Abstract:
As a preliminary step toward the establishment of a
70 mm. aerial photo timber inventory system capable of
generating accurate stand tables, stock tables, and gross
volume estimates, this project focuses on the development
of a system which (1) eliminates the need to measure tree
height on the photos through use of the tarif system and,
(2) incorporates the use of three distinct tree selection
methods-- fixed plot, variable plot, and line transect
sampling. The methodology is developed and applied to a
346 acre parcel of Douglas-fir forestland.
The accuracy of individual tree predictions for
diameter at breast height, volume, tarif number, and tree
height (which is derived without traditional photo
measurement) are evaluated by ground subsampling a 206-
tree validation set and comparing actual and predicted
values. The results show that an average underprediction
of ground-measured tarif number by about 5% occurs, which
in turn results in similar underpredictions of tree volume
and height.
The accuracy of the stand tables, stock tables, and
gross volume estimates are examined by comparison to
independently derived ground measurements. The photo
estimates of mean gross volume per acre, using each of the
three photo tree selection methods, are all within 5% of
the ground-derived estimates; in all cases the 68%
confidence intervals of the ground and photo estimates
overlap. Graphic comparisons of the photo and ground-derived
stand and stock tables are presented; they are
similar, except for the inability of the photo method to
accurately predict the stocking of trees less than twelve
inches in stem diameter.
The aerial photo system, using each of the three tree
selection methods, demonstrates an ability to produce
results comparable to those derived from conventional
ground inventory techniques. Future research is
recommended, and specific needs for this research are
identified.
