J. For. 112(1):51-54 EXPLORING THE ROOTS http://dx.doi.org/10.5849/jof.13-065

Accurately Measuring the Height of (Real)

Don C. Bragg

uick and accurate height1 measurement has always been a Q goal of . The techniques and technology to measure height were developed long ago—even the earliest textbooks on mensuration showcased hypsometers (e.g., Schlich 1895, Mlod- ziansky 1898, Schenck 1905, Graves 1906), and approaches to refine these sometimes remarkable tools appeared in the first issues of For- estry Quarterly, Proceedings of the Society of American Foresters, and the Journal of . For example, one such hypsometer based on the geometric principle of similar triangles (top of Figure 1) employed rotary mirrors to allow the user to simultaneously see the top and bottom of the tree in “proper parallax” (Tieman 1904). Other early hypsometers applied different approaches that used angles and dis- tance (e.g., Graves 1906, Detwiler 1915, Noyes 1916, Krauch 1918). Of these trigonometric hypsometers, those that calculated total tree height (HT) as a function of the tangent of the angles to the top (B2) and bottom (B1) of the tree and a baseline horizontal dis- tance (b) to the stem were most common (Figure 1). Because they are easy to apply and required only simple tech- nology, these approaches (hereafter, the similar triangles and tangent methods) have dominated tree height measurement. That is not to say the challenge of accurate tree height measurement was solved— many early foresters reported problems with getting consistent data Figure 1. Graphical representation of tree height measurement on in uneven terrain or dense understories or with the use of different sloping ground (the same mathematics apply for a level surface) types of hypsometers. Good measurement practices usually miti- using the similar triangles and tangent methods. Note that all gated these issues and became standard components of train- historic equations and figures referenced in this work follow the notation of these figures. ing programs. Others addressed these challenges by designing new techniques based on different trigonometric relationships. As an ex- ample, Haig (1925) proposed a slide rule solution that calculated tree Yet such corrections, although they improved the reliability of height using the sine law and slope distance from the observer to the indirect height measurement, failed to address other potential tree base (s1) sources of error. For these adjustments to work, the trees had to be ͑ ϩ ͒ truly vertical, i.e., the point at their base measured for the horizontal s1 sin B2 B1 HT ϭ distance was directly below the highest point at the top of the tree. As sin A 1 an example, if the subject tree happened to be leaning, the de-

(see the trigonometric example in Figure 1 for the angles involved). termination of A1 in Haig’s correction is less straightforward and In very steep terrain, the ability to use a slope distance rather than required additional measurements. Hypsometers based on similar horizontal distance (b) made height determination much simpler. triangles could be used on leaning stems, but the geometry of this McArdle and Chapman (1927) applied a different trigonometric approach requires the user to exactly match the angle of lean, a solution for sloping ground when determining tree heights. Using a daunting and imprecise task in most forested settings. For tree height ϭ fixed slope distance (s1 100 ft) from the eye of the observer to the measuring devices that used the tangent method, adjusting for lean ϭ ϩ base of the tree, they proposed the following: HT 100 sin B1 also presented a serious challenge. Falconer (1931, p. 744) succinctly 100 cos B1 tan B2. addressed this quandary:

Received August 19, 2013; accepted November 18, 2013. Affiliations: Don C. Bragg ([email protected]) is a research forester with the USDA Forest Service, Monticello, AR, and editor-in-chief of the . Acknowledgments: I recognize the contributions of the Native Tree Society for their efforts to develop and promote the sine method. Mike Shelton, Nancy Koerth, and Jim Guldin (all of the USDA Forest Service), Bob Leverett, and several anonymous reviewers all contributed to the development of this article.

Journal of Forestry • January 2014 51 The incorrect location or choice of the dle the plumb lines and the third to identify His seminal Forest Science monograph pro- point to which the measurement for dis- tance is taken is the only common and con- the point on which these two vertical planes vided such an insight: in his Figure 3, tinued source of error in measuring the intersect). Unfortunately, such an approach Grosenbaugh (1963, p. 8) provided the fol- ϭ Љ height of leaning trees. The error is due, not required further training of field crews, sig- lowing equation for “exact” height, H0 R to the actual chainage of the distance, but in nificantly increased the amount of time sin ␪Љ, where the height to the diameter mea- the location of the point to which the dis- tance is measured. The distance can be ac- needed to measure tree heights, especially in surement location (H0) was a function of a curately measured only when the point is heavy timber or rough terrain, and may have sloping baseline distance (RЉ) and angle be- located on the ground vertically below the necessitated adding staff for proper execu- tween the observer and the measurement lo- tip of the tree. tion. cation (␪Љ). He used sine rather than tangent In other words, because the tangent Such additional burdens did not endear at this point because the slope distance method projects height as a function of the themselves to forestry operations wanting to (whose measurement was possible using a horizontal distance from the observer to the streamline inventory procedures. Not sur- rangefinder-dendrometer) was required to apex of the crown, if a tree leans away from prisingly, then, less rigorous corrections for get the exact height above the ground for or toward the observer, the proper baseline tree lean and wide crowns became ingrained that point along the trunk. distance must be adjusted for accordingly. in mensuration textbooks and professional Grosenbaugh (1980, p. 204) later in- Think of how tree height measurement is curricula. The study of field-based tree troduced further measures to avoid den- often portrayed: the images are almost al- height measurement focused on optimizing drometry bias for leaning trees because ways straight, vertical, pointed crown coni- the efficiency of existing techniques (e.g., “…range, height, and (for optical forks) di- fers growing on level ground—an all-too- Barrett 1929, Morey 1931, Hunt 1959, ameter biases attributed to neglected lean to- often inadequate representation of “real” Bruce 1975, Rennie 1979, Long and Mohai ward or away from observer can be much trees in a more complicated forest environ- 1986, Williams et al. 1994) and the reintro- more serious and are not overcome by merely ment. duction and/or modification of largely un- tilting dendrometer and hypsometer to Hypsometers based on the mathemati- used tools (e.g., Curtis and Bruce 1968, match tree tilt in crosslevel.” This assertion cal principles of similar triangles and tan- Anunchin 1971, Buckner et al. 1977, Larsen questioned the use of certain ad hoc correc- gents incorporate assumptions that are only et al. 1987). Overall, most foresters were sat- tions for leaning trees. His use of spherical sometimes (rarely?) met in the field. Thick isfied with the degree of accuracy possible trigonometry yielded a more complete ver- undergrowth could be cleared and sloping from existing approaches (when the proper sion of the total tree height (Grosenbaugh Ϫ ground adjusted for, but tree lean and crown degree of fealty to adjusting for atypical trees 1980, p. 207), s2 sin B2 s1 sin B1, repre- irregularities proved much harder to ad- was observed) and that continuity in tech- senting the vertical difference between the dress. A fix for these was critical: Chapman nique over time trumped any possible im- crown high point (A) and the intersection of (1921, p. 246) noted that “(t)he error from pacts of systemic bias. To borrow a judicial the bole with the ground (C),where B1 and the measurement of broad-crown trees [un- phrase, it seems that height measurement B2 are the angles to the bottom and top of less corrected] is cumulative and tends to was considered “settled law.” the tree, respectively, and s1 and s2 are the over-estimate their heights.” To this end, However, in the latter half of the 20th corresponding “slant range” (the sloping Graves (1906) recognized the challenges of century, a new approach to tree height mea- baseline distance from his 1963 article) esti- measuring leaning and wide-crowned trees surement that could address the complexity mates from the observer to those points (Fig- and proposed corrections based on measur- of tree architecture appeared. Starting in ure 1). ing heights perpendicular to the lean and the 1950s, forest mensurationist Lewis R. For a truly vertical tree on level ground, moving further away to ensure that subordi- Grosenbaugh developed and refined den- this “sine method” yielded the exact same nate branches are not mistaken for the high- drometers capable of measuring tree heights, height as the tangent method (i.e., HT ϭ ͉b Ϫ ͉ ϭ ͉ Ϫ ͉ est point of the crown. Likewise, Krauch stem diameters, and bole volume under tan B2 b tan B1 s2 sin B2 s1 sin B1 ). (1918, 1922) suggested a number of adjust- many different conditions (e.g., Grosen- Figure 2 highlights the fact that for a leaning ments for the USDA Forest Service standard baugh 1954, 1963, 1980, 1981, 1991). tree, if the uncorrected distance to the stem ϭ ϩ hypsometer to improve its accuracy and Fundamentally, Grosenbaugh (1963, p. 27) (i.e., b b2 c1) is used to estimate height consistency in determining height. rejected the notion that it was not possible to rather than the adjusted baseline length (b2), Such ad hoc adjustments, designed for increase the precision and accuracy of tree the tangent method will overestimate height expedience, also came with their own as- measurement, faulting “[g]roup pessimism if the tree is leaning toward the observer and sumptions, for example, failure to identify after unsatisfactory trials of low-potential underestimate height if leaning away (not the actual nadir of the highest point of a dendrometers [that] has fostered a defeatist shown). Because the sine method directly leaning or wide-crowned tree (a distinct pos- attitude that has rejected dendrometers as measures slope distances s1 and s2, the proper ϭ ϩ sibility under many circumstances) means impractical but accepted the improper use of height (HT H1 H2) will be calculated that the horizontal baseline distance re- biased volume tables as inevitable” (emphasis regardless of the direction or magnitude of mained erroneous, and hence the estimate of added). In making this statement, Grosen- tree lean (Figure 2). In fact, the sine method total height remained inaccurate. Research- baugh recognized an inherent tendency of is also independent of ground slope, distance ers continued to offer improvements: Fal- many (perhaps most) foresters to embrace a from the stem, angle of observation, and vir- coner (1931) recommended using plumb correctable source of error. Although his tually all of the other assumptions of the bobs to identify the point on the ground work focused primarily on measuring diam- similar triangles and tangent approaches directly below the highest point of a leaning eter and estimating volume, accurately de- (Blozan 2006, Bragg 2008, Bragg et al. tree using a three-person crew (two to han- termining height was critical to this effort. 2011). This does not mean that foresters do

52 Journal of Forestry • January 2014 Figure 2. On level ground, it is easy to see the impact of an inappropriate baseline ؍ ؉ (b2 c1 b) on the estimated tree height using the tangent method: the proper base- line length (b2) must be determined to accu- rately estimate tree height. not face challenges in using the sine method to determine the actual total height of the tree—it is still possible to inadvertently se- lect a subordinate point on a crown (yet the sine height would still be correct for that branch). Modern laser rangefinders that in- clude the sine method can be used, however, to quickly scan multiple apparent high tops from the same vantage point (or even differ- ent viewpoints) to determine which is the maximum (i.e., total height) (Figure 3; see also discussion in Bragg et al. 2011). For Figure 3. An example of an offset crown and how this may influence height estimation using trees with broad, multileader crowns (the the traditional approaches. In these examples of mature eastern white pine (), deliquescent or decurrent growth form com- their apparent high points (white arrows) are noticeably offset from the vertical axis of their mon to most hardwoods), it can be particu- boles, necessitating adjustments with the similar triangles and tangent approaches to ensure accurate height measurements. It is also quite possible with the pine on the left that larly hard to identify the highest point (the the branch identified as the high point may actually represent a lower point in the crown “top”) using more traditional techniques, than a different top (black arrow). With laser rangefinders, it is possible to use the sine because each would have to be corrected for method to directly calculate the actual height of each of these points, permitting a direct independently to ensure proper height esti- comparison of multiple leaders to determine which is the highest (the “top”). mates (Bragg 2007). Although others had used elements of this approach (e.g., Haig 1925, McArdle lication on these approaches years after address the need for expensive equipment and Chapman 1927), Grosenbaugh’s appli- their introduction suggested that they had remained mathematically complex and re- cation of sine and slope distance for tilted failed to achieve much acceptance. Some of quired two observations points (Grosen- (leaning) tree hypsometry also translated to the barriers to acceptance probably rested baugh 1991), not exactly an efficient adjust- vertical trees growing on sloping ground and with the instrumentation: the rangefinding ment to make in the field, particularly in for trees with displaced tops (i.e., the highest dendrometer that he considered to be the rough terrain or heavy forest cover. point on their crowns not directly above the most efficient was admittedly expensive and Because Grosenbaugh’s focus was on base of the bole; Figure 3) (Grosenbaugh complicated (Grosenbaugh 1991). In addi- developing tools and techniques for foresters 1981, 1991). However, his continued pub- tion, the modified technique introduced to to calculate the dimensions and volumes of

Journal of Forestry • January 2014 53 leaning (slanted) trees, rather than their ver- Literature Cited new theory. Society of American Foresters, For- tical height, it can be understood that he did ANUNCHIN, N.P. 1971. An optical hypsometer. est Science Monograph 4, Bethesda, MD. not emphasize the potential of the sine P. 3–6 in Advances in forest mensuration (trans- 47 p. method for total height determination. After lated from Russian), Ovanesova, V.A., and GROSENBAUGH, L.R. 1980. Avoiding dendrom- etry bias when trees lean or taper. For. Sci. all, even though Grosenbaugh’s solution us- V.S. Uchenkov (eds.). Israel Program for Sci- entific Translations, Jerusalem, Israel. 26(2):203–215. ing sine (rather than tangent) was remark- AVERY, T.E., AND H.E. BURKHART. 1983. Forest GROSENBAUGH, L.R. 1981. Measuring trees that ably simple and elegant, it was plagued by measurements, 3rd ed. McGraw-Hill Book lean, fork, crook, or sweep. J. For. 79(2): the fact that it was very hard to measure Co., New York. 331 p. 89–92. slope (slant) distance with the technology BARRETT, L.I. 1929. Accuracy of Forest Service GROSENBAUGH, L.R. 1991. Tilted tree hypsom- then available. Fortunately, more recent ef- standard hypsometer. J. For. 27(5):587–588. etry. For. Sci. 37(6):1581–1590. BLOZAN, W. 2006. Tree measuring guidelines of forts led by canopy researcher Dr. Robert HAIG, I.T. 1925. Short cuts in measuring tree the Eastern Native Tree Society. Bull. East. heights. J. For. 23(11):941–944. Van Pelt, champion tree expert Robert Lev- Native Tree Soc. 1(1):3–10. Available online at HUNT, E.V. JR. 1959. A time and accuracy test of erett, Will Blozan, forester Bill Carr, www.nativetreesociety.org/bulletin/b1_1/ some hypsometers. J. For. 57(9):641–643. and numerous others have led to the adop- bulletin1_1.htm; last accessed Dec. 5, 2013. HUSCH, B., T.W. BEERS, AND J.A. KERSHAW JR. tion of accurate, relatively inexpensive, and BRAGG, D.C. 2007. The sine method as a more 2003. Forest mensuration, 4th ed. John Wiley portable laser distance measuring devices ca- accurate height predictor for hardwoods. P. & Sons, Hoboken, NJ. 443 p. 23–32 in Proc. 15th Central Hardwoods Con- pable of performing the sine method. Note KRAUCH, H. 1918. Some new aspects regarding ference, Buckley, D.S., and W.K. Clatterbuck the use of the Forest Service standard (gradi- that laser-based hypsometers have been on (eds.). USDA For. Serv., Gen. Tech. Rep. meter) hypsometer. J. For. 16(7):772–776. the market for decades and have become an SRS-101, Southern Research Station, Ashe- KRAUCH, H. 1922. A new method of measuring ville, NC. industry standard for many applications tree heights on sample plots. J. For. 20(3): BRAGG, D.C. 2008. An improved tree height (e.g., Williams et al. 1994, Carr 1996, Bragg 220–222. 2007, Farve 2010), but the default height measurement technique tested on mature southern pines. South. J. Appl. For. 32:38–43. LARSEN, D.R., D.W. HANN, AND S.C. STEARNS- MITH measurement programs built into most of BRAGG, D.C., L.E. FRELICH, R.T. LEVERETT,W. S . 1987. Accuracy and precision of the these devices still use the tangent method BLOZAN, AND D.J. LUTHRINGER. 2011. The tangent method of measuring tree height. and all its inherent assumptions. sine method: An alternative height measurement West. J. Appl. For. 2(1):26–28. Although digital analysis of light detec- technique. USDA For. Serv., Res. Note SRS- LONG, J.N., AND P. MOHAI. 1986. A simplified procedure for measuring tree heights directly tion and ranging (LiDAR) and other re- 22, Southern Research Station, Asheville, NC. 11 p. from slope distance. West. J. Appl. For. 1(2): motely sensed, high-resolution images offer BRUCE, D. 1975. Evaluating the accuracy of tree 58–59. new options for estimating total tree height, measurements made with optical instruments. MCARDLE, R.E., AND R.A. CHAPMAN. 1927. these tools will not be able to replace on-the- For. Sci. 21(4):421–426. Measuring tree heights on slopes. J. For. 25(7): ground measurements of individual trees for BUCKNER, E.R., J.L. RICHESON,J.RENNIE, AND 843–847. many applications. Hence, research into the W. BOYD. 1977. Modifying the Christen hyp- MLODZIANSKY, A.K. 1898. Measuring the forest efficacy of different height measurement someter for use in dense conifer stands. J. For. crop. USDA For. Serv., Bull. 20, Forestry Di- 75(3):139–140. techniques, particularly field studies de- vision, Washington, DC. 71 p. CARR, B. 1996. Using the Criterion 400 and Im- MOREY, H.F. 1931. A test of hypsometers on signed to compare various approaches across pulse 200 laser instruments to accurately mea- short trees. J. For. 29(2):233–237. a range of species, stand conditions, and to- sure trees. Compiler 14(4):6–12. NOYES, D.K. 1916. Comparative test of the pographies, should continue. The sine CHAPMAN, H.H. 1921. Forest mensuration. John Klaussner and Forest Service standard hyp- method offers a good blend of accuracy, pre- Wiley & Sons, New York. 553 p. someters. Proc. Soc. Am. For. 11(4):417–424. CURTIS, R.O., AND D. BRUCE. 1968. Tree heights cision, and efficiency, and its popularity will PHILIP, M.S. 1994. Measuring trees and , without a tape. J. For. 66(1):60–61. probably grow as its utility is increasingly 2nd ed. CAB International, Wallingford, UK. DETWILER, S.B. 1915. Saving labor in measuring 310 p. recognized. tree heights: A new principle applied to the RENNIE, J.C. 1979. Comparison of height-mea- hypsometer. For. Q. 13(4):442–444. surement techniques in a dense loblolly pine Endnote FALCONER, J.G. 1931. A method of accurate . South. J. Appl. For. 3(4):146- 1. To avoid confusion, this article applies the height measurement for forest trees. J. For. classic definition of total tree height as the 29(5):742–746. 148. SCHENCK, C.A. 1905. Forest mensuration. Univ. vertical distance between a horizontal plane FARVE, R. 2010. Evaluation of laser rangefinders. that intersects the bole at groundline and a USDA For. Serv., San Dimas Technology & Press of Sewanee Tennessee, Sewaneee, TN. parallel horizontal plane tangent to the high- Development Center, San Dimas, CA. 59 p. 71 p. est point of the crown (sensu Schlich 1895, GRAVES, H.S. 1906. Forest mensuration. John Wi- SCHLICH, W. 1895. A manual of forestry. Vol. III. p. 15–27, Avery and Burkhart 1983, p. 76, ley & Sons, New York. 458 p. . Bradbury, Agnew, & Philip 1994, p. 27, Husch et al. 2003, p. 99, GROSENBAUGH, L.R. 1954. New tree-measure- Company, London, UK. 397 p. and others). It is important to note that total ment concepts: Height accumulation, giant tree, TIEMAN, H.D. 1904. A new hypsometer. For. Q. tree height does not necessarily equal bole taper and shape. USDA For. Serv., Occasional 2(3):144–147. length or merchantable length under this Pap. SO-134, Southern Forest Experiment WILLIAMS, M.S., W.A. BECHTOLD, AND V.J. LA- definition, particularly in the case of leaning Station, New Orleans, LA. 25 p. BAU. 1994. Five instruments for measuring trees or specimens with broad, deliquescent GROSENBAUGH, L.R. 1963. Optical dendrometers tree height: An evaluation. South. J. Appl. For. crowns (see later discussion). for out-of-reach diameters: A conspectus and some 18(2):76–82.

54 Journal of Forestry • January 2014