Managing Forest Ecosystems Anthonie Van Laar Alparslan Akca RUST M LU CIRUCLROForest Mensuration By Anthonie van Laar Somerses West, South Africa anéd Alparslan Akca Universivar Gottingen, Germany ¥) SpringerAC.LP, Catalogue record for this book is available from the Library of Congress ISBN-13 978-1-4020-5990-2 (HB) ISBN-13 97%-1-4020-3991-9 (e-book) Published by Springer, PO. Box 17, 3300 AA Dordrecht, The Netherlands. lowm:springercom Printed on acid-free paper AIL Rights Reserved © 2007 Springer 'No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser af the work.CONTENTS PREFACE 00 eee vii SUMMARY . ow. ee eee eee x Chapter 1: INTRODUCTION : wee eee 1 Chapter 2: STATISTICAL PREREQUISITES ........ . 5 1 INTRODUCTION 5 2 SCALES AND UNITS 6 21S 6 2.2 Units of measurement 7 3. GRAPHICAL PRESENTATION OF DATA 7 4 DESCRIPTIVE STATISTICS 9 5S PROBABILITY DISTRIBUTIONS ..........- -. DB Normal distribution. 0.6... 1B Binomial distribution 15 Poisson distribution... cre cree ee 16 Distribution of 2 Lee Lee .ou Distribution of swdent’st 26... 00. ee ee 1B 6 ESTIMATION .... . vee eee es 20 6.1 Bias, precision and accuracy _ - . 20 62 Estimators... a . 22 63 Estimating accuracy + 23 7 REGRESSION AND CORRELATION ANALYSIS |. - - 25 7 iple linear regression... . a L235 7.2 Correlation analysis Lee .- .. 26 7.3 Multiple regression analysis. . 26 74 Nonlinear reg a . 0 8 MOVING AVERAGE 0.00.20. eee 227 xiimage not available9 10 FREEHAND FITTING Contents SMOOTHENING BY FITTING EQUATIONS .... . Chapter 3: INSTRUMENTS. . 2 RELASCOPES AND PRISMS Angle gauges 2... Kramer's dendrometer mirror relascope . . ¢ mirror relascope . .. . Bitterlich’s telerelascope Prisms 3 EWEIGHT 2.2.0... 00.0.0... 2-0006 Telescopic poles Hypsometers ... Hypsometer according to o thet trigonometric ec principle Baseline slope correction Senet ee Hypsometer according to the geometrical principle . . 4 BLUME-LEISS RANGE-TRACER DRUM... . 5 TREE CROWN AND FOLIAGE «2... 22 ee ee 6 SHORT-TERM RADIAL GROWTH RESPONSES .. . . 7 INCREMENTCORES..............-..-5 8 BARK THICKNESS . 9 RECENT DEVELOPMENTS IN INSTRUME TATION, . o1 Laser dendrometer LEDHA 9.2 Criterion 400 laser dendrometer 9.3 Digital hypsometer forestor vertex Chapter 4: SI '- TREE MEASUREMENTS 1 MEASUREMENTS ON STANDING TREE: Ld Age see eee 1.2 Diameter and basal area DIAMETER-MEASURING INSTRUMENTS . Ll Calipers. . : 1.2 Biltmore stick. . . 13 Diameter tapes. 14 Permanent diameter tapes 1.5 Wheeler's pemtaprism 1.6 Finnish parabolic caliper 1.7 Barr and stroud optical dendr 1.3 Tree height 14 Stem form 28 29 33 33 33 35 36 36 37 38 39 39 39 39 40 B 44 45 47 a7 47 47 Sl 54 56 57 37 58 59 59 60 62 62 63 63 63 64 67 68Chapter 6; TAPER TABLES AND FUNCTIONS I 2 Chapter 7: TREE VOLUME TAB! > i Chapter 8: TREE AND STAND BIOMASS: > Contents SPATIAL DISTRIBUTION OF TREES ..... 2... 8.1 Testsofrandomness 0... beeen 8.2 Spatial structure ee 8.3 Structural diversity STAND DENSITY . 20. ee ee 9.1 Area-related indices 92 Distance-related indices TAPER TABLES . STEM PROFILE MODELS 21 Introduction . 2.2 Taper functions . sae 23 Polynomials and segmented ‘polynomials ES AND EQUATIONS INTRODUCTION VOLUME EQUATIONS WITH ONE PREDICTOR VARIABLE, 21 iple tariff functions. . 22 The incorporation of height into the tariff functi tion. EQUATIONS WITH TWO PREDICTOR VARIABLES . . . - 3.1 Graphic methods 32 Regression equations oe EQUATIONS WITH MORE THAN 1 VARIABLES. . « . MERCHANTABLE VOLUME ..........------ INTRODUCTION 2... 0. ee eee BIOMASS COMPONENTS 2.1 Branches 22. Foliage 2. Stemwood wei he. 24 Bark weight 2.5 Root weight. . 2.6 Pooling of data _ 2.7 Randomized branch and importance sampling . TREE-LEVEL REGRESSION MODELS .......- ADDITIVITY OF BIOMASS COMPONENTS .. . . - 4.1 Stembiomass . . a 42 Branch and needle biomass. 2... 4.3 Root biomass 134 134 137 139 140 140 146 . 49 149 ist 151 152 156 . 163 163 165 165 167 170 170 Wi 175 v7 . 183 183 184 184 185 185 186 186 187 187 188 192 193 194 194Contents 5 6 Chapter > Chapter xv DUMMY-VARIABLES FOR TREE SPECIES .. . . . 196 RATIO ESTIMATORS AND CLUSTER SAMPLING .... 198 9: GROWTHAND YIELD... ........2025 - 21 DEFINITIONS . - 201 THE GROWTH OF SINGLE TREES - 202 21 Growth parameters eee sees ~. 202 22 Stem analysis... peewee eee vee 203 209 213 31l Introduction oe eae sane .. 213 32 index curves Pee arin) 215 33 index equations - wae -. 217 THE GROWTH OF STANDS «0. 0 ee 222 4d Estimating stand growth based on actual measurements . ~ wee eee -+ 222 42 Stand table projection see wee .. 226 4.3 Recentdevelopments .... 2... veces 227 10; SAMP! iG FOR FOREST INVENTORIES... . . . 229 INTROBUCTIO! . 229 1 Sampting its . fee nee -. 232 PLOT SAMPLING. 22.200 oe eee 232 21 Plotshape..... baer ween e eae see 232 22 Plotsize 233 23 Plots on stand boundaries 238 24 Slope Correction 240 POINT SAMPLING 2000 ee ee eee 3.1 Basic principles - .: 3.2 Choice of basal area factor... . sae 3.3 Nonsamplingermors. 6... ee 3.4 Efficiency of point sampling . . SIMPLE RANDOM SAMPLING Al Sample size. ERROR PROPAGATION STRATIFIED RANDOM SAMPLING 6.1 Basic principles... 2.2... 6.2 Allocation methods... . 63 Poststratification 64 Block Sampling . . bone REGRESSION AND RATIO ESTIMATORS... - 7.1 Regression estimators 7.2 Ratio estimatorsu 12 16 Contents DOUBLE SAMPLING (TWO-PHASE SAMPLING) .. . . . 276 8.1 Double sampling for regression estimators... 276 8.2 Double sampling for stratification 8.3 Double sampling for ratio estimators . CLUSTER SAMPLING 91 Definitions 92 Estimators MULTISTAGE SAMPLING. 10.1 Two-stage samplin; 10.2. Three-stage samplin STRIP SAMPLING . . . SAMPLING WITH UNEQL TAL SELECTION PROBABILITIES 6.0.0 eee eee 12.1 List sampling (PPS sampling)... 2.0... 12.2 3P sampling. . . SYSTEMATIC SAMPLING SAMPLING PROPORTIONS .. . . 14.1 Simple random sampling 14.2 Cluster sampling wee aan . ESTIMATING CHANGES oc e eee eee 15.1 Independent and matched sampling see 15.2. Sampling with partial replacement (SPR)... LINE INTERSECT SAMPLING ee Chapter 11; REMOTE SENSING IN FOREST MENSURATION . INTRODUCTION . . 3t7 2 FUNDAMENTALS OF Al 3 DENDROMETRIC DATA . 3.1 Tree height 32 Number of tree crowns 3. ‘Crown closure 34 Crown dimensions 350 Age . 3.6 Profile of the stand’ 's grow ing space oe 4 ESTIMATION OF STAND VOLUME... ...--.55 . S ESTIMATION OF VOLUME INCREMENT ......... 344 APPENDIX . 2. ee M7 BIBLIOGRAPHY . 2.2.2.0... 0.0 00000-000000- 353 INDEX 377Chapter | INTRODUCTION Tree growth results from a sequence of physiologic: the formation of new cells, cell enlargement, and cell differentiation. Tree phy- siology deals with the study of these processes and has made a major contri- bution towards a better understanding of the causal relationship between the production of dry matter and many influential interacting variables. The science of forest growth and yield emphasizes the construction of models describing the relationship between growth parameters and influential predictor variables, and is based on forest botany, ecology, climatology, and use of forest mensurational techniques and inferential statistics to model tree and forest growth. To a large extent, these studies are of a phenological nature, indispensable to the forest manager in observing and quantifying growth phenomena in relation to time, site, genetic factors, and stand treatment. Forest mensuration provides the methods and tools to conduct such stud- ies. It concentrates primarily on the quantitative assessment of tree and stand characteristics at a given point in time during the life of the tree and stand, and provides the data required for efficient forest management. In line with the North American literature, the authors of this book contend that a discussion of empirical and analytical growth models, because of their technical nature, falls within the discipl discuss advances in process models. In conclusion, forest mensuration deals with the technical aspec' and forest stand measurements, such as + Measurement of tree and stand variables, e.g... diameter, height, bas bark parameters, and volume of standing and felled trees + Determination of form and age of trees and forest stands * Determination of the volume of standing and felled tre: + Measurements of the live crown and quantity of foliage processes, consisting of soil science, It makes extensi ¢ of forest mensuration, It is not the purpose of this book to of tree2 Intraduction * Estimation of biomass and biomass components of individual trees and stands * Estimation of the total and merchantable stand volume and its size class dis- wibution + Estimation of the diameter, basal area, height, and volume growth of single ‘ees and forest stands + Estimation of the damages to and the quality of individual trees and forest stands In addition, it has to deal with the development of models for the construction of tree volume, taper and biomass functions, the construction of stand tables, as well as the development of growth and yield models, Traditionally, terrestrial methods have been used to measure tree and stand variables. More recently, large-scale aerial photography has been applied sue- cessfully to replace some of the methods used in conventional ground surveys and it is for this reason that remote sensing methods have been incorporated in this book. In a broader sense, forest mensuration also deals with the estima- tion of volume and growth of large forest tracts, for example, in regional and national forest inventories, which are needed as a basis for forest policy deci- sions. This implies the application of mader sampling concepts and sampling methods, which make it possible to draw inferences about the relevant popula- tions. Since so many excellent texi books about forest inventory are available already, this book will not deal in detail with the methodology of regional and national forest inventories. A surplus of wood and a limited demand for forest products occurred during the early Middle Ages; therefore, there was no direct need to measure the grow- ing stock at periodic intervals. Towards the end of the Middle Ages, however, the increasing demand for timber necessitated some form of yield regulation. Felling concessions were very much limited to designated areas and replaced single-tree forest exploitation, In Central Europe, the early decades of the 18th century witnessed an increased involvement in a more scientific approach to forest measurements, Attempts were made to classify ees and forests accord- ing to their dimensions and their usefulness to the local population, although no exact Measurements were carried out. During the first half of the 18th cen- tury, foresters made a beginning to improve the customary ocular methods for estimating standing timber. In France Duhamel Du Monceau (1764) initiated dendrometry as an independent scientific discipline and in Germany, Octtelt (1765) issued descriptions for the determination of the volume of felled trees and stacked wood. The last decades of the 18th century and the entire 19th cen- lury witnessed a relatively rapid development of forest mensurational methods, Hennert (1791) developed xylometric methods to determine the volume of treeIntroduction 3 sections by measuring the amount of water displaced by the timber. Hennert also introduced sampling as a device to assess the volume of entire stands, Paulsen (1795) developed the first stem form theory and constructed the first yield tables. Cotta (1804) introduced the caliper and constructed the first vol- ume tables. These early developments paved the way to a scientific basis for forest management and forest yield studies, During the 20th century, there has been an emphasis on the construction and further development of better and more reliable instruments. In recent years, electronic devices for measuring tree dimensions and ring widths have been developed, At the same time, the application of more sophisticated sampling methods made it possible to obtain better and unbiased estimates at lower cost. In 1948, Bitterlich introduced the revolutionary angie count cancept, initially to estimate the basal area per hectare, Some years later, Grosenbaugh (1952, 1958) redefined Bitterlich’s variable radius method as Sampling Proportional to Size. After carly and largely unsuccessful attempts to use medium-scale aer- ial photographs for forest surveys, Bickford et al. (1963) introduced a two- phase sampling procedure, which combined information from a large sample of photo-plots with that obtained from a subsample of plots, which were remea- sured with conventional terrestrial methods. During the last decades, large- scale color photography was introduced for forest mensurational studies, with emphasis on the measurement of the effect of stresses on needle losses and dis- coloration. During this period also, satellite images were used for a variety of purposes, but primarily to classify forests according to forest type and to mea- sure or (o estimate the areas covered by forest, High resolution satellite imagery, together with digital data processing, opened a new era in forest mensuration on a global seale. The rapid development of electronic data processing and the increasing availability of powerful microcomputers has been of immense importance for mathematical and statistical advances in forest mensuration, particularly because of the inherent possibility of data storage and high-speed processing of quantitative information, Peripheral equipment and computer software to carry out stem analysis was developed and widened the practical usefulness of stem analysis (Johann 1977; Nagel and Athari 1982). In Central Europe, the concept forest mensuration (“Holzmesskunde”) is synonymous with dendrometry and stand measurements, It covers primar- ily the methods for measuring trees and stands, whereas the concept forest inventory CWWaldinventur”) although being based on dendrometrical methods, deals with estimations and inferences of the volume and growth of larger tracts. Tischendorf’s Lehrbuch der Holzmassenermitilung, Prodan’s classic4 Intraduction Holzmesskunde, Akga and Kramer's Waldmesslehre, Pardés's Dendrometrie, and Anuchin's Forest Mensuration were structured in line with these ideas. In North America, Bruce and Schumacher’s classic Forest Menstration, Spurt’s Forest Inventory, Meyer's Forest Measurements, Husch, Miller, and Beers’s Forest Menstration, and Avery and Burkhart's Forest Measurements com- bined forest mensuration with inferential statistics, sometimes with modeling, although the emphasis on statistics and modeling varied,Chapter 2 STATISTICAL PREREQUISITES 1 INTRODUCTION Forest mensuration deals with the measurement of trees and stands, They are mathematical variables representing different phy variables with a probability distribution. Processing information about these variables requires the application of statistics and computer technology. Some characteristics are continuous variables, which implies that they can theoretically take on infinitely many values, The diameter of the bole, for exam- ple, may be measured in millimetres but continuity implies that a further sub- division can continue indefinitely, although it makes no sense to measure the diameter in 1/10 mm, Discrete variables can assume a countable number of values, The number of trees within fixed-radius sample plots, the number of branches within a tree and the number of needles within a branch are discrete variables. The aggregate of individuals (irces, stands forests), for which information is required, is denoted as popitlation. In management inventories, information is required about the single compartment, in regional and national forest inven- tories the population is defined as the forest in its entirety, although it may be stratified on the basis of tree species or species groups, age class and site quality. The population is described by parameters, which are fixed quantities, not subject to variation. They may be size parameters, for example, the mean diameter or mean height of a stand, but also parameters of the diameter distri- bution, those of functions, which describe the relationship between diameter and height, between diameter and stem volume, etc. The concept population refers to a certain point in time. The parameters of a regression equation, which predicts the m may change because of tree breeding which produces hybrids or clones with a higher growth potential. Similarly, the mean annual volume increment of all 10-year-old Eucalyptus stands within a region is influenced by climatic cycles, new silvicultural techniques, ete al entities, and statistical anqual increment of a given species from site variables,6 Statistical Prerequisites Because of the high cost involved, it is impractical to measure all trees within a stand or all stands within a forest. Forest mensuration relies heavily on sampling procedures to obtain quantitative information about the resources at reasonable cost. A sample is defined as the subset of actual measurements within a given population. It is a random sample if each sampling unit has the same chance of being included into the sample, In forest inventories, the n plots or individual trees, which constitute the sample, are measured without replacing these sampling units after being drawn, Since this implies that a given sam- pling unit cannot be drawn more than once, the relevant population, contrary, for example, to controlled experiments, is finite, The construction of volume, taper, biomass functions and the development of growth models assumes that sampling, which is required to estimate the parameters of the mode tinue indefinitely. Conceptually, this population represents an infinite unive It ignores the fact that climatic cycles or a permanent change of the physical environment of the trees may have a profound effect on the coefficients of a given equation. in con- 2 SCALES AND UNITS OF MEASUREME! 21 Scales of measurement Different scales of measurement may be used for measuring tree and stand characteristics, * The nominal scale, used for attributes, represents the weakest scale of mea- surement, The observation is assigned to one out of & discrete categories, Species, provenance, forest type and soil type, for example, are discrete vari= ables which cannot be arranged in a certain order, + The next-strongest ordinal scale is a ranking scale characterized by ordered categories and is used for ranked variables (discrete categorical variables) The scale is characterized by classes of different but unknown width. Forest soils, for example poor, medium or good, the vital- ity of trees as healthy, sick, dying or dead, social tree classes as dominant, co-dominant, dominated and suppressed. * Almost all forest mensurational characteristics, such as diameter, height, ‘basal area, volume and increments, are continuous variables, measured on a metric scale + The metric scale is sometimes subdivided into an interval scale, without a natural zero-point and a ratio scale, which assumes the existence of a natural zero-point, Temperature represents the classical example of a variable which could be categorizedGraphical Presentation of Data 7 is measured on an interval scale when expressed in degrees centigrade or Fahrenheit and on a ratio scale when measured in degrees Kelvin. + In many cases, the ranking seale is converted into a metric scale by assigning numerical values to the class midpoints. Forest soils, for example, might be measured on the basis of soil depth or moisture-storage capacity or by some linear combination of these variables, with different weights being assigned to each of them, 2.2 Units of measurement Quantitative variables are measured either in the metric or in the English sys- tem, The latter was originally used in Great Britain, in countries of the British Commonwealth and in the USA, but many countries within the Commonwealth have converted to the metric system, The most important linear, square, cubic and weight measures in the metric system and their equivalent in the English, Russian and Japanese system are presented in Appendix D. Data representing continuous variables should be recorded with an appro- priate number of significant digits. This number is obtained by counting the number of digits between the first nonzero number on the left and the last digit on the right. A tree diameter with a recorded diameter of 56 em has two signif- icant digits and implies that the tree has a diameter anywhere between 55.499 and 56.5cm. A record of 56.4cm implies that the diameter has some value between 55.349 and 56.45 cm. In consequence, when diameters are recorded in centimetres, no digits should be written to the right of the decimal point and when recording the diameter in millimetres, there should not be more than one digit to the right of the decimal point. Reproducing the data with too many significant digits gives misleading information and suggests a precision which was not achieved. However, when tree diameters (in centimetres) and tree heights (in metres) are recorded with one digit to the right of the decimal point, it is justified to record the sample mean with an additional digit to the right of the decimal point, 3 GRAPHICAL PRESENTATION OF DAT: ‘A graphical display and interpretation of survey and research data is useful for different purposes: + For the forest manager, a graph may be more persuasive than a summary of the results derived from a fitted model.8 Statistical Prerequisites * Itis frequently necessary to calculate confidence and prediction intervals for the true mean of a variable. This calculation is usually based on the assump- tion of a normal distribution, Confidence and prediction intervals, however, are sensitive to deviations from normality. In consequence, it is necessary to verify whether or not the assumption of normality is satisfied. This ean be done either with the aid of histograms or stem-and-leaf plots, but alterna- tively, by using a statistical package, which provides estimates of the stan- dardized coefficients of skewness and kurtosis, The hisiogram and frequency polygon are usually based on grouped data rep- resenting either continuous or discrete variables, In histograms the frequencies are represented by equally wide columns, with heights which are proportional to the observed frequencies, in frequency polygons the frequencies are plotted on the y-axis and connected by straight lines. Both serve as a elue for the distri- bution function to be fitted. The observed frequency distribution, based on such grouped data, is also useful to estimate probabilities, for example, Pla by where a and. / are selected points of a given diameter distribution, The selec- tion of an appropriate class width is important for the construction of a his- togram, which reflects the distribution adequately. Tao few classes obscure the true shape of the distribution curve, whereas too many classes induce excessive variability amongst the class frequencies, A rule of thumb is to draw a sample, which is sufficiently large to ensure that 10-12 classes are generated, Sturges (1926) proposed the following function to serve as a guideline k=1+ 144410 N where " = number of observations and & = number of classes. For W = 50, the number of classes should be between 6 and 7, for N 100 between 7 and 8, whereas 9 classes are adequate for N = 250. To reconstruct the diameter distribution of a stand it is impractical to define classes with fractions of | cm as class widths. One-centimetre classes are normally used to obtain the frequency table, but I mm classes are required for research purposes, whereas a width of 2 or 4em is adequate for management inventories Example 2.1. The breast height diameters of 253 trees in a Pinus radiata stand are given in Appendix B. The trees were measured in millimetres and grouped in 1 and4 cm diameter classes. The frequency polygons based on these class widths are shown in Figure 2-1. In recent years, stem-and-leaf plots have become increasingly popular, par- ticularly for small samples, The observations are summarized by a number ofavailableimage: not - available380 log grades, 90 log rules, 89, 92 lognormal distribution, 14, 113 M maximum likelihood estimators, 104 mean arithmetic, 10, 30, 96-97, 112, 116, 118-119 geometric, 10, 66 mean diameter, 96-99, 118-120, 127-128 arithmetic mean, 96-97, 259 quadratic mean, 97-99, 105, 114, 117-120, 127, 132-133, 168, 214-215 mean tree method, 127 median, 10-12, 98 méthode du controle, 163, 201 mirage method, 239, 248-249 mirror relascope, 40-41, 43-44, 76 wide-scale, 43 model, 6, 7, 21, 24, 59 multistage sampling, 287-291 N nadir point, 320 Newton formula, 88 Neymann distribution, 134 normal distribution, 13-15, 17-19, 22, 99, 254, kurtosis, 12, 110, 254 skewness, 8 12, 22, 109-110, 254 dndex P parallax, 321-322, 324-328, 330 difference, 322, 324-326, 328 measuring, 325-326 percentiles, 10-12, 113-114 Petterson dbh-height function, 126 photo interpretation, 317-318, 323 photogrammetry, 317-318 analytical, 329-330, 332 digital, 329-330, 332 plot ‘on stand boundaries, 238-239 re-locating, 238-239 shape, 232-233 size, 233-237, 240 point sampling, 130, 241, 251 efficiency, 249 non-sampling error borderline trees, 247, 250 hidden trees, 247, 248 instrument, 246 instrument position, 247 sampling position, 247 slope correction, 248 Poisson distribution, 16, 134, 135 population finite, 20, 23, 252, 258, 263, 272, 288, 291, 304 infinite, 15, 288 PPS-sampling, 241, 295 precision, 20, 24, 124, 234, 262, 301 precision of height estimates, 124 pre-stratification, 215, 281 principal point, 319, 322, 328, 338 prism, 45, RIndex range-tracer drum, 56. ratio estimators mean of ratios, 199, 274 regression adjusted R?, 192 coefficient, 19, 85, 156, 168, 196, 225,274, 309 constraints, 168, heteroscedasticity, 132, 161. 165, 172,176, 199 homoscedasticity, 190 outliers, 21, 31 R-square procedure, 172 segmented polynomials, 158, 160 simple, 25 standard deviation, 255, 336 studentized residuals, 172 weighting, 171, 173, 190, 193, 259 regression analysis. 26, 89, 111, 165, 175, 196, 334 regression estimator, 20, 78, 117, 271, 277, 344 remote sensing, 2, 317-344 s sample plots 6-tree, 237 circular, 57, 237, 239,293 concentric, 237 sample size simple random sampling, 251, 259, 279, 286 stratified random sampling, 192, 262-269 sample trees felled, 19, 33, 82, 131, 185, 204 sampling centroid sampling, 77 control-variate, 77, 78 critical height, 133-134. importance, 76-78, 187-188 independent, 264, 270, 308-309, strip, 292-295 with replacement, 252, 266-267, 277, 295, 303-304 without replacement, 252, 254, 256, 258, 266-267, 304 sampling fraction, 231, 236, 252, 270, 282, 287, 290 Sampling Proportional to Size (SPS), 37 sampling proportions, 303-308 sampling units, 189, 230, 232, 251-253, 262, 266, 268, 287, 289-290, 295-206, 298, 300-305, 309-310 sampling with partial replacement, 308-310 sapwood area, 83-85 sawlog classes, 90 Heilbronner classification, 90 scale, 6-7, 42-43, 49-7, 303-304, 319-320, 325-326 metric, 6-7, 303, nominal, 6 ordinal, 6, 303 ratio, 6-7 Scribner rule, 89 sectionwise measurements, 86, 132, an shadow length, 323-324 sidelap, 320-321 significant digits, 7 simple random sampling, 251, 254, 259, 295, 303, 305 sample size, 251, 254, 256-259, 265-267, 290, 293, 303-305 site index, 213-221, 225-226, 318, 343 anamorphic curves 216availableavailable