produced inferior results as the slope values and the corresponding
standard errors (SE) indicate a significant difference. However, it
must be noted that the reflection method shows the smallest residual
mean standard deviation (RMSD) and has the lowest bias. On
the other hand, the reconstruction method is not affected by a significant
bias throughout the range of the Hegyi index values as can
be seen from the data points (solid line) in Fig. 4. The reflection
method is clearly associated with a positive bias for small Hegyi index
values, i.e. for little competition pressure, and with a negative
bias for large competition index values. Using no edge-correction
leads to a large positive bias throughout the range of the Hegyi index
values (note the asymmetrical point cloud in Fig. 4 and the
large slope value for the corresponding model).
All options of modifying and optimising the tree patterns in the
reconstruction that are listed in Section 2.1 under (2) (a)–(d) and
their combinations have performed successfully and effective.