Photogrammetry has become increasingly important in the practice of forestry. Recently, the trend has been toward the development of photo-mensurational techniques for direct estimation of timber resources. The purpose of the present study was to assess the possibility of applying aerial stand-volume multiple-regression equations for the application of photo-mensurational techniques on several kinds of air photos. Field data were collected from sample plots located in the U.B.C. Research Forest at Haney as well as from the forest on the campus of the University of British Columbia, in Vancouver. Modifications in technique for the determination of tree height, crown width and crown closure were developed by the writer and are described in this study. Multiple linear-regression equations were used for the analysis of data. Application of the Electronic Computer Alwac III-E to solve all the multiple linear-regression equations is described briefly. Ease of typing was evaluated subjectively. The present study has indicated: (1) Using a spherical densiometer, a ground estimate of crown closure in per cent resulted in an over-estimate, as compared with the photo-estimate. (2) Tree count could not be used effectively as an independent variable in the construction of the photo-volume equation. (3) Best results were secured when photographs: were taken with a 12-inch focal length and a flying height of 15,600 feet above sea level. (4) For the construction of photo-volume tables, height, crown width and crown closure should be used as independent variables, especially when more than one interpreter is involved. (5) No significant differences were found among photographic papers or finishes used for the determination of photo volume. (6) Photography with a Representative Fraction (RF) of 1:15,840 should be satisfactory for forest typing. (7) The greatest variation was among photo-interpreters. (8) Photo-interpretation could be improved by the standardization of photo-interpretation procedures. / Forestry, Faculty of / Graduate

Even-aged well stocked aspen stands in Northern Alberta were sampled in order to derive "normal" yields per acre for number and size of trees, basal area, and various measures of volume. The stands occur chiefly on three qualities of site which have been classified into Pair, Medium and Good by the average height of dominant trees at 80 years. Exceptionally high mortality per decade is characteristic in fully stocked young aspen stands and the relation of number of trees per acre to average diameter is markedly different from the trends found by investigators for other species., A table of stand density units was constructed to permit the, rapid calculation of stand-density index when average diameter and number of trees per acre have been determined. The mean annual cubic volume growth on medium sites is maximum at age 40. For merchantable cubic and board feet, growth reaches a peak at 85 and 130 years respectively. Data from two other regions show that the amount of cull is high in aspen trees above 80 years of age, this indicates that the wood should be utilized before a stand-age of 130 years is reached if maximum return in board foot volume is desired. By 140 years aspen stands show signs of breaking up. The hardwood stands measured were composed mainly of aspen but both white birch and black poplar occurred on many of the plots. The growth of the black poplar is only slightly less than that for the aspen. It has consistently lower volumes per tree which permits the use of aspen volume tables when correction factors are applied. / Forestry, Faculty of / Graduate

A great many kinds and forms of volume equations can be found in any review of the literature on this topic. At least four procedures are available for comparing volume equations to judge which one gives the best estimate of volume. These procedures are evaluated in this thesis for seven volume equation forms: [equation forms omitted] and two new methods are introduced: (l) Taylor's series for estimating errors and (2) sum of absolute deviations. Comparisons of above equations with different procedures were carried out for two species, lodgepole pine (Pinus contorta Dougl. var. latifolia Engelm.) and white spruce (Picea glauca (Moench) Voss.). Merchantable volumes of both species were described by three subsets of data which were grouped according to the following Girard form quotients for 32 foot logs: 1. 80+ 2. 70-79 3. <69. It was found that none of the presently available procedures, including the two new ones, can by itself indicate the best tree volume equation, which led to the conclusion that whenever more than a quick approximation is required, at least two of the methods should be used simultaneously. Both relative and absolute measures of residual variation must be considered. It is concluded that choice of equation form depends very much on the specific requirements of the research or operational project involved. / Forestry, Faculty of / Graduate

Variations in western hemlock tree and stand growth were analysed to determine the relative importance of causative factors. Four phases in a stand's life-cycle were investigated (seedling establishment, juvenile growth, early stand formation, and middle-age). Detailed analyses of tree growth were made at the time of early stand formation (25 to 35 years). Stand growth and yield data (15 to 160 years) were analysed by multiple regression methods to ascertain the effects of varying density regimes on time and magnitude of maximum wood volume per acre. In all stands studied, tree height was unaffected by variations in stand density. Detailed analyses of current breast-height radial growth indicated that the factors of antecedent growth, tree d.b.h., stand density, age, and site quality were of decreasing importance in the order listed. Statistically, the best measures of stand density were ratios of crown width and live crown length with total tree height, although stand basal area per acre was only slightly less significant. Stand fertilisation with nitrogen stimulated tree radial growth and changed the relative concentrations of foliar nutrients within the crown. Limited (15 trees) biomass studies indicated that production per unit area of above-ground tree weight (oven-dry) increased linearly with greater stand density. Yield increased directly in proportion to the fraction of area occupied by trees (stocking) and curvilinearly with degree of crowding (stand density) within the area occupied. Mean annual net volume increment did not culminate even at stand densities of 500 square feet of basal area per acre. Both intensive and extensive forest management was discussed. Within the Coastal hemlock zone there could be many advantages for a shelterwood regeneration system and two-storey high forest with western hemlock as a supplementary species. / Forestry, Faculty of / Graduate

A new taper equation is presented, log d = b₀ + b₁ log D + b₂ log 1 + b₃ log H where d is the diameter inside bark in inches at any given 1 in feet,D is the diameter breast height outside bark in inches, 1 is the distance from the tip of the tree in feet,H is the total height of the tree in feet and b₀,b₁,b₂ and b₃ are the regression coefficients. Two methods of deriving a compatible system of tree taper and volume equations are discussed.One method involves conversion of the logarithmic taper equation into a logarithmic volume equation.The other involves the derivation of the logarithmic taper equation from an existing logarithmic volume equation to provide compatabllity in volume estimation and at the same time ensure as a good fit as possible for the estimation of upper bole diameters (taper). Tests for precision and bias of volume estimates, carried out on the British Columbia Forest Service taper curves and logarithmic volume equations,indicate that the latter approach is preferable to the former. / Forestry, Faculty of / Graduate

A serial multi-resolution stereo matching algorithm is presented that is based on the Marr-Poggio matcher (Marr and Poggio, 1979). It is shown that the Marr-Poggio feature disambiguation and in-range/out-of-range mechanisms are unreliable for non-constant disparity functions. It is proposed that a disparity function estimate reconstructed from the disparity samples at the lower resolution levels be used to disambiguate possible matches at the high resolutions. Also presented is a disparity scanning algorithm with a similar control structure, which is based on an algorithm recently proposed by Grimson (1985). It is seen that the proposed algorithms will function reliably only if the disparity measurements are accurate and if the reconstruction process is accurate. The various sources of errors in the matching are analyzed in detail. Witkin's (Witkin, 1983) scale space is used as an analytic tool for describing a hitherto unreported form of disparity error, that caused by spatial filtering of the images with non-constant disparity functions. The reconstruction process is analyzed in detail. Current methods for performing the reconstruction are reviewed. A new method for reconstructing functions from arbitrarily distributed samples based on applying coordinate transformations to the sampled function is presented. The error due to the reconstruction process is analyzed, and a general formula for the error as a function of the function spectra, sample distribution and reconstruction filter impulse response is derived. Experimental studies are presented which show how the matching algorithms perform with surfaces of varying bandwidths, and with additive image noise. It is proposed that matching of scale space feature maps can eliminate many of the problems that the Marr-Poggio type of matchers have. A method for matching scale space maps which operates in the domain of linear disparity functions is presented. This algorithm is used to experimentally verify the effect of spatial filtering on the disparity measurements for non-constant disparity functions. It is shown that measurements can be made on the binocular scale space maps that give an independent estimate of the disparity gradient this leads to the concept of binocular diffrequency. It is shown that the diffrequency measurements are not affected by the spatial filtering effect for linear disparities. Experiments are described which show that the disparity gradient can be obtained by diffrequency measurement. An industrial application for stereo vision is described. The application is automated measurement of logs, or log scaling. A moment based method for estimating the log volume from the segmented two dimensional disparity map of the log scene is described. Experiments are described which indicate that log volumes can be estimated to within 10%. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

Although the conventional methods of predicting growth of trees and stands by the measurements of variables on the ground yield a considerable degree of accuracy, they take a long time and are expensive. Therefore, prediction of growth of trees and stands by the use of aerial photographs as developed herein may be preferable in some situations. A survey was made of ten series of aerial photographs taken at various times during the year with the use of several kinds of photography. One series was omitted because of too much exaggeration of relief. Of the nine series, three were regarded as inadequate for growth studies due to poor photography. Repeated measurements were made for 135 sample trees on different series of photographs in terms of total height and crown width. The tree images were classified as good, medium and poor according to the visibility of both top and the base of tree, and the accuracy of height measurements was defined in terms of standard error of the mean difference from photo-measurements and the ground data. In addition to the conventional method of taking an average of four parallax readings or two identical readings, the writer set up a checking method by which a high degree of accuracy was secured. The standard error of the mean difference for 95 heights of good image trees was ± 0.23 feet. When the good image trees were classified as conifers or hardwoods, the standard errors of the mean differ- ences were ± 0.30 feet and ± 0.41 feet respectively. Generally the measurements of height for conifers appeared to be more accurate than those for hardwood. There was no consistent result indicating which species gives better measurements. For height growth estimation, Height/Age curves were used as a supplement to ground data, and a method of pooling errors was applied. The use of this method for determining the accuracy of growth prediction was assumed to be applicable, and height estimates based on direct measurements on different series of photographs yielded a considerable degree of accuracy, if good quality photographs were available. The accuracy of crown width measurements was fairly high. The smallest standard error of the mean difference was ± 0.21 feet for 53 Douglas fir and the largest standard error of the mean difference was ± 0.58 feet for 12 alders. For growth estimates of crown width, the indirect approach of using Crown width/dbh ratios was used as a substitute for ground control. It was concluded that the prediction of growth of crown width can be made from a series of aerial photographs. The application of theoretical values as criteria in determining the acceptability of growth estimates of crown width might be practicable because variation will be reduced by the joint computation of two pairs of observations. However, when Crown width/dbh ratios are used as control, there tends to be a great deal of error, partly because of the use of ratios. Accuracy of growth studies on three different groups of plots varies with the composition of stand. The group of young natural regeneration plots showed a relatively high degree of accuracy. It was concluded that where the stand is comparatively open, growth estimates on the photographs would give a useful degree of accuracy. / Forestry, Faculty of / Graduate

Classification of site of forest land is possible on aerial photographs. This classification can be based on topographic features, physiographic features, forest cover types, or on their combinations. Aerial photographs of the University Research Forest were typed using the following topographic features: exposure, percentage of slope, shape in profile, and shape in contour. Data on topographic and physiographic features were collected on 238 sample plots within topographic types in 30-year-old stands, on 83 permanent sample plots in 70-year-old stands, and on 26 sample plots in old-growth stands. Both graphical and mathematical analyses were carried out to determine relationships among site index and thirteen site factors. Simple correlation coefficients for site index of each of 320 plots were highly significant for each of local and general position on slope, per cent of slope, elevation, soil depth, moisture regime, permeability, soil texture, and thickness of A₂ later. Shape in profile was significantly associated with site index. Aspect, shape in contour, and thickness of the humus layer were not significantly associated with site index. The best of the single factors was moisture regime, but use of this by itself could only account for 20 per cent of the variation inplot site indices. Linear multiple-regression equations were computed to estimate site index from various combinations of topographic and physiographic variables. These equations were not used further in this study for determination of site index because of their relatively high standard error of estimate; however, several potentially useful equations were recognized. The best multiple-regression equation was highly significant statistically but accounted for only 31 per cent of the variation in plot site index. It included aspect, local and general position on slope, per cent of slope, shape in profile, elevation, and moisture regime. A procedure was developed to estimate site indices directly from aerial photographs by stereoscopic examination. Photo-estimation of site index was much more accurate than the computed equations based on all data collected in the field. Standard errors of estimate were reduced from 23 feet to 16 feet by direct estimation of site index. Regression equations were developed for conversion of site index of Douglas fir, western hemlock, and western red cedar from one species to another and to the average of all three species. Site maps were prepared for the 30-year-old stands which had not been mapped in the 1950 inventory of the University Research Forest. Preliminary site and forest cover types were recognized and general stand and stock tables were developed to describe these 30-year-old stands. / Forestry, Faculty of / Graduate

These soil studies were carried out on eight permanent plots at the University Research Forest, Haney, B.C. These plots ranged in Douglas fir site quality values 80 to 180 feet at 100 years. The soil of each plot was examined by means of soil pits and soil samples were taken for laboratory analysis. For each plot, topography, vegetation, general moisture condition, degree of stoniness, nature of soil and for each profile depth of horizons, soil color, texture, structure, consistence the amount and position of organic matter, drainage, physiologic depth, depth to mottling and special features of the profile such as hardpan, orstein, shot material, root mats were determined. In the laboratory, per cent soil skeleton, texture specific gravity of soils, bulk density, total-pore space, macro-pore space, micro-pore space, hydraulic conductivity and available water between 0.1 and 15.0 atmosphere (corrected for soil skeleton and/or soil stoniness), soil acidity, total cation exchange capacity, exchangeable calcium and potassium, and organic matter content were determined. Statistical analyses were conducted to determine the significancy of the values obtained in laboratory analysis. Regression equations were determined for evaluating site index from morphological soil characteristics such as horizon thickness, depth to mottling. The soils of the area were mainly derived from glacial till. Till is a coarse texture (coarse sandy loam to loamy sand) containing gravels, stones, and boulders. It was generally hard and impervious to water movement. The profiles generally have shown Aₒₒ, Aₒ, A₂, B₁₁, B₁₂, B₃ and C horizons, ^he development of eluvial and illuvial horizons (especially clay accumulation) appeared somewhat weak. Profiles have shown coarse texture similar to that of glacial till. The depth of profiles varied between 5 and 33 inches, the average range being from 20 to 30 inches. Two main soil groups were observed among the eight plots. Plots 2, 3 and 4 were minimal podzol (Order: Podzols), plot 1 was concretionary brown (Order: Brunosolic), and plots 5, 6, 7 and 8 are classified as intergrade between concretionary brown and minimal podzol. Generally the coarse skeleton comprises more than 50 per cent of the soil. All the soils were very low in clay content. Hydraulic conductivity was high. Micro-pore space showed higher per cent layers than macro-pore space. The available water was low in general due to texture. The preponderance of stones in the soil had an adverse effect on the water-holding capacity of the soil. Consequently, stoniness had to be taken into consideration in determining the actual volume of available water. All the soils were acid in reaction. The organic matter was high in upper horizons and decreased with depth. Total cation-exchange capacity exchangeable, calcium and potassium and calcium base saturation were low. Correlation studies between soil properties and site quality indicated that the physical soil properties were more important than chemical characteristics as an indication of soil quality. Texture, bulk density, porosities (macro-, micro and total), hydraulic conductivity did not show any correlation with site quality but were significant at the 5 per cent level between the plots (except micro-pore space). Available water was positively correlated with site quality when it was corrected for stoniness and coarse skeleton. Chemical properties were not significant between the plots and did not show any correlation to site quality. Morphological features such as thickness of B₁₁ horizon and depth to mottling were highly correlated to site quality and gave the smallest standard error of the estimate (± 21.3 feet). The study indicates that the volume of available water in the soil was most important for the growth of Douglas fir and that the site index of an area could be determined from a study of those morphological features of the soil which affects its water-holding capacity, as well as other morphological features such as depth to mottling. / Forestry, Faculty of / Graduate

The photo-interpreter has a difficult task when he is asked to identify the images of tree species recorded on aerial photographs. When a tree is examined on an aerial photo for such a purpose, the difficulty becomes more and more evident as it is realized that the different species can not always he identified by eyesight, even on large-scale photographs, but must be viewed stereoscopically, and the variations in appearance within the same species, even growing under the same conditions, are very great. In comparison with identification on the ground, the interpreter must take an entirely new approach in the determination of various species from aerial photographs. This approach involves training the eyes to recognize plants appearing with various hues and grey tones on black and white photographs, at much smaller scale than usual in ground studies, in most cases from above or half-oblique view of the tree, which is strange to the inexperienced interpreter. Most of the trees appear on aerial photographs in vertical or oblique views, when the branching habit and the crown shape of a tree are easily visible. For this purpose it is desirable to know the characteristic branching habit and typical crown shapes of trees in order to use these factors in species identification. This leads up to the problem that will be presented in this thesis. Factors which influence the ground characteristics of three major tree species in British Columbia are examined and analyzed. Various crown forms, with which Douglas fir, western hemlock and western red cedar occur in the vicinity of Haney and Vancouver, are described. The basic pictorial elements, with which these species appear and enable us to recognize them on aerial photographs, are analyzed. Influence of different films and filters on the appearance of species on air photos are also discussed. The thesis presents an analysis of identifications of species made by several interpreters. Requirements for photo-interpretation are also discussed. Finally, a dichotomous key is presented, which is constructed for Douglas fir, western hemlock and western red cedar, taking into account their appearance at various ages and locations. / Forestry, Faculty of / Graduate