Issuu

AUSTRALIAN FORESTERS in PAPUA NEW GUINEA 1922-1975

PNGAF MAGAZINE ISSUE # 9B -5B of 30th June 2021 PART A.

FOREST MENSURATION PNG till 1975.

Bulolo Forestry College Mensuration Classes using Suunto clinometer. Photo credit J Clifford.

Archer1, G.R. 1972. A simple tree height converter for measuring tree heights in steep topography with poor visibility. Commonwealth Forestry Review 51(3): 246-253. Editor R B McCarthy2 2021. 1 2

Gary Archer TPNG Forests 1963-1973. Dick McCarthy District Forester TPNG 1963-1975.

FOREST MENSURATION – the science of measurement applied to forest vegetation and forest products provides value for basic ecology and sustainable forest management. It is grounded in geometry, sampling theory, ecology, and practical forest experience. Measurement of forested lands requires knowledge of the location and current volume of timber resources. Because forests are biological systems, estimates of growth for various management strategies are also required. Forest measurements are considered part of forest management. The role is to supply the numerical data for management decision making. The field of forest management is concerned with direct measurements, sampling, and prediction. An understanding of statistical techniques and sampling methods is essential for practicing foresters. Because forested lands are typically extensive in area and contain thousands of trees, foresters commonly measure only a sample of trees and then expand the sample values appropriately to obtain estimates of the population of interest. In relation to forested land management, a knowledge of the elements of land surveying is essential to the inventory forester. Direct measurements of trees require appropriate use of instruments to obtain the desired data. Examples include measuring tree diameter at breast height and measuring tree height using clinometers. (hypsometers) Tree measurement here includes measuring: • • • •

Products cut from tree stems. Measuring attributes of standing tress – diameter, height, form, age. Quantifying stand characteristics as volumes per land area. Past growth and predicting future growth of individual trees and stands of trees.

In recent times, the demand for more and better information from forests has prompted the development and application of new instrumentation in addition to persisting with continued on-ground measurement. These include laser and ultrasound-based measuring devices, use of drones, terrestrial and airborne laser scanning, satellite imagery and other remote sensing techniques are used to measure and monitor trees and forests. New methods have been developed for conducting tree-ring analyses including technologies for assessing the physical, chemical, and anatomic properties of wood. Such measurements are being used for reconstruction of growth conditions and investigations of the impacts of environmental changes on forest growth.

PART A. FOREST MENSURATION PNG TILL 1975. TABLE OF CONTENTS “FORWOOD” FOREST MEASURATION TECHNIQUES Ground Measurement Direction Measuring Standing Trees Tree Girth Measurement Tree Diameter at BH Basal Area Bark Thickness Tree Age Tree Height Tree Height and Log Measurement Mechantable Height Log Form Classes Stem Volume Tree Volume Tree Form Expressions: Taper tables and functions Volume Tables Estimation Methods Sampling designs Sampling intensity Calculating the Number of Plots Required Final Reliability Calculation Component Volumes Growth and Yield Models Tree Volume Table for Mixed Rainforest Species PNG Evan Shield’s Thesis 1965 New Sampling Methods Tree Volume Tables Individual Species PNG John Davidson’s E. deglupta Growth Studies 1970/71 TingTing I Kamap Processing the Resource Assessment Data TingTing I Kamap Gen Methodology re National TPNG Helicopter Forest Resource Assessment An Example of Data Compilation Tiauru Pandi Forest Assessment 1970 BIBLIOGRAPHY ACRONYMS

page 3 page 5 page 8 page 8 page 9 page 11 page 11 page 13 page 13 page 13 page 13 page 15 page 16 page 16 page 17 page 18 page 19 page 19 page 21 page 22 page 22 page 22 page 23 page 24 page 24 page 25 page 25 page 27 page 31 page 40 page 48 page 51 page 58 page 67 page 73 page 86 page 90

Girth Stick Dept of Forests Photo 1969 from publication Turi and the Trees.

“FORWOOD” The science of forest measurement in PNG (forest mensuration) was reflected in the development of forest assessment techniques for the location and estimation of current volumes of commercial timber resources in PNG, coupled with detailed mapping of landforms and species identification and distribution, in the period from the 1940’s to the 1970’s through the efforts of Australian foresters working in PNG’s forests. During this thirty year period, foresters had to address war reconstruction, resource owners concerns, training of PNG’s foresters, botanical issues as identification, timber attributes of various species, mapping, development of a viable timber industry, metrication, currency changes, self government issues, computerisation, digitisation, photogrammetry and development of timber markets for many unknown species. As described in PNGAF Magazine Issue # 9B-5A OF 8th June 2021 p 45, from the Department of Forests publication New Horizons3 p 8 in 1972, PNG forests are divided into seven fairly homogeneous systems. Forest System Specified Area in Volume in Volume in Timber Area thousands thousands of thousands of of hectares cubic metres cubic metres (M3) (M3) 0.5 – 1.5 metres 1.5 metres + GAB* GAB* Western Papuan Not available 3000 (est) Not available Not available Papuan South Abau 90 3540 4365 Coast Aroa 45 1179 1179 South East Sagarai Gadaisu 48 4250 3303 Coastal Musa 80 4719 2478 Kumusi 28 2360 2360 Ioma 182 11780 10380 Mai’ama 24 2830 2950 Bismarck Open Bay 104 4480 7080 Ania Kapiura 110 7080 5660 Kandrian 121 7790 8260 Arowe 157 8260 8730 Kapaluk 103 6135 6135 Melkoi 27 2360 1650 Nakanai n/a 2360 4955 Kaut 10 470 470 Solomons Torakina 13 945 1160 Sepik Ramu Vanimo 239 8495 10620 Middle Ramu 131 5665 5665 Highlands Giluwe n/a Not available 1840 Jimi n/a Not available 5190 TOTAL 1,512 81,600 83,000 * Girth above buttress. 3

“New Horizons – Forestry in Papua New Guinea a Dept of Forests Publication 1973 Jacaranda Press.

The above table details an appreciation of the work performed by Australian foresters to establish the first PNG national forest inventory i.e., the standing assessed commercial forest resource by major forest types. The Department of Forests reported that the forest resources were extensive and varied in composition. Some 200 species had economic potential but only some 30 species accounted for the bulk of merchantable timber. By 1972, approximately onethird or 4.4 million hectares had been classed as having economic potential and assessed to a level where development planning was possible. Forest Systems 4 Dept of Forests PNG 1970. In 1946 some 14,518,219 hectares had been investigated (figures reviewed in 1957) of which 5273240 hectares were under 10 degree slope and the area of potential forest type was estimated to be some 274731 hectares. The national forest inventory in PNG was largely completed by 1968 describing standing log volume and areas for specified timber areas. This was described in New Horizons 19725p7. Of the PNG total land mass of 46,886,000 hectares, tree cover was some 40,000,000 hectares of which commercially accessible forest was some 13,000,000 hectares. By 1975 some 4,800,000 ha had been covered. By the late 1980’s, over 7,400,000 hectares had been covered. After the mid 1960’s, - foresters had the advantages of developments in transportation such as helicopters. In the late 1960’s and early 1970’s data processing moved from mechanical calculators to the advent of then primative elctronic computers. By 1975, foresters had established the location and commercial volumes of timber resources by various botanical forest types throughout PNG. They had developed global techniques for the assessment of commercial sawlog and pulpwood volumes together with volume table compilation for the mixed rainforest species. They had identified botanical and timber characteristics for the various species . This magazine (PNGAF Magazine Issue 9B-5B) describes the forest mensuration criteria under which the Department of Forests undertook the national forest inventory. It was a team effort because as previously described, it required correct mapping (refer PNGAF MAG ISSUE # 8 of 1 February 2021 – Department of Forests Mapping Section), correct botanical 4 5

Dept of Forests PNG 1970 Forest Systems of Territory of Papua and New Guinea. New Horizons – Forestry in Papua New Guinea a Dept of Forests Publication 1973 Jacaranda Press.

attributes (refer PNGAF MAF ISSUE # 6 of 8th February 2021- Lae Botany) and timber attributes (forthcoming PNGAF MAG ISSUE). It was after 1975, that advances in photogrammetry occurred with the introduction of satelllite coverage of forested areas. This enabled a much more rapid appreciation of forest cover although detailed species and volume aspects still required much ground work. In recent times since 1975, although PNG was and is still interested in the commercial volumes of standing timber, the demand for more and better information from forests has prompted the development and application of new instrumentation in addition to persisting with continued on-ground measurement. These advances allow PNG to undertake: • • • • • •

Growth and yield/productivity studies in natural and planted forests. Measuring the effects of silvicultural management practices on stem form, wood quality, tree growth and stand development. Development of total and merchantable volume equations for estimating the biological asset value of tree stands, and biomass tables for carbon estimation. Development of non-destructive sampling methods for volume, biomass, and wood quality estimation. Studying the impacts of long-term environmental changes to the growth of trees and productivity of forests. Advanced scientific methods for measuring and monitoring the impact of environmental conditions such as climate change on forest productivity, structure, and diversity.

Forest Measurement Techniques . Ground measurement using a steel tape called a chain and usually 100 feet in length. The term chain comes from the form of the early tapes which were composed of 100 links, each one foot long. Brass tags were fastened at each ten links. For sloping countryside, techniques using the topographical abney to correct for slope categories were used.

Photo credit Wikipedia. The fundamental unit of horizontal measurement is the surveyors (Gunther’s) chain of 66 ft. (22 yards or 20.1m). The chain is divided into 100 equal parts known as links (7.92 inches or 201mm). It was designed and introduced in 1620 by English clergyman and mathematician Edmund Gunter (1581–1626). It enabled plots of land to be accurately surveyed and plotted, for legal and commercial purposes.

Photo credit Wikipedia. Link chains were later superseded by the steel ribbon chain. The length of a cricket pitch is one chain (22 yards).

Start of survey measuring the distance. Dept of Forests Photo 1969, from publication Turi and the Trees. 8

Direction A compass is a magnetometer that shows the geographic cardinal directions (or points) used for navigation. Usually, a diagram called a compass rose shows the direction north, south, east, and west on the compass face.

Figure p 3 Arthur Ramsay6 1966 Volume 2 Forest Surveying PNG Forestry School TPNG.

Compass bearing. Dept of Forests Photo 1969, from publication Turi and the Trees. 6

Arthur (Blue) Ramsay TPNG Forests 1960 – 1975.

Forestry College Bulolo group photo was taken of the staff and students. L to R Norma Collis; Olga Woolcott; Heiner Streimann; Arthur Ramsay; Leon Clifford; Evan Shield; Gerry Cullen; Vivienne Shield; Pat Cattanach. Photo Credit Janelle Clifford 1966.

Measuring Standing Trees Tree Girth Measurement Girth is a measurement of the distance around the trunk of a tree measured perpendicular to the axis of the trunk. It is measured at breast height or at 4.3 feet (1.3 stick m) above ground level or above the buttress. Girth Note that the breast height value is a measurement grandfathered from decades of forestry applications. It was developed because of the ease and simplicity of measurement. There is a modern day move to change it to chest height.

Itam with a girth stick measuring T. brassii Mom Hill swamp, September 1964. Photo credit Ken Granger.

Girth Stick Dept of Forests Photo 1969 from publication Turi and the Trees. 11

Ken Granger 7 recalled that it was John Lake8 who thought of using girth sticks on the 1964 Bougainville survey. Ken was unaware if this method was ever documented. They may have been developed by Eric Hammermaster9 in the earlier surveys of Tonolai Harbour where T. brassii were dominant and too big to measure with a girth tape. Ken used them first on the Empress Augusta Bay survey in 1964 (the photo of Itam was measuring a T. brassii on that survey). They became standard issue from then on using milled timber rather than a bit of bush carpentry to make the stick. The markings were derived from a diameter tape and each mark represented one foot of girth. The technique was to line up the edge of the stick on the left side of the tree and then count off the number of “feet” marks. For smaller trees they always used a girth tape at breast height or above the buttress if necessary. Gary Archer10 recalled that unlike the photo in Ken's document, the girth sticks in later (late 1960s, early 70s) assessment surveys were mounted on poles and held above the buttresses. The sticks were, as Ken stated, recorded in girth units in feet rather than diameter. The ones Gary used had alternate black and white sections for each girth class and large numbers painted on them for easy reading at a distance. The Forest Department's publication "Turi and the Trees" 1968, (also translated in Pidgin as "Turi na Diwai"), had photos of the girth sticks in use. They were straight sticks on a pole, not modified for line of sight like a Biltmore stick, and when being used the reader had to read one end and then step sideways by the approximate width of the bole to read the other end. Bill Finlayson11 developed a variant of the Biltmore stick for use above buttresses (Finlayson and Archer, 1964), but it required using a measuring stick to stand 20 feet away from the tree. For this reason, Eric Hammermaster favoured using the simpler unmodified version, which required stepping sideways when reading it. John Davidson12 recalled that when he was on the Ioma survey in 1967 they had girth sticks about four feet long marked up to read girth in feet directly from the large numbers painted on the stick in alternating black and white sections. In the centre at the back a large metal D was attached with screws. This was to accept a pole to raise the stick horizontally above any buttress. The pole therefore was not permanent and a new one was cut from the bush to the required short or longer length and fitted into the D on the girth stick if required and discarded before moving on to the next plot. If none of the trees in a plot had a tall buttress (say all less than shoulder height) the stick was held up against the trunks directly by hand without the need to cut and fit a pole. Not having any length of pole permanently attached also meant that the girth stick was convenient to carry on the helicopter pannier (the survey team usually folded it up in their tent fly so it could not slip out and fall overboard during the flight!).

Personal Communication April 2021 Ken Granger TPNG Forests. John Lake TPNG Forests 1962-1968. 9 Eric Hammermaster TPNG Forests 1956-1979. 10 Personal communication April 2021 Gary Archer TPNG Forests 1963 to 1973. 11 Bill Finlayson TPNG Forests 1963-1965. 12 Personal Communication April 2021 John Davidson TPNG Forests 1962-1980. 8

Tree Diameters. Tree diameters are measured at breast height (1.3 m) outside bark from the uphill side using diameter tapes, tree calipers or Biltmore stick or above buttress. Diameter at breast height (DBH) is a standard method of expressing the diameter of a tree or stem of a standing tree. It is measured at breast height of approximately 1.3 m (4. 3 ft) above ground. Diameter at breast height

photo credit Wikipedia The diameter tape in measuring the girth of the tree is calibrated in divisions of p1 centimetres (3.14159cm). The measure assumes the stem has a circular cross section and gives a directly converted reading of the diameter. DBH is used in estimating the amount of timber volume in a single tree or stand of trees utilizing the allometric correlation between stem diameter, tree height and timber volume. Basal area and mean diameter. Tree stem diameter measurements are often converted to cross sectional areas. The cross-sectional area at breast height is called basal area. Bark thickness is measured using a bark gauge, taking a minimum of two readings.

. Tree Age. The age of a tree is defined as the time from the gemination of the seed or sprouting from the cutting from which the tree developed. For rainforest stands, they are normally referred to as uneven aged, as individual trees vary in age and size classes. For plantations, the tree age is taken from the date it was planted. Estimating age from annual rings (examining cross sections in felling operations or use of increment borers in living trees) is only possible where there is a distinct growing season e.g., northern hemisphere. In many tree growing areas of the world, e.g., tropical, and southern temperate zones, tree growth is generally not characterized by annual rings. Hence it may only be possible to approximate individual tree ages.

Bulolo Forestry College Mensuration Classes lower montane forest. Photo credit Janelle Clifford.

Tree Heights Instruments used for measuring tree heights in PNG.

Figure p 3 Arthur Ramsay 1966 Volume 2 Forest Surveying PNG Foresry School TPNG .

Bulolo Forestry College Mensuration Classes using Suunto clinometer. Photo credit Janelle Clifford.

Bulolo Forestry College Mensuration Classes. Photo credit Janelle Clifford.

Tree Height and Log Measurement Tree height is the vertical distance between the base of the tree and the highest tip at the top of the tree. Tree height can be measured several ways with varying degrees of accuracy. Short trees using a long pole. The basic trigonometric principle is described below. Stick Method. Tree heights can also be measured remotely from the ground. The most basic remote height methodologies are all variations of the stick measurement. Stick Measurement Method (Wikipedia)

Tangent Method. A second method uses a clinometer and tape method, or the tangent method is commonly used in the forest industry to measure log length. Some clinometers handheld devices used to measure slopes. The(1965-6 user can sight Ken Granger13are recalls that before the introduction of theangles Suuntoofclinometers I think) to the top of a tree using such a clinometer and read the angle to the top using a scale in we used Abney levels to measure tree height/log length. the instrument. Before clinometers, topographic abney levels were used for tree height measurement. Abney levels are calibrated so when read at a distance of 66 feet (20 m) from the tree, the height to the tree above eye level can be directly read on the scale. Sine Height Method A third method to overcome the limitations and errors associated with stick method and the tangent method can be overcome by using a laser rangefinder in conjunction with a clinometer. Today a hypsometer incorporates both devices into a single unit.

(Wikipedia) Issues in tropical forest assessment. E.g., the relascope has a slight hole in the back and a clear window at the front to allow the user to sight through the tool. There are three light inlet holes which are used to light the scale. This is one of the major problems with relascope because the way it is designed, it cannot be used in low light as frequented in rainforest environs. Merchantable Height refers to the usable portion of the tree stem. This is defined as the usable length from stump height to an arbitrary upper stem dimeter still considered utilizable.

Log Form Classes The form class of all logs in PNG was recorded using the following definitions. Note that form classes A to D must have a minimum log length of twenty feet. FORM CLASS A B

C D

E F

DESCRIPTION Round and straight. A good peeler log Round but not straight. A marginal peeler log. Within any 20-foot section, logs are permitted to sweep in one direction only and the sweep must not exceed one quarter of the mid - diameter of the section (see diagram). Straight but not round. A good sawlog. Neither straight nor round. A marginal sawlog. Within any 20-foot section, logs are permitted to sweep in one direction only and sweep must not exceed one quarter of the mid - diameter of the section. With bends or sweeps more than these allowed for in classes B and D on a log shorter than 20 feet. Suitable for chipping. Unsuitable for any purpose due to defect or fault or shorter than 20 feet.

Form classes and Bulolo Forestry College Mensuration Classes Natural A. hunsteinii (klinkii) tree. Photo credit Janelle Clifford.

STEM VOLUME is function of a tree's height, basal area, shape, and depending on definition, bark thickness. It is therefore one of the most difficult parameters to measure, because an error in the measurement or assumptions for any one of the above factors will propagate to the volume estimate. Volume is often measured for specific purposes, and the measurement and interpretation of the volume estimate will depend on the units of measurement, standards of use, and other specifications. For example: •

Biological volume is the volume of stem with branches trimmed at the junction with the stem, but usually excluding irregularities not part of the natural growth habit (e.g., malformation due to insects, fungi, fire, and mechanical damage).

•

Utilisable or merchantable volume excludes some volume within irregularities of the bole shape caused by normal growth in addition to those irregularities not part of natural growth. For example, the volume contained in the swelling around a branch node may be excluded because this volume could not be utilised (by a nominated user).

•

Gross volume estimates would include defective and decayed wood.

•

Net volume estimates would exclude defective and decayed wood.

Thus, the type of volume use measured must be reported for reliable interpretation. For example, the net merchantable volume of sawlogs in a tree will be significantly different to the gross biological volume. Calculations of merchantable volume may also be based on true cubic volume or product-oriented volume. Product oriented volume is the volume of a nominal product that could be cut from the log or stem under specified conditions and assumptions. Direct and indirect methods for estimating volume are available. The direct methods tend to divide the stem into theoretical or actual sections and measure the volume of these sections: • •

Graphical method Standard sectional method

The indirect methods include: • • •

Volume tables Volume equations Integrating taper equations

TREE VOLUME When tree stems are cut into lengths, they are called logs. The process of measuring volumes of individual logs is called scaling. Logs are scaled in terms of cubic metres. Tree volume is the most widely used measure of wood quantity. It is usually estimated for the assessment of economic value or commercial utilization potential. Depending on management objectives, predictions of tree volume in this magazine refer to commercial tree volumes above ground to a certain merchantable limit excluding bark. To calculate tree volume, the tree is divided into several segments with the successive diameters being the bottom and top of each segment and segment length being equal to the difference in height between the upper and lower diameters and the corresponding volume calculated. Cumulative trunk volume is calculated by adding the volume of the measured segments of the tree together. Tree Form Expressions: Taper tables and functions Because trees taper, often irregularly from stump to top, it is common to make some evaluation of stem form in the construction and application of tree volume tables. The rate of tree taper varies with species, age, dbh, and tree height. If a series of diameter measurements are taken at intervals along the stem, average taper rates may be derived for groups of trees characterized by a particular shape or form category. Such tabulations are referred to as taper tables.

Construction of log volume and log taper tables Gogol Survey Camp 1970. Photo credit Ian Whyte. 19

Although tree cross sections rarely form true circles, they are normally presumed to be circular for purposes of computing cross sectional areas. However, as logs taper from one end to another, only short sections could be treated as cylinders. There are several common geometric solids from which truncated sections could be extracted to approximate log forms as described in the following figure of geometric shapes assumed by different portions of tree stems.

Volumes of these solids of revolution are computed as follows. Name of solid Volume computation Paraboloid Area/2 x length Conoid Area/3 x length Neiloid Area/4 x length As a rule, trees approximate the shape of truncated neiloids while the effects of butt swell are apparent. Logs from middle sections of tree stems are similar to truncated paraboloids, while upper logs approach the form of conoids. Cubic volumes of all solids of revolution are computed from the product of their average cross-sectional area and length. Three common formulas applied are listed below.

Areas and volumes are computed under bark. Limitations of the three formulae are recognised by foresters in volume calculations. • • •

Huber’s formula assumes the average cross section area is found at the midpoint of the log but that is not always true. Smalian’s formula requires measurements at both ends of the log. It is the easiest and least expensive to apply. Newton’s formula requires measurement of logs at the midpoint and at both ends. It is more accurate than the above two formulae but more expensive.

FAO described the estimation of tree volume from dimensional variables such that as diameter and height. Classical volume models include the so-called combinedvariable equation:

And the more general model:

where, in both cases, α, β and γ are coefficients, D refers to diameter (usually measured at 1.3 m above ground level), and H refers to total height, merchantable height, or merchantable length as defined for a given application. The latter model is often used after logarithmic transformation:

where estimates of β often approach 2 while estimates of γ approach 1. These basic equations implicitly assume a single-stemmed form and may require modification or replacement for species with a more complex form. A TREE VOLUME TABLE is a tabulation that provides the average contents for standing trees of various sizes and species. These tables are based on volume equations and use correlations between certain aspects of the tree to estimate the volume to a degree of certainty. The principal variables associated with standing tree volume are diameter at breast height (dbh), merchantable stem length, and tree form (this type of volume table is referred to as a standard volume or multiple entry table). The preferred method of constructing tree volume tables is by regression analysis. Difficulties occur when estimating the form class for sawlog merchantability of the tree in question. 1- way table – normally diameter at breast height (dbh)or girth above buttress (gab) or basal area is the only measurement. (Sometimes called a local table). 2-way table. Dbh or gab and height. (Sometimes called a standard table). 3- way table. Measurements that correspond to bark thickness and taper. (Regional table). Before the general availability of computers, volume tables were compiled by summarizing the volume of trees in dbh or gab and height classes. Volume equations are developed now.

Tree volume equations like tree volume tables are a statement of the expected volume of a tree of nominated dimensions in a particular stand or population. The input variables to the equations can also include diameter (at breast height and other heights), height, taper, and interaction terms. Some equation forms have been given specific names.

Volume curve: An equation that relates dbh to volume as a polynomial or log-linear line. The relationship is normally curvilinear and convex to the abscissa. Volume curves include the following relationships (where v denotes volume, d denotes diameter at breast height and a1-a3 are constants):

: Volume line: An equation that relates volume to basal area as a straight line (v=a+b*g). The volume line approach is simple, and fitting requires fewer sample tree measurements than the volume curve. Estimation Methods Sampling Designs. Sampling versus complete enumeration. The objective of sample surveys is to gain information about a population. When all individual units of the population are observed, the survey is termed a complete enumeration. Complete enumerations are extremely expensive and time consuming to perform. The objective of all sampling is to make some inference about a population from the observations composing the sample. The method of selecting the non-overlapping sample units to be included in a sample is referred to as the sampling design. Many statistical procedures assume simple random sampling. By this approach, every possible combination of sample unit has an equal and independent chance of being selected.

Sampling intensity To plan a forest assessment that is statistically and practically efficient, enough sample units should be measured to obtain the desired standard of precision. There is a need to ensure plot size is selected on the most efficient size to commensurate with variability produced. 22

Sampling Intensity It is important that the approximate level of reliability is required before beginning an assessment. All available information should be utilised to estimate the number of sample plots required for each forest type. Guidelines may be obtained from previous reconnaissance information in the area itself or similar areas by using the formula given below. In the absence of such information, the following sampling levels may be taken as approximate guidelines. Sawlog Volume Approximately 200 acres of sample (approximately 600 x 66 feet radius plots) are required for reliability of + or – 5 %. Approximately 50 acres of sample (approximately 150 x 66 feet radius plots) are required for reliability of + or – 10 %. For reconnaissance level surveys, it has been found that a minimum of 40-50 x 66 feet radius plots is required to give any meaningful estimate. Pulp Volume Approximately 60 acres of sample (800 x 33-foot radius plots) are required for reliability of + or – 5 %. Approximately 15 acres of sample )200 x 33-foot radius plots) are required for reliability of + or – 10 %. Calculating the Number of Plots Required for a Meaningful Sample The actual number of plots required may differ from the guidelines according to the quality and variability of the forest being assessed. Sampling requirements during the survey can be reviewed as data becomes available by extracting a sub - sample of say 30-40 plots, compiling the plot volumes by hand from the relevant volume tables and then analysing the results.

Final Reliability Calculation After an assessment has been completed the reliability of the estimate of the mean is calculated as follows.

Component Volumes The above guidelines apply only to estimates of total volume of all species combined. If volume estimates for individual species or size classes are required with comparable reliability, then a considerably larger sample may be necessary. It may in fact be uneconomic to take sufficient sample to reliably predict the volumes of species or size classes of limited occurrence. The highest sampling intensities used by the Department of Forests has been a 5 % sawlog sample for logging/working plan assessments. This gives 50 acres of sample within each 1000 acres of forest, which in general permits total sawlog volume of all species for each 100-acre unit to be estimated within approximately + or _ 10 % but gives volume estimates of considerably lower reliability for single species or size classes. The amount of sample to be taken should be a management decision based on the area of the forest unit for which a reliable estimate is required, the costs of sampling and the value of the resource being investigated. If the area contains a reasonably high proportion of an economically important species (e.g. Anisoptera, Intsia, Pometia), then there may be a case for calculating the size of the sample required for a specific species. This can be done by extracting plot volumes for the single species and analysis of these. Stand dynamics (i.e., growth, mortality, reproduction, and associated changes in the stand) can be predicted through direct and indirect methods. There are many difficulties in modelling uneven aged stands such as natural rainforest stands.

Growth and Yield Models. Forest management decisions are predicted on information about both current and future resource conditions. Inventories taken at one instance in time provide information on current volume stand related statistics. Forests are dynamic biological systems that are continuously changing, and it is necessary to project those changes to obtain relevant information for prudent decision making. TREE VOLUME TABLES FOR MIXED SPECIES IN PAPUA/NEW GUINEA14. Based on V=80.31549 +2.18592G² - 1.15235H + 0.64224G²H Where V = log volume under bark in super feet true measure Where G = girth over bark above buttress in feet Where H = log length in feet (log height above buttress height) and a Bark allowance 3inch off girth. Gary Archer15 took the Original Imperial equation, with girth and height in feet and volume in super feet: V = 80.31549 + 2.18592 G^2 - 1.15235 H + 0.64224 G^2 H And converted it to metric, with diameter in centimetres, height in metres, and volume in cubic metres: V = 0.185582 + 0.0000536586 D^2 - 0.00873586 H + 0.0000517235 D^2 H In relation to a sustained yield concept in PNG, the felling cycle was based on a 35-year rotation on a selection logging model.

Based on 1965 thesis, Evan Shield14 completed his dissertation into PNG forests volume tables titled The Application of New Sampling Methods to Previously Inaccessible Tropical Forest Areas, with Reference to Papua New Guinea. Commonwealth Forestry Review vol 55 No 1 March 1976. https://ora.ox.ac.uk/objects/uuid:56c8977d-af05-40d4-9d3a-54959ffcfe00

Personal communication Gary Archer 13th June 2021 TPNG Forests 1963-1973.

TREE VOLUME TABLE FOR MIXED RAINFOREST SPECIES in PNG16

Extract page 28 Table 2 Tiauru Pandi TA Logging Plan 1970-75 R B McCarthy TPNG Forests 1963-1975.

Evan Shield’s Thesis 1965 The Application of New Sampling Methods to Previously Inaccessible Tropical Forest Areas, with Reference to Papua New Guinea In 1965, Evan Shield17 completed his dissertation into PNG forests volume tables titled the application of new sampling methods to previously inaccessible tropical forest areas, with reference to Papua New Guinea. Commonwealth Forestry Review vol 55 No 1 March 1976.

The link to the thesis on the Oxford University Research Archive is: https://ora.ox.ac.uk/objects/uuid:56c8977d-af05-40d4-9d3a-54959ffcfe00 This thesis is a classic mensuration piece of its era. But, at the same time as reflecting on the development of forest mensuration methodology in PNG, it attempts to move forward from the 1960’s, addressing the advent of forest statistical methodology and sampling methods combined with the arrival of computers and their combined impact into the field of tropical natural forest mensuration. In his thesis, Evan addresses the need to evaluate the forest resource as a preliminary to forest industry development. The purposes of forest resource inventory were defined as supplying information for: • • •

long term policy planning the evaluation of forest areas for exploitation investment direction of more detailed management investigation.

The requirements of a single inventory were to provide data on the physical, biological, and economic aspects of a forest area. In reviewing the application of forest resource inventory in PNG as of 1965, considering existing and potential forestry development, at no time had entirely objective methods been employed in assessing the forest resource. Subjective methods have no place in long term extensive forest inventory. Evan discussed the use of sampling techniques in inventory and the nature of the statistical populations which must be recognised. These populations are based on utilisation

Evan Shield TPNG Forests 1958-1971.

requirements although silvicultural requirements cannot be overlooked. He delineates that populations based on species, size and form are essential to both requirements. The necessity to delimit, within an inventory area, those areas of similarity of occurrence of particular populations was demonstrated in the light of utilisation and silvicultural requirements. Such demarcation was termed stratification. A review was made of the methods of stratification of an inventory area by aerial photo interpretation. Species identification, photo-mensuration and canopy-aspect mapping were the methods reviewed. He concluded that canopy-aspect mapping provides the greatest potential for reaching the ideal stratification based on species and size composition. He then discussed improvements as improved photography (which today would imply satellite coverage) and multi staged sampling techniques. Evan thinking back then in 1965, was of the difficult task of inventory of a large area of virtually inaccessible tropical forest as the PNG forests. He paid much attention to the use of aerial photographs to determine both the forest area and to stratify it broadly by type. Improvements to inventory mensuration methods are addressed. e.g., the use of small circular plots is superior. Evan discussed issues that need further addressing as: • •

the influence of errors on volume calculations defect assessment.

On page 79, Evan discussed the concept and use of log volume estimation as “limiting volume” (L.V.). This was an extension of an approach adopted in Tasmania (Lawrence 1960). This “limiting volume” concept was to avoid confusion with the term log volume as accepted by the trade where the volume was calculated by multiplication of log length and sectional area one half log length. This was Huber’s formula. On page 80, in the volume estimator for PNG, the standard volume was used as the dependent variable. Independent variables were girth at the base of the log (a more concise term than above buttress) and log length the latter being the difference between log height (from ground level to break of crown or major defect) and buttress height (from ground level to the base of the log). Both independent variables were readily measured in practice Evan discussed the impact of electronic data processing on volume table construction on page 77. He refers to Appendix 4 (page 105) and the testing of the fourteen mathematical models against which volume data for PNG species were tested. The model selected as best fit was based on an index developed by Furnival (1961). On page 82 of the thesis, Evan detailed a standard volume equation which had application as a general, all species model for native tropical rainforests in PNG.

Gary Archer18reported that the above equation is not only unusual in that it includes H2 components, as Evan notes in his thesis. It is also unusual in lacking a constant term, because volume models for larger trees do not automatically pass through the origin when extrapolated back to zero. In Evan’s thesis, model 14 in Appendix IV is ranked the highest, with the lowest Furnival Index. Model 14 is in fact the Australian equation, not the above model with H2 terms in it. The Australian equation generally performs well in most cases, so that does not surprise Gary. Evan Shield 19 confirmed that the first volume table for PNG rainforest mixed species was a component of his work at the University of Oxford in 1965. He had collected the raw data for this project over the couple of prior years and had played around with graphical and formula fitting without great success. After substantial assistance of Howard Wright at the CFI in Oxford, together, they managed to get the data onto punched paper tape and Howard generously allocated some part of his allotted time on a University of Oxford computer somewhere on the campus. (He cannot remember exactly). Several mathematical models were tested and - again without now being completely sure - a weighted, multi-variate model was deemed the superior one. (Refer Appendix four page 105 Shield Thesis). Even at this earliest stage (1965), Evan was aware that there remained much more work to be done, especially in considerably increasing the range of species and the tree count in the raw data. From the 1960’s to now, probably little has changed.

18 19

Personal Communication Gary Archer 26th May 2021 TPNG Forests 1963-1973. Personal communication Evan Shield 28th April 2021 TPNG Forests 1958-1971.

Appendix IV p 105 Evan Shield Oxford Thesis

TREE VOLUME TABLES PNG INDIVIDUAL SPECIES. Gary Archer20 reported much work was done in the period 1973-75 on detailed volume and merchantable conversion factors by the Commonwealth Forestry Institute in Oxford. This was from measurements taken on small-sized exotic plantation trees, mostly less than 30 cm diameter, for E. robusta, E. grandis, P. patula and P. caribaea, mostly from Kindeng and Lapegu plantations with similar work done for A. cunninghamii and A. hunsteinii, apparently up to 50-55 cm diameter where measurements were presumably made on plantation fellings at Bulolo.

Personal communication Gary Archer 29 May 2021. TPNG Forests 1963 to 1973.

Professor John Davidson21 in the following section describes the work that was undertaken to produce the volume table for E deglupta as part of his PhD studies.

Personal communication John Davidson and Dick McCarthy 26 th April 2021.

Professor John Davidson’s Project re E deglupta Growth Studies 1970/71 One of the conditions that John was permitted to go to Canberra to do his PhD on a Commonwealth Scholarship was that he would continue to work on E. deglupta and continue work on volume tables and productivity investigations (See K J White’s22 letter). (He was eventually granted continuity of service and promotions as they fell due by the PNG Public Service Commission for the time he spent in Canberra, and the ANU granted him leave to return to Keravat for field work from November 1970 to January 1971, an almost unheard-of concession, granted only by Kevin White being deputised as his external PhD supervisor for his time in PNG!)

Hundreds of trees harvested from thinning trials, provenance trials, spacing trials and plots were felled specifically for volume compilation in different age compartments at Keravat, Dami and Baku. The total height, and over-bark and under-bark diameters at 10 cm and 50 cm height above ground, breast height (1.3 m), 2 m above ground then at one-metre intervals along the stem to the tip were measured by Jeff Fairlamb, John Dalton and Alan Williams and John Davidson. Heights from 10 cm above ground (stump height) to each of 5, 10 and 15 cm diameter small end under-bark were also measured. Alan Williams working in the research office at Keravat derived the total volume and volumes to each of the small end diameters of all the trees by a graphical method on large sheets of millimeter graph paper using the metric system. The computing services of the ANU were used on the data. While John was at ANU, data were also sent to Dr Howard Wright, Commonwealth Forestry Institute (CFI), Oxford, UK who provided without charge computer computation, interpretation, and compilation of one of the sets of volume tables. Complicating this project was that the ANU worked with the metric system, while the data coming in from PNG were in imperial units from the growth plots and metric from the research section at Keravat. The imperial data had to be converted to metric in Canberra for use at the ANU.

K J White TPNG Forests 1957 – 1977.

However as the CFI was working in imperial units Dr Wright was sent both original imperial data where it already existed plus data from Keravat that were converted from metric to imperial units before dispatch.23As well as the primary variables of volume (V) in cubic feet, diameter at breast height (D) in inches and height (H) in feet, various functions and ratios of these variables such as D2, 1/D, 1/D2, Log (V), DH, D2H, V/D2H, 1/DH, 1/D2H, H/D2, H/D and Log (H) were derived to provide additional variables for testing in form factor and volume regression analyses. The metric data set was transformed at the ANU using a Fortran program to also create an equivalent imperial data set, which was sent to Oxford. A preliminary volume table for E. deglupta in imperial units based on 200 trees was received in Canberra from Oxford in late 1971. This comprised nine sets of tables giving total volume (V) under bark (UB) and merchantable volume UB to 3 inch, 4-inch, 8 inch and 12-inch diameter upper limits for diameter at breast height (D) and total height (H) as variables, and the same for diameter and “predominant height” (Hdom). 24

Metric equivalents were 7.6 cm, 10.2 cm, 20.3 cm, and 30.5 cm, respectively. Predominant height was defined in PNG at the time as the mean height of the tallest 20 trees per acre. A typical half-acre plot was divided into ten parts and the tallest tree in each part was measured. (This was equivalent to 50 trees/ha or 10 trees on a 0.2 ha plot.) 25

In Oxford, for total height (H), 15 regression models were tried for best fit to the data using different variables where V, D and H are as described above, a is the regression constant and b, c and d are regression coefficients. The logarithmic functions are to the base e (natural logarithms - ln). 1.

V = a + bD

V = a + bD + cD2

V = a + bD2

V = a + bD2H

V = a + bD2 + cH + bD2H

V = a + bD2 + cDH + bD2H

On 12 June 1970, the Australian Metric Conversion Act passed by the Australian Parliament was given assent. This Act created the Metric Conversion Board to facilitate the conversion of measurements from imperial to metric. PNG followed the same timeline as Australia over the protracted period of introduction of the metric system across various industries. However, the forest and timber industries took several years to be fully metricated.

(lnV) = a + b(lnD)

(lnV) = a + b(lnD) + c(lnH)

V/D2 = a + b/D2 + c/D

10. V/D2 = a + b/D 11. V/D2H = a + b/D2H 12. V/D2H = a + b/D2 + cH/D2 + dH 13. V/D2H = a + b/D2H + cH + d/D2 14. V/D2 = b/D2 + cH/D + dH 15. V/D2H = a + b/D2H + c/H + d/D The regression model of best fit was chosen by reference to various parameters describing the regression including the lowest Furnivall index and highest multiple correlation coefficient. Regression model number 8 from the list above was best. The resulting equation, based on diameter at breast height over bark (D in inches) and total height (H in feet), was: ln V = -6.298 + 1.6458(lnD) + 1.2368(lnH) For predominant height (Hdom) the best model was the so-called Shield equation weighted by D2Hdom. V = -6.194 + 0.16Hdom + 0.002823D2Hdom – 0.001151Hdom2 – 0.00000334D2Hdom This equation produces negative volumes for some of the extreme values of D and Hdom because of the very quadratic nature of this regression. However, these extreme values were well outside of the range of actual data used and would also be well outside the range of D and Hdom to which the equation would be applied in practice. The 3-, 4- and 8-inch merchantable volume conversion factors (F) to be applied to reduce total volume to merchantable volume under bark were derived from an exponential function of breast height diameter (D) of the type: F3,4 or 8 = a + bcD

The values of the regression constant a and regression coefficients b and c to be substituted in the above equation to arrive at each conversion factor were as follows for each of the three smaller merchantable diameter limits: Merchantable top Constant a diameter 3 inches (7.6 cm) 0.9944 4 inches (10.2 cm) 0.9923 8 inches (20.3 cm) 0.9850

Coefficient b

Coefficient c

-5.0949 -4.2986 -7.5477

-0.6476 -0.5039 -0.2871

The 12-inch (30.5 cm) diameter factor was derived from a quadratic regression: F12 = -2.1036 + 0.2514D – 0.00529D2 for values of D up to 23.0 inches [58.4 cm]. For diameters greater than 23 inches a constant 0.883 could be used as the conversion factor. Fortran computer programs were written to generate tables of volumes and conversion factors across a range of D, H and Hdom. A stand table was prepared to show the range of data used and it was pointed out that caution should be observed if extrapolating outside that range. Within the range of D from 4 to 24 inches (10.2 cm to 61 cm) and H or Hdom from 40 feet (12.2 m) to 180 feet (55 m) the percentage errors in the mean predicted volume were likely to be small as indicated in the following table. D (inches) 4 8 8 8 12 12 12 16 16 20 20 20 24

H (feet) 40 60 80 100 100 120 140 140 170 140 160 180 180

Total Height 1.9% 0.8% 0.8% 1.3% 1.5%

Predominant Height 2.2% 1.3% 1.2% 1.3% 1.2% 2.1% 1.4% 2.6% 2.7%

Using the stand volume tables and error estimates given here to determine the volume of individual trees is not appropriate. Values of D, H and Hdom entered in the formulae should be stand mean values derived from measuring a prescribed number of trees in plots to return a stand mean volume result. From 1972, Oxford began to work in the metric system, which greatly simplified matters! 43

Keravat 1973 (four photos below). Harvesting plots of E. deglupta for pulpwood volume table preparation. Felled stems were measured for diameter over and under bark at ground level (10 cm stump), breast height and one-metre height intervals and for total height. Billets were cut to one-metre length and stacked with bark on and with bark off to calculate stacked roundwood volumes. The data were processed using graphical methods to provide whole tree volumes over and under bark. Two-way volume tables were prepared based on diameter at breast height and total height and separately to several top diameter limits. The metre-long billets with bark on and off were stacked as shown in the photographs to determine the stacked volume of roundwood to simulate measuring pulpwood logs on log transport or in the log yard at a mill, that is to derive conversion factors for stacked roundwood volume to true solid wood volume for various diameter classes.

Keravat 1973. Thinning trials in E. deglupta. Top: Row thinning (complete removal of every second row). Bottom: Conventional thinning from below on an area basis, favouring the release of potential final crop trees (marked in advance with a band of white tape at breast height).

After they had compiled a database of some 800 trees it was possible to divide up the data to study separately the effects of any change in bole shape on the computation of stand tables. 45

Genetic (provenance), environmental (site quality, silvicultural and management practices) and developmental (age and size effects) could be studied by deriving form factors on the partitioned data. The results are in John’s article “Rainbow Eucalypt Man” for eight provenances grown at two sites, 5 different initial spacings, 3 site quality classes and 3 different age classes. (There were differences in bole shape in all the partitions, including age/size.) As a result of the work on bole shape, updated metric volume tables were prepared for E. deglupta. One was for application to pulpwood plantations of Keravat provenance grown on land of site quality 1 or 2 and an initial spacing of from 2.5 x 2.5 m to 4 x 4 m or other arrangements of spacing between rows and between trees that would produce an equivalent stocking, diameters up to 20 cm and heights up to 25 m. The regression of best fit led to the following: V = 0.0005601 + 0.0000804D2 + 0.0000205D2H – 0.0000314H2 + 0.0000111DH2 where: V = total volume in m3, D = mean diameter at breast height over bark in cm, and H = total height in m. For sawlog plantations diameters between 20 cm and 100 cm and heights between 25 m and 85 m, the regression of best fit was the following: V = 0.0000636D1.9037H0.9139 where V, D and H are as above. A measurement plot of not less than about 1/20 ha was recommended. The height H of all trees in a plot could be measured easily using height rods if the stand was not much more than a year old. For older (and thus taller) trees, the measurement of fewer trees in the plot was desirable. Only 10 tallest trees could be measured (200/ha) in stands 18 months or more in age. The mean of these trees was the “predominant height” of the stand (Hdom). A good relationship was found between predominant heights and mean heights at different ages enabling predominant height (Hdom) to be converted into mean height (H) for use in the above volume equation. At 18 months: H = 0.92Hdom – 1.3 (r = 0.94, significant at the 0.1% level) At 30 months: H = 0.97Hdom – 2.03 (r = 0.92, significant at the 0.1% level) At 42 months: H = Hdom – 3.0, at 54 months H = Hdom – 4.1, at 66 months H = Hdom – 5.2 (all r = 0.90, significant at the 0.15 level).

There were A4 page tables produced that enabled the pre-calculated volumes V to be looked up, with D down the left side of the page (rows) and H across the top (columns), which saved the hassle of working out the equation every time.

Personal communication John Davidson and Dick McCarthy 26 th April 2021.

Gary Archer28 described in his thesis in 1977, that forest inventory provides the basis for most forest management decisions. He highlighted the issue that relatively little work has been done on the problems of estimating stand volumes in tropical forests. His thesis, although in reference to the tropical forests of the Northern Territory, reflects his prior experiences in PNG in relation to stand volume estimation. He emphasised that the PNG forest assessments were to give basic estimates of the volume of utilisable wood in the various forest areas. In PNG, tree volume tables related tree volume to independent variables as girth above buttress, log length and form. Resource inventories in PNG are applied to provide information concerning the size of the resource and its commercial log volume content based on sampling at a very low intensity which has rarely exceeded 1% and has often been less than 0.1%. At first the inventories were systematic parallel transects, with better statistical design and accompanying analysis, inventories utilised the random radial line system but used the lines as transects. later to become circular plots of 20 metre radius located every 100 metres along traverses where the bearings are chosen at random and radiating from clearings used as helicopter pads. The traverses are generally 3.2 kilometres long but could be extended. Each plot was treated as a separate sample. The aim was to sample to an intensity of up to 5 %. The mean commercial volume over 50 cm diameter and occasionally the mean volume of trees less than 50 cm was calculated. With the results, suitable correction factors were used to allow for sampling errors etc. Volumes were calculated in forest inventories from a tree volume equation derived from data collected by the Department of Forests. Gary Archer29 recalled that the volume table used for trees >5' girth were greatly assisted by the work of Evan Shield during work for his Dip. For. at Oxford in the 1960s. Gary Archer30 emphasised that the volume table finally used by the Department of Forests was the one shown in the Tiauru Pandi report.

Volume tables for pulpwood trees under 5' girth were developed by David Num from measurements which Gary Archer supervised during the Madang assessment survey in 1970 and measurements by Dave Num from the Cape Rodney Forest assessment survey. 27

Cartoon from Bob Brown’s Grass Roots Guide to PNG Pidgin South Pacific Post. Gary Archer TPNG Forests 1963-1973, MSc Thesis 1977 Tree Volume Models for Tropical Broadleaved Forests with Particular Reference to the Forests of the Northern Territory. 29 Personal communication Gary Archer 27th April 2021 TPNG Forests 1963-1973. 30 Personal Communication 13th June 2021 Gary Archer TPNG Forests 1963-1973. 28

Wood samples were collected at the Madang Forest assessment survey and later the Vanimo assessment survey for a CSIRO investigation of pulpwood properties supervised by V. "Bill" Balodis. Dave Num31 reported that the volume calculation for >5Girth at BH was using an established set of tables in existence when he started surveying. He was certain that it was calculated using Hubers method but whether 10' or 20 ft log length were used he was uncertain. The relationship between volume and total log length was established using linear regression. He understood logs from a couple of hundred trees were used. Evan Shield and Eric Hammermaster were involved. Similarly for 2-5ft girth trees, pulpwood, he established the table using trees from the Madang and Cape Rodney Forest assessment survey. Dave Num32 reported that he worked with Vince Bary a (P)STO with CSIRO Forests Canberra when he was doing his programming training at CCAE in 1971. Vince Bary was the CSIRO statistician and expert programmer who wrote the programs that analysed the resource surveys that PNG NFS used to send down to him for analysis on the CSIRO CDC3200. Dave Num confirmed that he worked on preparing the PNG inventory data for processing at CSIRO. The volume equations were included in the FORTRAN programs. PNG would punch the basic field book onto the cards locally before sending to Canberra. Chris Borough and Dave Num started pre-processing them for errors locally, writing some Fortran programs for the IBM 1130 in the Boroko computer bureau owned by David Lee and managed by Andrew Podger. They were then sent on to Vince Bary for him to shepherd through processing on CSIRO’s Black Mountain CDC3200. The FORTRAN programs NFS were using of course had the equations in them for calculating volumes. Other programs Dave Num wrote for the HP9810a? used the formula for some processing work in PNG in the 70's. The HP 9800 series was introduced in 1972 and discontinued in the late 1970’s. HP Model 9830A calculator. Operating system – ROM basic. 16 kB memory. Photo credit Wilkipedia.

Gary Archer33 reported that he knew Vince Bary who was an STO at CSIRO who wrote FORTRAN programs for processing CSIRO data and fitted volume models to tree measurement data. He and Chris Borough both worked for CSIRO Forest Research, before Chris left to join FORTECH and then later set up his own forest consultancy business. Chris Borough34 reported on the FAO 1989 mensuration report confirming 20" diameter above buttress (48 cm =19.0"). Chris though was never sure what H means as the height to the top of the tree is virtually unmeasurable in normal rainforest. All operations that he saw (including log exports) only harvested logs until the first branch. All PNG’s measurements of log length were from the dab/gab (diameter or girth above buttress) point to 31

Personal Communication Dave Num April 2021 TPNG Forests 1966-1979. Personal communication Dave Num 29th May 2021 TPNG Forests 1966-1979. 33 Personal Communication Gary Archer 29th May 2021 TPNG Forests 1963-1973. 34 Personal communication Chris Borough 30 May 2021 TPNG forests 1960-1971. 32

the first branch and all volume calculations were of log volume - not total volume which in these forests was not a useful measure. Chris emphasized that dab/gab in the case of PNG rainforest assessment was diameter or girth over buttress and log length was estimated from the point where the buttress stopped. Diameter at breast height was not generally applicable within the rainforest situation. Chris Borough35 reported that for natural rainforest log volume tables in PNG they were based on the typical taper of rainforest trees; it was diameter over buttress not diameter at breast height and log length was estimated from the point where the buttress stopped. The measurement of diameter at breast height was not generally applicable to the great majority of rainforest trees measured. Gary does not have any experience on the use of the volume tables listed for the other species (presumably plantation species) in TREE VOLUME TABLES for PNG, page 20-24 i.e., E. robusta, E. grandis, P. patula, P. caribaea, E. deglupta, A. cunninghamii or A. hunsteinii. Gary understood David Num fitted volume equations to the tree measurements he supervised for rainforest trees < 5' girth above buttress during the Madang survey in 1970, but Gary never saw those volume equations before he left PNG in 1973. Gary supervised measurements of montane forest trees while based in Mount Hagen in 1971. Gary fitted height-diameter models of the form: H = a + b.D + c.D2, where H was total tree height rather than merchantable log length, and V = a + b.D2H, where V was total volume including small branch wood. It was not possible to fit any fancier models in Mount Hagen, because they were limited to the use of hand calculators. They did some limited defect studies, but he has no record of these results.

Personal Communication Chris Borough TPNG Forests 1960-1971.

Processing the Resource Assessment Data – Before the invention of modern calculators, people would calculate by using various counting devices. Early human beings may have used their fingers and toes for some problems. As the earliest civilizations grew, merchants may have used stones or seeds to help them with equations.

PNG Tolai Shell money from the Gazelle Peninsular New Britain. Photo credit Dick McCarthy.

The predecessor to the modern abacus was the counting board, which featured groves or lines between which pebbles, or beads were moved. The abacus, on the other hand, featured rods, and could have been made from stone, metal, or wood. Some historians hypothesize that the original abacus was invented in China thousands of years ago, around 500 B.C.E. This device features both an upper and a lower deck with rods and beads and was called a "suanpan." An abacus (plural abaci or abacuses) is a device (also called a counting frame) composed of beads that slide along rods, which fit into a frame. In ancient times, the abacus was used as a calculator; it aided in performing mathematical processes like counting, addition, subtraction, and multiplication. Photo credit Wilkipedia

The abacus was used in the Ancient Near East, Europe, China and Russia, centuries before the adoption of the written Arabic numeral system. The exact origin of the abacus is unknown. The abacus consists of rows of movable beads, or other objects, which represent digits. One of two numbers is set up, and the beads are manipulated to implement an operation involving a second number (e.g., additions) or rarely a square or cubic root. In its earliest use, the rows of beads could be loose on a flat surface or sliding in grooves. Later the beads were made to slide on rods built into a frame, allowing faster manipulation. Abacuses are still made, often on a bamboo frame with beads sliding on wires.

There are distinctive modern implementations of the abacus. The Japanese soroban has been used for practical applications of up to multi-digit numbers.

Today’s Japanese abacus is a 1:4 type, four-bead abacus, introduced from China. It is still manufactured in Japan today. The use of the soroban is still taught in Japanese primary schools as part of mathematics, primarily as an aid to faster mental calculation.

Dick McCarthy’s Unused Soroban. Photo credit Dick McCarthy.

Above 1960’s Publication. To the side. Use of pencils, ruler, rubber, triangle, protractor, tables, was in big demand by TPNG foresters. 52

Other devices used before calculators include the slide rule. This slide rule is positioned to yield several values: From C scale to D scale (multiply by 2), from D scale to C scale (divide by 2), A and B scales (multiply and divide by 4), A and D scales (squares and square roots). Photo credit Wilkipedia.

1963 Reprint.

1962 Publication.

1964 Publication.

1967 Publication.

1968 Publication.

From the early 1900s through the 1960s, mechanical calculators dominated the desktop computing market. These devices were motor-driven and had movable carriages where results of calculations were displayed by dials. Nearly all keyboards were full - each digit that could be entered had its own column of nine keys, 1 thru 9, plus a column-clear key, permitting entry of several digits at once. In these machines, addition and subtraction were performed in a single operation, as on a conventional adding machine, but multiplication and division were accomplished by repeated mechanical additions and subtractions. Bulky calculating machines were essential office tools before the digital revolution. Various desktop mechanical calculators used in the office from 1851 onwards. Each one has a different user interface. This picture shows clockwise from top left: An Arithmometer, a Comptometer, a Dalton adding machine, a Sundstrand, and an Odhner Arithmometer. Photo credit Wilkipedia.

Calculators and computers are relatively new inventions, rendering the use of the abacus a bit obsolete. However, the abacus is still an important learning device, and can still make complicated calculations easier. An office calculating machine with a paper printer. Photo credit Wilkipedia.

Early 1970’s Metrication Publications.

The IBM System/360 (S/360) purchased by Treasury PNG was a family of mainframe computer systems that was announced by IBM on April 7, 1964 and delivered between 1965 and 1978. The slowest System/360 model announced in 1964, the Model 30, could perform up to 34,500 instructions per second, with memory from 8 to 64 KB IBM System 360 Model 30 Released 1964. Memory 8-64 K core. Photo credit Wilkipedia.

Adler 81S pocket calculator with vacuum fluorescent display (VFD) from the mid-1970s. Photo credit Wilkipedia.

The Casio CM-602 Mini electronic calculator provided basic functions in the 1970s. Photo credit Wilkipedia.

The 1972 Sinclair Executive pocket calculator. Photo credit Wilkipedia.

The HP-35, the world's first scientific pocket calculator by Hewlett Packard (1972). (Des Harries36 still has his) Photo credit Wilkipedia.

Des Harries TPNG Forests 1955 - 1976

The IBM 1130 Computing System, purchased by the University of Papua New Guinea in 1968 was IBM’s least expensive computer at that time. A binary 16-bit machine. succeeding the decimal IBM 1620. Typical installations included a 1-megabyte disk drive that stored the operating system, compilers, and object programs, with program source generated and maintained on punched cards. Fortran was the most common programming language used. IBM 1130 with peripherals, including paper tape reader punch, card reader/punch (rear) and Calcomp plotter. IBM 1130 console Photo credit Wilkipedia.

John Davidson reported (21/2/21) on bushies getting ready for computerisation. LOG ON: Adding wood to make the barbie hotter LOG OFF: Not adding any more wood to the barbie. MONITOR: Keeping an eye on the barbie. DOWNLOAD: Getting the firewood off the Ute. HARD DRIVE: Making the trip back home without any cold tinnies. KEYBOARD: Where you hang the Ute keys. WINDOWS: What you open when the weather's hot. SCREEN: What you shut in the mozzie season.. BYTE: What mozzies do MEGABYTE: What Waigani swamp mozzies do. CHIP: A pub snack. MICROCHIP: What's left in the bag after you've eaten the chips. MODEM: What you did to the lawns. LAPTOP: Where the dog/pig sleeps. SOFTWARE: Plastic knives and forks you get at Chinatown. HARDWARE: Stainless steel knives and forks - from Chinatown. MOUSE: The small rat that eats the rice. MAINFRAME: What holds the shed up. WEB: What spiders make. WEBSITE: Usually in the shed or under the verandah. SEARCH ENGINE: What you do when the PMV won't go. CURSOR: What you say when the PMV won't go. YAHOO: What you say when the PMV does go. UPGRADE: A steep hill. SERVER: The person at the pub who brings out the counter lunch. MAIL SERVER: The bloke at the pub who brings out the counter lunch. USER: The one talk who keeps borrowing things. NETWORK: What you do when you need to repair the fishing net. INTERNET: Where you want the fish to go. ONLINE: Where you hang the washing. OFFLINE: Where the washing ends up when the pegs aren't strong enough. 57

“GEN” Neil Brightwell37 recalls many of the concerns are probably wrapped up in the New Britain General 90-10-1 (1) file under Brightwell 1966. If memory serves (and it often does not) the volumetric calculations and area assessments, for the Tiauru-Pandi and Open Bay surveys which were conducted in the field in 1965 together with the KandrianGasmata, Arawe and Kapaluk areas which were conducted in 1966, took so long that they decided they would be best rolled into one general report for the island of New Britain. All these surveys were run by Neil Brightwell together with Ken Granger. This report may also include a note on the Nakanai plateau, visited and sampled in the Tiauru-Pandi assessment. The 90-10-13 Keravat-Vudal report Brightwell 1965, however, does confound him. He has no recollection of either doing or reporting on this area except as may be included in the 9010-1- (1) file noted above. This file may refer to a small survey done out of Keravat in 1963 by Ken Granger independently. In 1963 Neil was based in Keravat and produced a short (five(?) year) working plan for the Keravat plantation and timber stand improvement areas. Neil doubts that such a report (lacking any volumetric data) would have been filed in the 90 series files. On recall Alan White38 may well have placed it in the round file. It would have been impossible to file the report on the Bulolo plantations assessment 1970 in the conventional filing system as it was produced on large size continuous computer paper and was quite bulky. This survey was run by Neil Brightwell using variable sized triangular plots and a Hoop pine tree volume table (ex -Queensland)”, programmed in Fortran by Neil Brightwell and heavily transcribed by Dave Num into a language acceptable to Treasury's new IBM 360 computer. In relation to the Bulolo Plantation assessment 1970, Neil suspects that the field plots were actually quadrilateral, with the areas being calculated using bearings and distances from the corners to calculate and sum two triangles. “With the loss of all those records, it must give you some satisfaction to realise that your magnum opus may shed some light onto yet another dark age.” Quote Neil Brightwell. Neil regrets that he is unable to assist with anything that transpired between Sept 1967 and December 1969, as he was largely tied up with Antarctic activities and some long service leave in Africa, Europe, and southern Asia. Chris Borough39 recalls in 1968-70 the small Forest Management hut at Konedobu was a hive of activity – processing the massive amount of information that had been pouring in from various surveys.

Personal communication 18th April 2021 Neil Brightwell TPNG Forests 1961 – 1984. Alan White TPNG Forests 1956-1977. 39 Personal communication Chris Borough TPNG Forests 1960-1971. 38

Department of Forests Konedobu Photo credit Cliff Southwell.

Each survey book was identified by Assessment, Survey Line, Survey Plot, Forest Type, and Plot Area. This was at a time when electronic calculators were just becoming available. Our collective eyes widened when this new electronic machine appeared that could add, subtract, multiply and divide with ease. Simultaneously the University of Papua New Guinea took delivery of their new computer – the IBM 1130 with the massive capacity of 8Kb of RAM. Located with a range of peripheral equipment in an air-conditioned room, the 1130 opened up the capacity to process rapidly large amounts of data. Chris was sent to the University to learn the programming language - FORTRAN IV. The course was an eye opener as how to add two data elements in a single floating-point number – this enabled the limited programming capacity (8Kb) to handle a large amount of data. As an example, log length and diameter above buttress could be recorded as say 20 (log length in feet) and 24 (diameter in inches). This could be recorded as 20.24. By using the CUT command one number could be generated (in this case 20). Subtracting 20 from 20.24 this leaves 0.24. Multiplying by 100 and you generate the two numbers (20 and 24) from which timber volume could be calculated. By using these cunning techniques, the whole of the PNG Resource data could be processed. The secret to the whole operation was a superb comptometrist (name forgotten) who operated a card punch machine. The data for each survey plot was entered and a punch card generated. This would then be checked by entering the identical data and, if correctly verified the card would be accepted. The next piece of machinery in use was a card sorter where punch cards were sorted such that all the plot data would relate to a specific forest type within a survey area. As the boxes of sorted cards mounted, he had the task of writing the program (on punch cards) that would fit into the 1130. FORTRAN IV was very sensitive to such things as brackets, spaces etc. He could only use the 1130 overnight (cheaper than daytime use) so dropped his program cards and some data cards in at the University on the way home - ready to collect the next morning. It took him many weeks to accurately write the program and have it run. It was then a case of taking a sorted box of plot data cards, commencing with the FORTAN IV program cards, and being delivered a mass of paper output. Having success at last it was back to the card sorter to generate the next box of plot data cards. This was part of the huge effort by the Department of Forests to map and estimate the timber resources of PNG; an astounding outcome that involved so many under difficult conditions. 59

Chris Borough40 reported the IBM 1130 put in place around 1968 had an 8k capacity which was largely taken up by the program. About all that could be done was to add up for each attribute. We had to use every canny trick in the book to allow the analyses to be done. The card sorter had to be used for every run to isolate a specific set of data by location or forest type for example. To determine an estimate of Taun in a mixed forest in a certain area would require a card sort to take out everything but Taun data from that area. I simply do not recall the function used to calculate volume, but I have a gut feeling that it was very simple. It sure was early days in computing whereas previously all the calculations were done on a flash mechanical calculator by a dedicated operator who later switched to preparing punch cards. Chris Borough41 recalls he has some recollection re ERAVE - Pai - A but he did do a small survey out of Kikori and maybe this was the report? All others seem correct. Re Tuolumne Corp. Chris’s interaction was with John from Tuolumne on a visit to Canberra in July 1969 to run a simulation model of the entire forest resource using the CSIRO supercomputer (The IBM1130 with 8k ram at UPNG was a bit short on capacity!). The concept was to use the flash model to optimise the use of the entire PNG forest resource. Chris was very dubious of the practicality of the goals of the project - it seemed to fizzle out after he returned to POM and Tuolumne was departed. Chris thinks Don McIntosh42 must have got cold feet on the project. Chris Borough43 complemented Gary Archer’s approach re - a two-way table to determine form factor. Chris does not recall this being used in the early days and that a single form factor was probably used. Nevertheless, the fact is that a suitable equation existed and could have been used in the early analyses. Gary Archer44 provided an update on form factors. Chris Borough had made the important point of examining form factors, as a possible alternative to complicated volume models. While the rest of us were making guesses of form factors ranging from 0.6 up to 0.9, Chris actually calculated them from the PNGFA two-way (Shield) equation. Gary calculated form factors over a range of DABs and bole lengths using Evan's equation. Results are shown here.

Personal communication 12th June 2021 Chris Brough TPNG Forests 1960-1971. Personal communication Chris Brough 21st April 2021 TPNG Forest 1960-1971. 42 Don McIntosh TPNG Forests 1948-1973. 43 Personal Communication 12th June 2021 Chris Borough TPNG Forests 1960-1971. 44 Personal communication June 2021 Gary Archer TPNG Forests 1963-1973. 41

PNGFA two-way volume equation (Shield equation) - form factors

b0 b1 b2 b3

coefficients 0.185582 0.0000536586 -0.00873586 0.0000517235

DAB(cm) 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80

6 0.885 0.877 0.869 0.863 0.856 0.851 0.846 0.841 0.837 0.834 0.830 0.827 0.824 0.821 0.819 0.817

8 0.818 0.812 0.807 0.803 0.799 0.795 0.792 0.789 0.786 0.784 0.782 0.779 0.778 0.776 0.774 0.773

Length of merchantable bole (m) 10 12 14 16 18 0.777 0.750 0.730 0.716 0.705 0.773 0.747 0.729 0.715 0.704 0.770 0.745 0.727 0.714 0.703 0.767 0.743 0.726 0.713 0.703 0.764 0.741 0.724 0.712 0.702 0.762 0.739 0.723 0.711 0.702 0.759 0.738 0.722 0.711 0.702 0.757 0.736 0.721 0.710 0.701 0.756 0.735 0.721 0.710 0.701 0.754 0.734 0.720 0.709 0.701 0.752 0.733 0.719 0.709 0.701 0.751 0.732 0.718 0.708 0.700 0.750 0.731 0.718 0.708 0.700 0.749 0.730 0.717 0.708 0.700 0.747 0.730 0.717 0.707 0.700 0.746 0.729 0.716 0.707 0.700

20 0.695 0.695 0.695 0.695 0.695 0.695 0.695 0.694 0.694 0.694 0.694 0.694 0.694 0.694 0.694 0.694

Gary notes that the elusive Shield volume equation, converted to metric, in Appendix 7 of the ACIAR inventory_review.pdf. This equation was created in the 1960s from Evan Shield's measurements of mixed rain forest trees, with volume equations fitted and tested on computer at Oxford. In Evan's thesis his equation 14, the Australian equation used here, gave the best fit to the data. In Appendix 7 of inventory_review.pdf the author states: "The Table in Appendix 7 suggests that the PNGFA 'two way' volume function may be providing reasonable estimates of aggregate log parcel provided that inventory crews can reliably estimate merchantable bole length in a standing tree." Gary recommends that use be made of the Shield equation if possible. The above table of form factors illustrates the way it varies with tree size, from 0.9 down to 0.71, being highest for short fat logs just above the buttress and lowest for longer boles extending up to below the tree crowns. The higher form factors (0.85 to 0.9) with short log lengths could be partly an artefact of the data Evan collected. If form factors are calculated from Evan's equation for log lengths of 4 metres or less, they start to go above 1, which is biologically unlikely. Gary suspects two reasons for this: (i) Evan's tree data probably did not include very many short boles, so extrapolating the equation down to short log lengths is risky. (ii) Evan's model includes a negative coefficient for height (-0.0089213), implying that to some extent volume decreases as height increases. 61

Point (ii) is biologically unlikely but would probably have been caused by the shortage of short boles in his data, leading to the least squares fitting process spitting out a (small) spurious negative result. His equation still gives a good fit overall but should obviously be used with caution for short log lengths. Paul Ryan 45 recalls he did not have the volume table or formula included in the Tonolei working plan inventory of 1969. But then he does not think many other inventory reports included the volume table at that time. Paul is sure they used the same table as in the Tiauru Pandi Working Plan as to the best of his knowledge there was only one being used at that time, though there was discussion about how accurate it was for individual species. A paper by Kev White Lowland Rain Forest Regeneration in Papua New Guinea with reference to the Vanimo Sub Province, May 1996, Tropical Forestry Research Note SR.32, makes no mention of volume tables or formulae in Kev’s calculations of volumes. Paul noted in his Tonolei report that they did do some check volume measurements for pulpwood (< 5ft gbh) using the sectional method and from this volume table check it was determined that the Tonolei trees were 14 percent larger than the Cape Rodney trees used to compile the pulp volume table using the Huber Method. Gary Archer46 recalls stumbling on this whilst working in Port Moresby. He believes the original version of Evan's volume equation was fitted to data with girths in inches instead of feet. If you convert his equation to girths in inches, it becomes: V = 80.31549 + 0.01518 G^2 - 1.15235 H + 0.00446 G^2 H As all the coefficients are now expressed to exactly 5 decimal places, which makes Gary think that is how they were printed out from the regression fitting program in Oxford. Then later, when the equation was to be used with girths in feet instead of inches, Evan adjusted coefficients 2 and 4 by multiplying them by 144. This point confirms for Gary that the Tiauru Pandi equation is EXACTLY the same one he was looking at fifty years ago in PNG. Gary Archer47 provided the following update re the volume equation from the Tiauru Pandi Assessment report of 1970. Remember when using the converted metric version of the volume equation, a bark allowance of 2.4 cm should first be deducted from the diameter, The equation, after subtracting the bark allowance of 3 inches (0.25 feet) off girth, gives the same results in super feet. Gary converted the Imperial equation from girth and height in feet and volume in super feet, to its metric equivalent of diameter in centimetres, height in metres and volume in cubic metres. Gary, cross checked the conversion by converting the cubic metre results back to super feet and confirming that they agree exactly with results from the original equation. Original Imperial equation, with girth and height in feet and volume in super feet:

Personal communication 17th June 2021 Paul Ryan TPNG Forests 1962-1972. Personal communication 13 June 2021 Gary Archer TPNG Forests 1963-1973. 47 Personal communication 13 June 2021 Gary Archer TPNG Forests 1963-1973. 46

V = 80.31549 + 2.18592 G^2 - 1.15235 H + 0.64224 G^2 H Converted to metric, with diameter in centimetres, height in metres, and volume in cubic metres: V = 0.185582 + 0.0000536586 D^2 - 0.00873586 H + 0.0000517235 D^2 H The converted equation is not exactly the same as the one given in the Keenan ACIAR Appendix 7 of the earlier document: V = 0.189523 + 0.0000547982 D^2 - 0.0087213 H + 0.0000528219 D^2 H Gary’s conversion checks out exactly, so he does not know why the equation in the ACIAR report page 94 re their Appendix 7 is a little different. It gives slightly different results - up to 1 cubic metre difference in the bottom right corner of the table on page 10. Neil Brightwell 48recalls most of the calculations being done in the 1960's were in the imperial system using inches(?) and feet and resulting in millions of super feet true volume. As Dr Google assures me that Australia did not move to the metric system until 1974, Neil suspects that that the 1970 Bulolo plantation assessment would have also been presented in Imperial form. Bredan Bailey49 recalls purely from memory, two types - hand crank mechanical calculator and motorised (electrical) mechanical calculator. He recalls just old-fashioned pencil and paper calculations. From memory, some of the mechanical calculators had a paper tape/roll. He knew there were some special scientific mechanical calculators around, but they were not my stock and trade. He has painful memories of just filling-out columns of figures with paper and pencil/pen. Brendan recalls vividly at Forest Headquarters Port Moresby, that Dave Num was the first to introduce an electronic computer for calculations. (A side issue - when Brendan left PNG in 1974, he received a parting gift of one of the newfangled battery-powered pocket calculators. He still uses it. His kids think it is cool. It just adds, subtracts, multiplies, and divides. It can do decimal calculations. The screen is quite small so that you have a button to allow the screen to display larger numbers by bringing up the numbers that do not fit on the screen. It is his go-to device. It does not provide a tape printout.) Professor Phil West50 advises that he took the chook shed report and wonder of wonders, the sawlog equation there is the same as the equation on p. 66 in the Brack et al. (2001?) Inventory and Planning Review Report for ACIAR Project FST 98-118 that Gary Archer sent to you and that he concluded (and in his 12 June email) was Shield’s two-way volume function for native species. So, does this mean that Shield has been the only thing that anyone has used for sawlog? From your chook shed report Prof Phil deduced

Personal communication 11th June 2021 Neil Brightwell TPNG Forests. 1961-1984. Personal communication 23 June 2021 Brendan Bailey TPNG Forests. 1969-1974. 50 Personal communication 14th June 2021 Prof Phil West. 49

1. Girth at breast height over bark (gbhob) (G) was defined as girth at breast height or, where a buttress was present, immediately above buttress height. 2. Sawlog volume (V) was estimated for trees with a gbhob of 5 feet or more and a merchantable log length (H) greater than 20 feet. A merchantable log was defined as being the stem section between the logical docking point above the buttress and crown break or some point below crown break that effectively terminated the log. Those points below the crown break could be either the height at which stem diameter over bark fell to 15 inches or where a major fault occurred (twist, bend, broken top). 3. Pulp volume (P) was estimated for trees with a gbhob of 1½−5 feet and a pulp log length (L) greater than 10 feet. A pulp log was defined as being the stem section between the logical docking point above the buttress and crown break or some point below crown break that effectively terminated the log. Those points below the crown break could be either the height at which stem diameter over bark fell to 4 inches or where a major fault occurred (twist, bend, broken top). 4. The equations used to estimate volumes were. V = 80.31449 + 2.18592G2 – 1.15235H + 0.64224G2H

(1)

and P = 17.417 + 2.5495G2 – 0.66954GL + 0.72675G2L ,

(2)

with units of V and P (super feet) and G, H and L (feet). Note that the last term in the equation for P is shown incorrectly in the original document as (GL)2. But volume estimates with that term deviate substantially from those for V, whereas they are acceptably close using G2L. 5. No statement seems to be made as to what volume (if any) was calculated for trees with gbhob big enough for sawlog, but with a merchantable log length less than 20 feet. Were they treated as pulp and, if so, how was the volume calculated? Then, 6. In his thesis, Shield (p 78) defined H as above, but G as girth at the base of the log. This meant it would have had the same value for trees with a buttress, but some value different from girth at breast height, which he called girth at base of, for trees without a buttress (see his p. 84). Following Gary Archer’s interpretation of the Brack et al. (2001?) ACIAR Report, Shield’s two-way volume function was. V = 0.189523 + 0.0000547982D2 − 0.0089213H + 0.0000528219D2H (3) where D was stem diameter rather than girth with units of V (m3), D (cm) and H (m). Converting the units of Equation (1) to metric and girth to diameter, Equations (1) and (3) become the same. 7. If the units of Equation (2) are changed to metric and girths to diameters, it becomes. P = 0.04109954 + 0.00006391D2 – 0.00053427DL + 0.00005977D2L . (4)

This gives volume estimates that are slightly different from those of Equation (3), as would be expected since it is estimating pulpwood, not sawlog, volumes. Professor John Davidson51 offered the following observations re form factor or taper regarding E. deglupta: "It is interesting that for E. deglupta, Lane Poole in 1925 estimated a form factor or taper of 0.5, and Heather in 1955 a form factor of 0.4 for the merchantable bole of large trees of E. deglupta based on limited data. In Indonesia, a form factor of 0.56 is often applied in the absence of any precise information on stand composition." Reference page 55 in Lane-Poole C E 1925 Forest Resources of the Territories of Papua and New Guinea. Government Printer, Melbourne. 209 pp and Page 277 in Heather W A 1955 The Kamarere Forests of New Britain. The Empire Forestry Review, 34:255-278. John referred to his notes sent back by Howard Wright in Oxford to him in Canberra in 1970 after they had run the first 200 E. deglupta trees for volume table creation. And he quotes Howard as follows: "For predominant height (Hdom) the best model was the Shield equation weighted by D2H2dom. V = -6.194 + 0.16Hdom + 0.002823D2Hdom – 0.001151Hdom2 – 0.00000334D2H2dom" This was in imperial units of the time (V = cu ft); D = inches; H = feet; (Pre)Dominant height was defined in PNG at the time as the mean height of the tallest 20 trees per acre. Until now John had never made any connection! Coming just five years after Evan Shield was in Oxford completing his thesis and with Howard Wright involved in his supervision, is “the Shield equation” he refers to, in returning the results to John in 1970 actually the Evan Shield equation? If so, he quotes it as above with added weighting! It is not unlike the form of the equation on p82 of Evan’s thesis. Gary Archer52 referred to John Davidson’s suggestion that some might use simple assumptions such as a form factor of 0.5 to approximate volumes for PNG rain forest species. Earlier Gary had suggested maybe 0.6. Chris Borough53 advised that the shield report confirmed 20" diameter above buttress (48 cm =19.0"). Chris was never sure what H means as the height to the top of the tree is virtually unmeasurable in normal rainforest. All operations that he saw (including log exports) only harvested logs until the first branch. All DOF measurements of log length were from the dab point to the first branch and all volume calculations were of log volume - not total volume which in these forests was not a useful measure. John Davidson54 advised that he had downloaded a copy of Evan Shield’s thesis successfully. Wow! Shades of the old technology - typed on foolscap, tables and formulae handwritten in

Personal communication 7th June 2021 Professor John Davidson TPNG Forests 1962-1980. Personal communication 6th June 2021 Gary Archer TPNG Forests 1963-1973. 53 Personal communication 30th May 2021 Chris Borough TPNG Forests 1960-1971. 54 Personal communication 30th May 2021 John Davidson TPNG Forests 1962-1980. 52

India ink! He noted that Evan prefers to use “girth at the base of the log” rather than "girth above buttress”. John pointed out that it was interesting in Chapter 7, 14 mathematical models for testing volume regressions were already in use at Oxford in 1965, based on the work of Spurr in 1952 and Wright in 1964. Wright applied these same models to John’s Kamarere data in 1970/71. John pointed out that you must go to Appendix 4 (E Shield thesis) to find the volume regression model that Evan fitted to the 461 samples of mixed PNG species. Model 14 was the most satisfactory, followed by 12, 9, 6 and 7 in descending order. Evan also ranked separate models for Anisoptera polyandra (62 samples) and Pometia spp (211 samples). However, the models presented in the thesis are without regression constants and regression coefficients, so one cannot (re)construct the actual volume table(s). Presumably, this was done separately, to produce the first volume table used in PNG?

Methodology re National TPNG Helicopter Forest Resource Assessment55 Up until 1963, forest surveys were ground based but to complete the comprehensive coverage of PNG’s forest resources to meet the Commonwealth Government’s timeframe, helicopter transport was introduced in 1964 to take over most of the support tasks and to position field parties on site. Over 3.6 million hectares were assessed in the six years 1964 to 1969. This was documented by Don McIntosh and Eric Hammermaster in 1968. Major surveys then undertaken included timber areas as Vanimo, Empress Augusta Bay, Open Bay, Wanigela, West New Britain, Middle Ramu, Bulolo Wau, Ioma, Sagarai Gadaisu, Cape Rodney, Tonolei Harbour etc. From 1963 - 1969 approximately 3,600,000 ha were assessed when helicopter techniques of inventory were introduced.

56Gogol

Survey Boss Boys 1963 L to R Norm Endacott; Johnny Lowien; David “dokka” Reid, Eric Hammermaster, Jim Cavanaugh57, Evan Shield, Peter Eddowes, Pilot Hurrell, Don McIntosh, Bill Jenkin, unknown, unknown, Kevin White58, unknown. Photo credit Mary Jenkin.

McIntosh D H and Hammermaster E T 1968 Forest Resource Assessments using Helicopter Transport in the Territory of Papua and New Guinea. Paper 9th Commonwealth Forestry Conference 1968 56 Personal Communication Eric Hammermaster 11/3/21.TPNG Forests 1956-1979. Gogol trial survey Nov to Dec 1963. Based at Utu Catholic Mission Station. Unknowns would include helicopter mechanic of Helicopter Utilities, soils staff of DASF as Paul Aland and DIES. One of the PNG forestry assistants was Gallope Mato who trained under Jim Cavanaugh pos- war. 57 Jim Cavanaugh TPNG Forests 1938-1972. 58 Kevin White TPNG Forests 1957-1977.

Photo credit Professor John Davidson. TPNG Forests 1962-1980. 72

An Example of Data Compilation Tiauru Pandi Forest Assessment 1970

Tiauru Pandi and Open Bay Timber Areas 1970 TPNG. Dick McCarthy, Bob McKeowen, and Francis Tigi, Tiauru Pandi Resource Survey1970. Photo Credit Dept of Forests PNG 1970. (Isaac Passingan in field.)

Tiauru Pandi Land Use Plan New Britain by DASF and Dept of Forests 1970

Mount Ulawun Volcano. Photo credit Prof John Davidson. Map by Bob McKeowen Senior Draftsman DOF Tiauru Pandi Vegetation Types 1970

Bibliography (Includes relevant documents tracing PNG’s Forest Mensuration History over Time). ACIAR 2000. Project FST/1995/123 Testing the utility of the North Queensland rainforest growth and timber yield model in PNG. ACIAR 2001. Project FST/1998/118 Review Report. Planning methods for sustainable management of timber stocks in Papua New Guinea’s Forests. ACIAR 2002. Project FST/1998/118 Annual Report 2001-02. Planning methods for sustainable management of timber stocks in Papua New Guinea’s Forests. ACIAR 2005. Project FST/1998/118 Final Report. Planning methods for sustainable management of timber stocks in Papua New Guinea’s Forests. ACIAR 2005. Project FST98-118 Improved timber inventory and strategic forest planning in Papua New Guinea. Authors include R Keenan, Chris Brack, Ian Franks, Adam Gerrand. ACIAR 2011. Native Forest management in Papua New Guinea: advances in assessment, modelling and decision-making. Editors Julian Fox, J Keenan, Chris Brack and Simon Saulei. ACIAR 2021. Project FST 98-118 - Planning methods for sustainable management of timber stocks in PNG forests Review of forest inventory and mapping systems for forest planning report based on trip to PNG from 19-29 November 2001 C. Brack, C. Bragg, I. Frakes, A. Gerrand, R. Keenan, P. Tickle, J. Vanclay. Alder D 1977 A growth and management model for coniferous plantations in East Africa. PhD thesis Oxford University. Alder D 1992. Simple methods for calculating minimum diameter and sustainable yield in mixed tropical forest. Proceedings of the Oxford Conference on Tropical Forests Wise Management of Tropical Forests. Oxford Forestry Institute. Alder D and Synott TC 1992. Permanent Sample Plot techniques for mixed tropical forests. Tropical Forestry Paper No 25 Oxford Forestry Institute. Alder, D. 1998 PINFORM: a growth model for lowland tropical forest in Papua New Guinea Unpublished Report to ITTO, Project PD 162/91. Alder, D., Oavika, F. and Yosi, C. 1998 Data, programs and models for natural forest growth and yield Unpublished Report to ITTO, Project PD 162/91. Archer, G.R. 1972. A simple tree height converter [for measuring tree heights in steep topography with poor visibility]. Commonwealth Forestry Review 51(3): 246-253. Archer G R 1977 Tree Volume Models for Tropical Broadleaved Forests with particular reference to the forests of the Northern Territory. Thesis submitted for the degree of Master of Science at the Australian National University 1977. Avery T E and Burkhart H E FOREST MEASUREMENTS Fourth Edition. McGraw-Hill Series in Forest Resources. ISBN 0-07-113204-X. Barraclough H 1963 ABRIDGED MATHEMATICAL TABLES Parts 1 and 2. Sydney University. Publisher Angus and Robertson P/l. 85

British Standard Code of Practice CP 112: Part 2: 1971 The Structural Use of Timber Part 2. Metric Units the Council for Codes of Practice British Standards Institution Gr 8. Brack CL 2011 Improving Forest inventory in Papua New Guinea: moving away from the 1% strip line survey. In report by ACIAR 2011. Native Forest management in Papua New Guinea: advances in assessment, modelling and decision-making. Editors Julian Fox, J Keenan, Chris Brack and Simon Saulei. Carron L T 1968 An Outline of Forest Mensuration with Special Reference to Australia. ANU Press. Aus. 68-2022 ODC 5. Department of Agriculture Forestry and Timber Bureau Australia 1975 TABLES FOR FORESTERS ISBN 0 642 01779 4. Department of Forests TPNG 1966 Forest Surveying Vol 2 Papua and New Guinea Forestry School Manual prepared by Arthur Ramsay. Department of Forests PNG Manual 1969 Forest Assessment Surveys Cartography Instruction Manual No 2 prepared by Bob McKeowen. Department of Forests PNG Forestry College Training Manual Volume # 7 FOREST ASSESSMENT and WORKING PLANS June 1973 compiled by Alan White, Dick McCarthy, Gary Archer and Dave Num. Department of Forests PNG 1969 Publication. Turi and the Trees. Dept of Forests PNG 1970 TIAURU PANDI TIMBER AREA WNB LOGGING PLAN 19701975 prepared by Forest Officer R B McCarthy. Department of National Development Forestry and Timber Bureau Australia 1964 FORESTRY TABLES Part 1 Section 1 – Mensuration. Department of National Development Forestry and Timber Bureau Australia 1964 FORESTRY TABLES Part 2 Section 2 – Management, Section 3 Utilisation, Section 4 Silviculture, Section 5 Meteorology. Department of Primary Industry Forestry and Timber Bureau Australia 1973 METRIC LOG VOLUME TABLES ISBN0 642 00381 5. FAO. 1948. Forest Resources of the World. Unasylva, 2(4). FAO. 1968. World Forest Inventories. Unasylva, 22(3): 2. FAO 1980 1980 D Forest Volume Estimation and Yield Prediction Forestry Paper 22/2 vol 2 prepared by D Alder Rome. FAO 1987. Continuous forest inventory in lowland and lower montane forests. Working Document No.5 project PNG/84/003, prepared by B Kingston Dept of Forests PNG.

FAO 1988. Report on Diagnostic Sampling conducted in Oomsis Forest Morobe Province. Working Document No.5 project PNG/84/003, prepared by B Kingston Dept of Forests PNG. FAO 1988. Master Plan for Timber Stand Improvement in Lowland and Lower Montane Forest. Working Document No.10 project PNG/84/003, prepared by B Kingston Dept of Forests PNG. FAO 1989. Forest Management Research Plans for the Natural Forests of Papua New Guinea. Working Document No.4 project PNG/84/003, prepared by B Kingston Dept of Forests PNG. FAO 1989. Compilation of Tree Volume Tables for the Lowland Forests. Working Document, No 13 project PNG/84/003, prepared by B Kingston Dept of Forests PNG. FAO 1988. Growth and Yield of Mixed Tropical Forests. Unpublished consultancy report to FAO by D Alder Rome. FAO. 2001. Global Forest Resources Assessment 2000 - main report. FAO Forestry Paper No. 140. Rome. Available on the Internet: www.fao.org/forestry/fo/fra/main/index.jsp. Finlayson, W., and Archer, G.R. 1964. A modified Biltmore stick for use on buttressed trees in tropical rain forest. Letter to the Editor, Commonwealth Forestry Review 43(4): 285. Freyne D 1997 unpublished report Resource Inventory: Developing the management database Team leader of the Land Mobilisation Project (ACLMP59) 1997, presentation Foresters’ Refresher School FRI Lae 1997, as part of the PNG Forestry Human Resource Development Project funded by AusAid. Furnival G W 1961 An Index for Comparing Equations used in Constructing Volume Tables. For. Sci. Vol 7 No 4 pp 337-341. Gael Keig, Robin L. Hide, Susan M. Cuddy, Heinz Buettikofer, Jennifer A. Bellamy, Pieter Bleeker, David Freyne and John McAlpine: CSIRO and land research in Papua New Guinea 1950–2000: part 1: pre-Independence, CSIRO Publishing Historical Records of Australian Science https://doi.org/10.1071/HR18019. Hammermaster E T and Saunders J C Forest Resources and Vegetation Mapping of Papua New Guinea 1995 PNGRIS Publication No. 4 ISBN 064219605 2. ITTO/PNG project PD 162/91 Intensification of growth and yield studies in previously logged forest. https://www.itto.int/project/id/PD162_91-Rev.1-F. Johns R J 1977 The vegetation of Papua New Guinea Training Manual for the Forestry College Vol 10 Bulolo December 1977. Reprinted 1984. Keenan RJ, Ambia V, Brack C, Gerrand A, Golman M et al 2005. Improved timber inventory and strategic forest planning in Papua New Guinea. Bureau of Rural Sciences Canberra and FRI Lae.

ACLMP Australian contribution to the World Bank funded Land Mobilization Program.

Lawrence P R 1960 A Rational Approach to Tree Volume Measurement. Australian Forestry Vol XX1V. No 2. Pp 107-111. Luke R H 1946 An estimate of the Timber Resources of New Britain. Australian Forestry, 10:1, 72-80, DO1: 10:1080/00049158.1946.10675246. MATHESON’S 1974 Australian Saw-Millers Log and Timber ready Reckoner. METRIC Cole Publications ISBN 0 909900 051. McAlpine J and Quigley J Sept 1998 Forest Resources of Papua New Guinea. Summary Statistics from the Forest Inventory Mapping System (FIM) Coffey for AusAid/CSIRO. McIntosh D H and Hammermaster E T 1968 Forest Resource Assessments using Helicopter Transport in the Territory of Papua and New Guinea. Paper 9th Commonwealth Forestry Conference 1968. Metric Conversion Board Australian Government 1970 METRIC FARMING. Minister for Forests 1973 New Horizons - Forestry in Papua New Guinea – Brisbane. Paijmans, K. 1975 Vegetation Map of Papua New Guinea and Explanatory Notes to the Vegetation Map of Papua New Guinea - Land Research Series No. 35 – Melbourne. Papua New Guinea Forest Authority 1996 NATIONAL FOREST PLAN FOR PAPUA NEW GUINEA. PNGFA/JICA 2019 Papua New Guinea Forest Resource Information Management System (PNG-FRIMS) ISBN 978-9980-908-76-6. Peki M 2001 Stand Structure and Growth of Logged over Natural Forest in Papua New Guinea. Master of Agricultural Science Thesis Faculty of Agriculture. The University of Tokyo 145 pp. Peki M 2004 The Growth Analysis and its Application for Management Selective Cutting Natural Forest in Papua New Guinea. PhD thesis. Graduate School of Agriculture and Life Sciences Faculty of Agriculture. The University of Tokyo 249 pp. Peki M 2021 Progress on the studies of growth of logged over natural forests in Papua New Guinea. Phillips F H (CSIRO) and Harries E D (FPRC DOF PNG) 1975 The Pulping and Papermaking Potential of Tropical Hardwoods 1. Phillips F H (CSIRO), Logan A F (CSIRO), Balodis V (CSIRO) 1975 The Pulping and Papermaking Potential of Tropical Hardwoods 11. PNGRIS No 6 Papua New Guinea Inventory of Natural Resources. Population Distribution and Land use handbook. Prepared by Bellamy J A and McAlpine J R (1995) ISBN 0642196109. PNGRIS No 2 Forest Resources of Papua New Guinea ISBN 0642196036.

Poesi M, Piafu M, Abe H, Turia R editors 2008. Papua New Guinea’s first multi-purpose National Forest Inventory Project Proceedings of a Research Conference held in Lae 14-15 February 2018. NFI Proceedings 188p. Quigley J 1998 FIM: Forest Inventory and Mapping System, Version 2.1. User Guide. October 1998 91 pp. Romijn K 1994. PSP standards and procedures: A permanent sample plot program to predict growth and yield in previously logged forests. Part A-E. ITTO Project PD 162/91. PNG Forest Research Institute Lae PNG 219 pp. Shearman PL, Bryan JE, Ash J, Hunnam P, Mackey B, Lokes B 2008. The state of forests of Papua New Guinea. Mapping the extent and condition of forest cover and measuring the drivers of forest change in the period 1972-2002. University of Papua New Guinea 2009. ISBN 9980-937-48-3 Shield E H 1965, The application of new sampling methods to previously inaccessible tropical forest areas, with particular reference to Papua New Guinea. Commonwealth Forestry Review Vol 55 No 1 March 1976. Shillinglaw A W the Military Survey of the Forest Resources of the Commonwealth Territories, Australian Forestry, 9:2, 57-59, DO1:10.1080/00049158.1945.10675232. US Department of Agriculture 1962 ELEMENTARY FOREST SAMPLING Agricultural Handbook 317 prepared by F Freese Statistician. US Department of Agriculture 1967 ELEMENTARY STATISTICAL METHODS FOR FORESTERS Agricultural Handbook 232 prepared by F Freese Statistician. Wallace W G 1966 Cutting Pads in the Forest for Bell 47G-3B-1 Helicopter Operations. Government Printer Port Moresby 1966. West P W 2014 Tree and Forest Measurement Third Edition. Springer ISBN 978-3-31914707-9. WIGG & SON P/L 1966 THE WIGG TABLE BOOK for use throughout Primary Education ISBN 13: 9780959902419. Yosi C and Ambia V 2001 A Review of Sampling Methods for Forest Resource Inventories in Papua New Guinea. Paper to seminar on status and progress of natural forest management research activities in Papua New Guines PNG FRI Lae 28th June 2001 17 pp.

ACRONYMS ACT ACIAR ACLMP AEC AFS AFPNG AIF AMF ANBG ANGAU ANU APMF APPM ASIO ASOPA AusAID BA BCOF “Beer Time” BFC BGD BUC C cm CALM CFA CNGT CRE CRE

CSIRO CHAH DASF DBH/ dbh DEPT DPI DOF e.g. Etc FAO F &TB FIM FPRC

Australian Capital Territory Australian Centre for International Agricultural Research AusAid funded World Bank Land Mobilisation program. Administrators Executive Committee Australian Forestry School Association of Foresters of PNG Australian Infantry Forces Australian Military Forces Australian National Botanical Gardens Australian New Guinea Administrative Unit Australian National University Australian Paper Manufacturers Forestry Pty Ltd Australia Paper and Pulp Manufacturers Australian Security Intelligence Organisation Australian School of Pacific Administration Australian Aid Agency basal area British Commonwealth Occupational Force 1945-52 Any time. Bulolo Forestry College Bulolo Gold Dredging Company Bulolo University College Commonwealth Centimetre Western Australian Department of Conservation and Land Management Commonwealth Forestry Association Commonwealth New Guinea Timbers Bulolo Commander Royal Engineers CRE is a term inherited by RAE from RE and is the term for the Commanding Officer of a RAE unit which is headed by a Lt Col. Although the officer is called the CRE the name is also used for the name of his unit. E.g., CRE Aust Forestry Group or 1(NG Forests). Commonwealth Scientific and Industrial Research Organisation Council of Heads of Australasian Herbaria Dept of Agriculture, Stock and Fisheries Diameter at breast height Department Department of Primary Industry Department of Forests For example et cetera (more of the same) Food and Agriculture Organisation Forest and Timber Bureau Canberra Forest Information System Forest Products Research Centre Hohola 90

FRI Forkol GAB Gbhob Gubab GIS ha IBRD IFA ITTO JICA L of N LRRS m3 MAG MM NAA NARI NB NFCAP no. NG NGF NGIB NGO NGVR NZ NSW NTSC OISCA

Forest Research Institute Lae Bulolo Forestry College Girth above buttress Girth breast height over bark Girth under bark above buttress Geographic Information Systems Hectare International Bank for Reconstruction and Development Institute of Foresters of Australia International Tropical Timber Organization Japanese International Cooperation Agency League of Nations Land Resource Soils Survey (branch of CSIRO) cubic metre magazine Military Medal National Archives Australia National Agriculture Research Institute New Britain PNG National Forestry and Conservation Action Plan Number New Guinea New Guinea Forces (relates to plant collection of Lae Herbarium) New Guinea Infantry Battalion Non-Government Organisation New Guinea Volunteer Rifles New Zealand New South Wales National Tree Seed Centre PNG Bulolo Organisation for Industrial, Spiritual and Cultural Advancement International Japan. P or p or pp page PIB Papuan Infantry Battalion PIR Pacific Islands Regiment PNG Papua New Guinea PNGAA Papua New Guinea Australia Association PNGAF Papua New Guinea Australian Foresters Magazine Series PNGFA Papua New Guinea Forest Authority PNGFIA PNG Forest Industries Association PNGRIS Papua New Guinea Resource Information System PNGUT PNG University of Technology POM Port Moresby Q Queensland QF Queensland Forestry RAE Royal Australian Engineers/Australian Army RPC Royal Papuan Constabulary RRA Rapid Resource Appraisal SFM Sustainable Forest Management SP South Pacific UK United Kingdom 91

UN Unasylva UNE UNEP UNI UNITECH UNRE UPNG UQ US USA TPNG TUBL TA TA TRP Vol VSF WA WB WW2

United Nations Journal of FAO of UN University of New England Armidale NSW United Nations Environment Program University University of Technology Lae PNG University of Natural Resources and Environment University of Papua New Guinea University of Queensland United States United States of America Territory of Papua and New Guinea Territory United Brewery Ltd Timber Area Timber Authority Timber Rights Purchase volume Victorian School of Forestry Western Australia World Bank WORLD WAR 2