribution of first returns, we estimated parameters that succinctly describe SDDs that are most consistent with each 0.25-ha LiDAR tile across a 30,000 ha forest landscape. Tests with independent validation plots showed that the diameter distribution of stems was predicted with reasonable accuracy in most cases (half of validation plots had R 2 ≥ 0.75, and another 23% had 0.5 ≤ R 2 < 0.75). The predicted frequency of larger stems was much better than that of small stems (8 ≤ x < 11 cm diameter), particularly small conifers. We used the predicted SDDs to calculate aboveground carbon density (ACD; RMSE = 21.4 Mg C/ha), quadratic mean diameter (RMSE = 3.64 cm), basal area (RMSE = 6.99 m 2 /ha) and stem number (RMSE = 272 stems/ha). The accuracy of our predictions compared favorably with previous studies that have generally been undertaken in simpler conifer-dominated forest types. We demonstrate the utility of our results to spatial forest management planning by mapping SDDs, the proportion of broadleaves, and ACD at a 0.25 ha resolution.

Forest Site and Type Variability in ALS-Based Forest Resource Inventory Attribute Predictions over Three Ontario Forest Sites

“…The %RMSE observed in this study were comparable to other studies that modeled similar FRI attributes at the same forest sites [10,13,44,45,52]. These prediction accuracies were calculated at the plot level (or pixel level once wall-to-wall prediction surfaces of the FRI attributes are generated).…”

“…For example, Pitt et al [44] observed that the quality of FRI attribute predictions varied with forest type and that the largest errors were observed in boreal mixedwoods in HF, regardless of which prediction method used. Contrary to the findings in the boreal zone, Spriggs et al [45] found that in HFWR, prediction accuracy was better for broadleaved species and postulated that estimation models were most suited to the dominant forest type of a site. In the GLSL forest sites, we did not observe either of these trends consistently over the three FRI attributes and two sites.…”

“…Among our three sites, structural complexity and species composition were expected to be less variable for the boreal forest site, i.e., HF is largely dominated by conifers (70% versus 30% mixed and intolerant hardwoods) [13], and more complex for the temperate forest sites. In HFWR, however, species composition was fairly homogenous, i.e., the dominant species in almost all plots is sugar maple [45]. Hence prediction accuracies in both HF and HFWR were better than in PRF, which was the most complex forest site in terms of species composition and structure.…”

“…The variability in FRI attribute prediction accuracy within forest types has been addressed either by predicting FRI attributes separately by forest type [9,10,13,[40][41][42][43][44][45] or by predicting FRI attributes over the entire forest site and subsequently extracting predictions (and error) by forest type (e.g. [13,44,45]). For example, Pitt et al [44] found that the quality of FRI attribute predictions varied with forest type regardless of which prediction method used.…”

“…Studies in boreal and temperate regions in Europe (e.g., [40,41,43,46]) found more accurate predictions of FRI attribute estimates for conifers than deciduous or mixed forest types. However, Spriggs et al [45] found that BA and S were more accurately predicted for broadleaved species than conifers in the Great Lakes, St. Lawrence (GLSL) forest region in Ontario. The authors postulated that estimation models fit dominant forest type better.…”