An assessment of uncertainty in volume estimates for stands reconstructed from tree stump information

Assessment of pre-harvest stand conditions after unplanned tree removals often requires reconstruction of the stand based on stump information. Prediction of diameter at breast height (d.b.h.) from stump measurements is a common practice because d.b.h. is usually a necessary precursor for estimating diameter distributions and predicting tree volume. Although not a widespread exercise, tree volumes are sometimes predicted directly from stump dimensions. Regardless of the approach taken, statistical models are invariably used in some manner and the model predictions are erroneously assumed to be without error. In this study, several methods for tree volume prediction arising from stump information were evaluated for the contribution of model-related uncertainty to the error in population estimates of total volume. When the entire population was enumerated, the model-related uncertainty was 1-2 per cent of the estimate depending on the volume estimation method. Sampling approaches based on individual stumps and 0.042 ha plots were evaluated, where the total uncertainty due to both model and sampling error was considerably larger when using the plot-based method. Generally, the smallest amount of error was present when predicting d.b.h. and then estimating tree volume from d.b.h. The uncertainty was largest for estimation of tree volume directly from stump dimensions when sampling proportions were similar to 0.35 or smaller; otherwise, the largest uncertainty resulted from prediction of d.b.h. and merchantable height which were both used as predictor variables in the volume model.