0.31 with a range of -0.70 to 0.94 and a standard deviation of 0.31. A linear regression of tarif number on diameter was also fit for each plot of the form: Effects of the Tarif Number/ Diameter Relationship on Volume and Height Estimates T2 Gary W. Clendenen, USDA Forest Service, Pacific Northwest Research Station, Forestry Sciences Laboratory, Olympia, WA 98502 ABSTRACT. Volume estimates using the tarif system often assume a single mean tarif number for an even-aged stand. Con­ sistent trends of tarif number with dbh within plots were found in a set of 67 plots in western Oregon and Washington. Ef­ fects of these trends on plot and diameter class volume estimates, and on heights es­ timated from these volumes, were exam­ ined. Total plot volume estimates were ap­ parently not biased, but substantial biases were found in estimates of volumes and heights by dbh ,class. Height samples should be represlitative of the range of di­ ameters, adequate to detect trends in tarif number, and adequate to support alterna­ tive methods of Jolume and height estima­ tion. f West. f. Appl. For. 5(1):9-12. The Tarif system of volume estima­ tion was introduced to American for­ estry in 1963 with publication of Com­ prehensive Tree-Volume Tarif Tables by Turnbull et al. (1963). The tarif system was further described and extended by T u r n b u l l and Hoyer (1965), Brackett (1973), and Chambers and Jenkins (1976). The tarif system is often used for volume estimation, par­ ticularly when height samples are in­ adequate for deriving conventional height/diameter equations. Individual tree tarif numbers are averaged to es­ timate the tarif number for the stand or particular stand component. This procedure assumes that the tarif number is not related to dbh for the stand components of interest. Also, heights can be assigned to trees not m e a s u r e d in the f i e l d by using volumes estimated with the tarif system and rearranging a standard volume equation to give estimates of tree height. This paper reports a case study of the effects on volume estimates and heights estimated from these volumes when this constant tarif assumption is not met. METHODS Tree measurements were originally assembled on 67 plots in a study of the effects of burning slash after logging on the growth of young, even-aged stands in western Oregon and Wash­ ington (Miller et al. 1986). Average breast height ages for trees on these plots ranged from 14 to 33 years and site index (King 1966) ranged from 71 to 130 with a mean of 105. During analysis of the data, it became ap­ parent that tarif was positively corre­ lated with dbh on many plots. This vi­ olated a basic assumption of the tarif system-that in even-aged stands, tarif number is constant across dbh within a given stand-and raised a question about the applicability of the tarif system to this and similar data sets. Tarif numbers (T) were estimated for each tree by using volume esti­ mates (V) from the volume equation for young growth Douglas-fir b y Bruce and DeMars (1974) and equa­ tions given by Brackett (1973, p. 7). Mean tarif (Tl) was computed for each of the 67 plots. The mean slope was = b0 + b1 * dbh. (1) The slope coefficient, b1, was tested and found to be significantly different (P < 0. 100) from zero on 32 plots (48%). All these 32 plots had positive slope coefficients. In fact, 58 of the 66 plots with Douglas-fir (Pseudotsuga menziesii) present had positive slopes (Figure 1). The assumption of a con­ stant tarif number for all diameters within plots is clearly untenable for these data. Use of a single mean tarif number when tarif number is actually a func­ tion of dbh must introduce error in volume estimates. Two plots with more than 95% Douglas-fir by basal area were selected from the 66 plots with positive slopes, as examples of the effects of the tarif number/dbh re­ lation on volumes and on heights esti­ mated from these volumes. Plot A had the most significant slope coefficient (b1 0.73) with a t­ statistic of 6.18 (P < 0.001); plot B had­ the least significant slope <::_ e ficient (b1 0.07) with a t-statistit· of 0.44 (P > 0.500). Plot A had 84 Douglas-fir larger than 2.0 in. dbh, of which 27 were measured for total height. Plot B had 123 Douglas-fir larger than 2.0 in. dbh, of which 25 were measured for total height. Only Douglas-fir trees larger than 2.0 in. dbh were used. Other selected stand statistics for each plot are given in Table 1. Height trees on both plots were selected across the range of diameters. Total stem cubic-foot volumes, in­ cluding stump and tip (V1 and V2), were computed for each tree on each plot by using the two estimates of tarif (Tl and T2). Scribner board-foot = = 12 �------, 10 8 6 LL 0 0:4 w ID :E :::> 2 z -0.6 -0.4 ·0.2 0 0.2 0.4 0.6 0.8 1.0 SLOPE COEFFICIENT CLASS Fig. 1. Number of plots by tarif equation slope class. WJAF 5(1)1990 Reprinted from the Western Journal of Applied Forestry, Vol. 5, No. 1, January 1990. 9 Table 1. Plot statistics.1 Plot B Plot A Plot size in acres Age Site index Number of trees per acre Quadratic mean diameter Arithmetic mean diameter Mean tarif Percent Douglas-fir by basal area Total stem cubic-feet per acre Scribner board-feet per acre, 16-ft logs, to a 6-in top 0.45 27 92 187 7.4 7.0 22.1 98.4 1113 0.25 23 117 492 6.6 5.7 24.1 96.6 2544 2384 6640 1 Volumes computed using tarif adjusted for dbh. Table 2. Individual tree summary of results for Plot A and Plot B. Plot B Differences Plot A Differences Differences Max. Mean Max. Mean T1 minus n V1 minus V2 51 minus 52 H1 minus H2 5.05 -1.61 -9.38 5.60 1.59 0.09 -0.49 2.38 0.53 -0.47 -3.79 -0.90 0.28 0.00 -0.16 0.33 variables: H = total tree height in feet H1 = Has a function of T1 H2 = Has a function of T2 5 = Scribner board-foot volume, 16-ft logs, to 6-in. top 51 = 5as a function of T1 52 = 5 as a function of T2 T1 = mean tarif number T2 = tarif number fit as a linear function of dbh V = cubic-foot volume including stump and tip V1 = Vas a function of T1 V2 = Vas a fu ction of T2 1 , 30 I T 1 =MEAN TARIF NUMBER T2=TARIF NUMBER ADJUSTED FOR DBH 25 T2 -- PLOT A MEAN TARIF ----- PLOT A ADJUSTED TARIF -·-·-· PLOT B MEAN TARIF PLOT B ADJUSTED TARIF 15 �--�--�--��6 7 8 9 10 11 12 4 5 3 13 14 15 2 DBH CLASS {INCHES) Fig. 2. Mean tarif and tarif adjusted for diameter. 10 WJAF 5(1)1990 H1 H2 . DISCUSSION ' volumes, 16-ft logs, to a 6-in. top (51 and 52) were computed in the same way. Total height was estimated for each tree on each plot by solving the volume equation (Bruce and DeMars, 1974) for height and using the mea­ sured diameters and estimated volumes corresponding to the two tarif numbers (Tl and T2). These esti­ mated heights (H1 and H2) were fit as functions of dbh for comparison pur­ poses as follows: cant trend in tarif number across dbh exists as on plot A, there is no prac­ tical difference in total volume per acre, both cubic and Scribner, be­ tween the two tarif-based methods of estimating volume. Large differences between tarif esti­ mates by dbh class on plot A and small differences on plot B are evident in Figure 2. Plot A shows 2 8% more total stem cubic-foot volume for small /diameters and 8% less total stem ;, cubic-foot volume for large diameters when the mean tarif number is used instead of the tarif number adjusted for dbh (Figure 3); however, these dif­ ferences were of no practical impor­ tance in total stem volume per acre es­ timates by dbh class on plot B (Figure 4). Differences in Scribner board-foot volumes are even greater on plot A, ranging from 16% for 6-in. trees to -14% for 11 in. trees (Figure 3). Volume differences on each plot are a direct reflection of the differences in tarif number shown in Figure 2. There is a large difference in tarif based estimates of tree height (35% for small dbh trees) on plot A where a sig­ nificant trend in tarif across dbh exists and a small difference (generally less than 5%) on plot B where no signifi&..­ cant trend across dbh exist (Figure 5). = = f(V1,D) 4. 5 + EXP(b0 + b1 * D ** b2), (2) = f(V2,D) 4. 5 + EXP(b0 (3) + b1 * D ** b2), = RESULTS Plot A showed apparent differ­ ences, among dbh class, between T1 and T2, V1 and V2, 51 and 52, and H1 and H2. Plot B had no practical differ­ ences among any of the comparisons made (Table 2). Even when a signifi­ On most plots, the underlying as­ sumption of constant tarif across the range of diameters was not met. In­ stead, a consistent trend of increasing tarif number with increasing diameter was found. Such a trend would seldom be detectable with the limited height samples commonly used in ap­ plications of the tarif system. As in this study, Rustagi and Alegria (1981) also found that tarif number and dbh were not independent on several plots they examined at Pack Forest, WA. If tarif number is related to diam­ eter, and height samples are confined to a few large trees (a common prac­ tice among users of the tarif system), then estimates of mean tarif number will be biased, and estimates of both total volume and volume by indi­ vidual diameter classes will also be biased. Turnbull et a!. (1963) specify that trees should be sampled for tarif numbers from the group or �QJJ.PS of interest across the entire range of,clJ-, ameters present in the group, that a minimum of six trees should be sam­ pled on a given plot, and at least 20 trees should be sampled within each group of interest for the stand. Six trees are probably adequate if similar plots are to be combined into a stand estimate, but it is not sufficient to de­ tect tarif/dbh trends for individual plots. If individual plots must stand alone in an analysis (often the case in _ • SCRIBNER fillll CUBIC V1>V2 V1=VOLUME ESTIMATED FROM MEAN TARIF V2=VOLUME ESTIMATED FROM TARIF ADJUSTED FOR DBH V1 < V2 2 3 4 5 6 7 8 9 10 11 12 DBH CLASS {INCHES) 13 14 15 Fig. 3. Percent difference, by dbh class, on plot A between volume estimated from mean tarif vs. volume estimated from tarif adjusted for dbh for total stem cubic-foot volume and Scribner board-foot, 16 ft logs, to a 6-in. top. 30.--------------------------------25 lliU CUBIC V1>V2 w20 ----. • SCRIBNER u 15 a: lJJ u.. u.. 10 ' 1( Ci 5 o u , fiiJL V1=VOLUME ESTIMATED FROM MEAN TARIF V2=VOLUME ESTIMATED FROM TARIF ADJUSTED FOR DBH 171· rm. =·-=- - · .... ffi ·5 CL _, . ., .. V1 < V2 -10 ·15+--.--.--�-.--.� 2 3 4 5 6 7 8 9 10 11 12 13 14 15 DBH CLASS Fig. 4. Percent difference, by dbh class, on plot B between volume estimated from mean tarif vs. volume estimated from tarif adjusted for dbh for total stem cubic-foot volume and Scribner board-foot, 16 ft logs, to a 6-in. top. research), and anything more than total volume is wanted, then a larger sample is needed. Berry and Wiant (1967) and Rustagi and Alegria (1981) suggest using a prism sample for tarif tree selection, thereby weighting the selection to­ ward the larger trees. The most effi­ cient metliod of selecting the trees to be sampled will depend on how the ,data is to be analyzed, but it is essen­ /tial that the requirement for sampling across the entire dbh range be ob­ served. If height samples are distributed across the range of diameters, as spec­ ified in the original instructions of Turnbull et al. (1963), any bias in total volume per acre estimates introduced by use of a single mean tarif number is probably of little practical importance. A trend of tarif number with diameter, however, will result in substantial and potentially important biases in esti­ mates of cubic-foot and board-foot volume by diameter classes (as illus­ trated by plot A). These biases are more pronounced for estimates of board-foot volume, because board­ foot to cubic-foot volume ratios in­ crease with diameter. Heights can be assigned to trees notmeas ured i n the field1by using volumes estimated with the tarif system and rearranging a standard volume equation to give estimates of tree height. This procedure is some­ times used for purposes such as esti­ mating top heights on remeasured plots when height samples for a given measurement date are missing or in­ adequate and when tarif numbers by plot are smoothed over time to pro­ vide interpolated values. Biases sim­ ilar to those for volume estimates can be expected in estimates of dominant height or top height derived from a single-plot mean tarif number. RECOMMENDATIONS 40 w () z w a: w () a: w a. PLOT A PLOT B 30 25 20 H1=HEIGHT ESTIMATED FROM MEAN TARIF H2=HEIGHT ESTIMATED FROM TARIF ADJUSTED FOR DBH !:!: 15 c !z • l£ll H1> H2 35 10 5 0 5 10 2 3 4 5 7 8 9 10 11 12 13 14 15 DBH CLASS {INCHES) 6 Fig. 5. Percent difference, by dbh class, between height estimated from mean tarif vs. height estimated from tarif adjusted for dbh. It is therefore recommended that height samples for tarif estimation should be distributed across the full range of diameters of interest for volume computations. These samples should be sufficient to allow a sensi­ tive test of the slope coefficient in the regression of tarif number on diam­ eter. If the slope coefficient differs sig­ nificantly from zero and thert'f1s-1n­ terest in any tarif-derived stand s1'a""' tistic other than total stand volume, then either a separate tarif number should be used for each diameter class as calculated from the tarif/diameter regression, or alternatively, height/di­ ameter or volume/diameter curves can be used. D LITERATURE CITED BRACKETT, M. 1973. Notes on tarif tree volume computation. State of Washington, Dep. WJAF 5(1)1990 11 Natur. Resour., Olympia, WA. Resour. Manage. Rep. 24. 26 p. BRUCE, D., AND D. J. DEMARS. 1974. Volume equations for second-growth Douglas-fir. USDA For. Serv. Res. Note PNW-239. 5 p. CHAMBERS, C. J., AND D. F. JENKINS. 1976. Com­ prehensive log scale tree-volume tarif tables for Douglas-fir. State of Washington, Dep. Natur. Resour., Olympia, WA. 138 p. KING, J. E. 1966. Site index curves for Douglas-fir in the pacific northwest. Weyerhauser For. Res. Cent., Centralia, WA. For. Pap. No. 8. 49 p. MILLER, R. E., R. E. BIGLEY, AND S. N. LITTLE. 1986. The effects of slash burning on stand es­ tablishment and stand volume growth. Un­ publ. study plan on file at For. Sci. Lab. , Olympia, WA. RUSTAGI, K. P., AND J. ALEGRIA. 1981. The tarif system: An evaluation. Unpubl. manuscript on Purehosed by the Forest Service U.S. Department of Agriculture, for official use 12 WJAF 5(1)1990 file at Coli. of For. Resour. , Univ. of Wash­ ington, Seattle. 11 p. TURNBULL, K. J. , AND G. E. HoYER. 1965. Con­ struction and analysis of comprehensive tree­ volume tarif tables. State of Washington, Dep. Natur. Resour. , Olympia, WA. Resour. Manage. Rep. 8. 64 p. TURNBULL, K. J., G. R. LITTLE, AND G. E. HoYER. 1963. Comprehensive tree-volume tarif tables. State of Washington, Dep. Natur. Resour. Olympia, Wash. 127 p.