I.1What is ForestPro?

It is a program for forest inventory, mensuration, biomass, carbon and agroforestry that runs in the browser, with nothing to install. It takes a forestry job from the field sheet to the report: measure the trees, sample the stand, expand to the hectare, convert into biomass and carbon, project growth and report.

It is meant for those who collect and analyze inventories —forest technicians, thesis students, people in charge of management plans or carbon projects— and for coffee farms and other agroforestry systems, which seldom fit tools made for the forest. It scales trees and logs and fits volume, taper and height equations; expands the inventory with its sampling error under four designs; estimates biomass and carbon with a catalog of 31 allometric equations, wood density and its uncertainty by Monte Carlo; fits growth and site index curves and projects the stand; measures diversity, similarity among sites and the spatial pattern of the trees; characterizes the shade of a coffee farm by counting trees or from a canopy photograph; puts the plots on a map, with sampling grids and interpolation; and does the sums of the sawmill and the plantation.

Everything is computed with an engine written from scratch for this program, with no libraries by others, and every result comes with its method, its sources and an interpretation in words.

I.2Questions it answers

A good way to find out whether the program suits you is to see which questions it answers, and in which block.

I.3Six ideas behind the manual

Everything that follows rests on six ideas. If you keep them in mind, many of the mistakes made when processing an inventory are no longer possible.

1. One measured tree stands for many. Each tree in a plot of a square meters is worth 10,000/a trees per hectare; in a nested plot, those of the subplot its diameter assigns it to, and in angle-count sampling, BAF/g. Getting a plot’s area wrong makes everything that follows wrong. That is why Block 3 asks for the design of each plot and shows it on screen.
2. A mean without its error is not an estimate. The inventory is a sample of the property: the mean per hectare is worth what its sampling error is worth, and that error depends on the design with which the plots were taken —random, stratified, in a systematic grid or in clusters—. The app gives the error under each design and says which one applies.
3. An equation is only valid where it was fitted. Volume, biomass and height equations were fitted with trees of a certain size, species and region; outside that range they extrapolate. The app warns when a tree falls outside the diameter range of its equation, and lets you add the local equation that exists for your species.
4. Biomass has three sources of uncertainty, and sampling is only one of them. The other two are measurement error (diameter, height, wood density) and the error of the equation itself. Block 4 propagates them by Monte Carlo and adds them to the sampling error.
5. Carbon sits in several pools. Living trees are the most visible one, but not the only one: roots, deadwood, litter and soil count, and on a coffee farm the soil usually holds more than the shade trees. The app brings them together with their combined error.
6. What is not reported cannot be checked. The design and area of the plots, the equation and density used, the growth model, the confidence level of the interval and the program version belong in the methods. Every result comes with its Method and sources box, and the Block 10 report can carry them all.

I.4Data it accepts and what it delivers

The starting point is almost always a table: one row per tree, plot, section, point or log. Every section that needs data has a data card that reads it from an example, a file, a pasted or hand-typed table, or the inventory already loaded in Block 3, and then matches its columns to the fields the method asks for (Section I.6).

The blocks also pass data to other apps of the suite: the similarity matrix of Block 6 is sent to ClusteringPro and PCAPro, and the soil carbon of Block 4 reads SoilPro’s layer tables.

I.5What you need and how to open it

Just a modern browser, in a version from the last few years, on a computer, tablet or phone. Nothing needs to be installed and no internet is required; the whole program fits in one folder.

1. Double-click index.html. It opens in the default browser, on the home page. It is the simplest way and works just like everything else: the computations and the example data sets are inside the program.
2. Or double-click Open ForestPro.bat. If your computer does not allow pages to be opened as local files, it starts a minimal server on your own computer (you can also run server.ps1 with PowerShell) and opens the app at http://localhost:9480.

Preferences —the language, the light or dark theme, the animations— are stored in the browser itself and remembered the next time. Data are not: every time you open the app it starts empty, and to resume a job there is the .json project of Block 10. If you try to close the tab with data loaded or unsaved results, the app warns you.

I.6A tour of the interface

All the apps of the suite share the same bar, the same navigation and the same figure tools, so what you learn here serves you in the others.

Finding your way around

On the left is the bar of the ten blocks; on the right, the In this block list, with the sections of the open block. All the blocks are open from the start and each one can be used on its own. At the foot of each block two buttons lead to the previous and the next one. When you change a control and the results are recomputed further down, a floating button appears that takes you back to the control you touched, and another one to go back to the top. Almost everything can also be done from the keyboard:

1 2 3 4 5 6 7 8 9 10 11 12

• 1Name of the app and its tagline. The name leads to the home page (Block 1).
• 2LABG Suite. Back to the portal with the other apps.
• 3ES · EN. Switches the language of the whole interface, including the figures and the report.
• 4Theme, animations, report and shortcuts. Light or dark; animated waits on or off; reporting a problem on the project’s public page; the list of keyboard shortcuts (key ?).
• 5The bar of the ten blocks. The number turns into a check mark when the block has a result, and the count at the top says how many do (1/10 in the figure); [ collapses it.
• 6The path. The app, the block and the section you are in.
• 7Document · Focused. All the sections of the block one after another, or a single one at a time; the button beside it collapses or expands all the sections.
• 8Search (Ctrl + K). A block, a section, a control or an action, by its name.
• 9In this block. The sections of the open block; the current one is highlighted.
• 10The block title, with its number and what it does in one line, and a short presentation below.
• 11A card. Each section is a card with its data and controls at the top and its results below.
• 12Search, Focused mode and Help, also at the foot of the block bar.

The data card

Almost every section that needs data receives it through the same piece, the data card, so you learn it once. From left to right: the list of examples for that section with Load example (a note below says what the chosen example contains); Open file…; Paste table; Type by hand, which opens an empty table with the columns the section asks for; Use the loaded inventory, in the sections that can work with the Block 3 inventory; and ⤓ Template, a CSV with the column names and three example rows.

After reading a file or a pasted table, the app matches the columns: those it recognizes by their name, in Spanish or English and with several synonyms, come already chosen; the others are chosen from a list, and the required ones carry an asterisk. Use these data loads them. Rows missing a required value are left out with a warning, and See the data shows what was loaded.

What every result brings

The results of every section have the same shape: a few tiles with the numbers that matter; the figures and tables, with their ⤓ CSV and ⤓ XLSX buttons; an interpretation in words, written with the numbers of your data; the warnings, when something deserves attention; and, at the end, the collapsible Method and sources box, with the formulas and the publications they come from.

The figures

Every figure in the app can be opened in the figure studio, a window shared by the suite where the texts, the font size, the legend and the colors are changed without touching the data, and which exports to the widths and resolutions of journals. Next to each figure, the ✎ button opens that figure’s editor, to change its titles, axes, series, legend and annotations. Figures are always exported with the colors of the light theme, even when you are working in the dark one. Chapter 10 explains it in depth.

I.7The four example data sets

To learn without data of your own, the app brings four data sets. Three are simulated with a fixed seed and realistic orders of magnitude: the species are real, but the trees, plots and properties do not exist, and they serve to explore the app and for the tests, not to describe any site. The fourth gathers reference tables. They open with the Open in … buttons of Block 1›Example data, which open the block and load the example, or from the list of examples of each data card.

How each data set was simulated —the models, the parameters, the seed— is written in data/LEEME.md, next to the script that generates them; the tables are also in data/ as CSV, with column names the app recognizes without matching. The text says “the example coffee farm”, not “the coffee farm”.

I.8How a forestry job is worked through

The blocks follow the order of a real forestry job. You need not go through all of them: those that do not apply are skipped, and each module can also be used on its own, with its own table. The How it works section of the home page sums it up in six steps.

1. Measure (Blocks 2 and 9). The diameters and heights of standing trees; the volume of felled trees and logs; the volume, taper and height equations that will serve the rest.
2. Sample and expand (Block 3). Load the trees and the plots with their design; the app expands them to the hectare and estimates the mean, its sampling error and the total of the property, with the stand structure and the sample size for next time.
3. Convert (Block 4). Biomass tree by tree, roots, deadwood, litter and soil; the carbon and CO₂e of each pool with its uncertainty, and the change between two inventories.
4. Project and describe (Blocks 5, 6 and 7). Growth, rotation and site quality; diversity and spatial pattern; the shade, yield and carbon of an agroforestry system.
5. Locate (Block 8). The plots on the map, the area of the property, the grid for a new inventory and the interpolation of a variable among plots.
6. Report (Block 10). The report with the figures, the tables and the methods; all the tables in one workbook; the project to resume the work.

I.9How it was checked

A computing program is worth what its numbers are worth. Every ForestPro computing engine runs the same in the browser as outside it, so each block was compared, with the example data sets, against independent reference implementations written by others in an open-source statistical computing language, and against formulas written separately where no implementation exists. Each comparison has its tolerance.

The comparisons also served to find errors and differences of convention, which were documented and, the errors, corrected before publishing. For example, the first version of the Block 4 Monte Carlo corrected the log-scale bias twice and gave a mean biomass 6.6% higher than the direct computation: the coefficient 0.0673 of Chave et al. (2014) already includes that correction. Or that in the IPCC (2006) root-to-shoot table the “50–150” class includes 150. The chapters say, where it matters, which convention the app follows and why. The details, block by block, are in the program’s tests/ folder.

I.10What the program does not do

Knowing where a tool ends is as useful as knowing what it does. The app’s home page says it in What it does and what it does not:

1.1The home page and its buttons

When you open index.html, the app starts in Block 1. Unlike the other blocks, it has no numbered title and no cards: it is a presentation page that you read from top to bottom, and the In this block list on the right takes you to each of its sections.

At the top is the hero, a dark panel with the motto From the measured tree to the metric ton of carbon, a line with the four subjects of the app (Forest inventory · mensuration · biomass and carbon · agroforestry) and a paragraph that sums up what it does. On the right is an original drawing: the end of a log with its growth rings, labeled DBH 38.4 cm · 43 rings · 1.30 m, whose outer rings leave the trunk and become the contour lines of a map with inventory plots. The caption of the drawing says it in one line: The rings of a trunk and the contour lines of the stand: what ForestPro measures.

1 2 3 4 5

• 1Start with an example → Opens Block 3 and loads the coffee farm example into it, trees and plots: the quickest way to see the app with data.
• 2See the modules. Scrolls down to The nine modules, on the same page.
• 3How to cite. Scrolls down to the How to cite section, at the end of the page.
• 4The four badges. Runs entirely in your browser, Checked against reference implementations in R, Every method with its source and Spanish and English.
• 5The end of the log. A drawing, not a result: it does not respond to the data you load.

The Start with an example → button does the same as the Open in Inventory button on the coffee farm card in Example data (Section 1.4). The other two only scroll the page; if you asked your system settings to reduce motion, the jump is immediate instead of smooth.

1.2How it works: the six steps

The first section below the hero sums up the order of a forestry job in six boxes, each with its step number, its name and what goes into it. The subtitle explains it: The order is that of a forestry job: measure the trees, sample the stand, expand to the hectare, convert into biomass and carbon, project and report. Figure I.7 of the introduction shows this section.

The second half of the subtitle matters just as much: The modules also work on their own: each one takes its own table. Nothing forces you to go through Block 3 to scale logs in Block 9 or to fit a growth curve in Block 5. The order helps you not to forget steps; it does not impose them. Section I.8 of the introduction walks through it in more detail.

1.3The nine modules

The section The nine modules has one card for each block from 2 to 10, with a vignette drawn for that block, its number, its name and the same line that the app shows below the title of the block. Clicking anywhere on the card opens the block. As its subtitle says, A module is a block of the bar and a chapter of the manual. The card of Block 2 corresponds to Chapter 2 of this manual, that of Block 3 to Chapter 3, and so on up to 10.

The vignettes are also original drawings made by code, like the end of the log in the hero, and they follow the language of the interface. The short names of the cards are the same as in the block bar (Inventory and sampling, Maps, …); the full title of each block is the one its chapter carries in this manual (Forest inventory and sampling, Maps and light geoprocessing, …).

1.4Example data

The Example data section has four cards, one per dataset, with its name, a line of figures, a description and one Open in … button for each block where it is used (Figure I.6 of the introduction). Its subtitle warns that they are simulated, with realistic orders of magnitude, that they are not measurements of a real site and that the script that generates them is in data/generar_datos.R.

Each button does two things: it opens the block and loads the example into its data cards, replacing whatever was there. Each data card also has its own list of examples, so you can load, for instance, the cloud forest trees into Block 2›Standing trees even though the home page has no button for that. The datasets, what they contain and what each is for are described in Section I.7.

1.5What it does and what it does not

Four boxes say up front what you can expect from the program. The first is a promise; the other three, limits.

• It computes with your data, on your computer. Nothing is sent to any server. The only exception is voluntary: the online map background of Block 8, off by default, which when switched on requests only the map tiles.
• It does not replace the local equation. The biomass equations of the Block 4 catalog are references; if there is one for your species and region, add it and use it.
• It is not a geographic information system. Block 8 computes areas, grids and interpolations and exports GPX, KML and GeoJSON; full spatial analysis is done with a mapping program.
• It does not certify carbon. The carbon figures with their uncertainty serve studies and management plans; a carbon-market project also follows the rules of its standard.

Section I.10 of the introduction adds two limits that are not on the home page: the program does not plan the sampling for you and does not work with tree-level models or remote sensing.

1.6How to cite, the tests and the validation

The last section gives the citation of the program in two formats, one below the other: the APA-style reference and the BibTeX entry (with the key BarreraGuzman2026ForestPro, version 1.0.0, the license GPL-3.0-or-later and the DOI). The Copy APA and Copy BibTeX buttons copy it to the clipboard, and a brief notice confirms Copied. The text above it recalls what must also be cited: the equations and methods you used, which every result names, in its Method and sources box.

The same citation is in Block 10›How to cite, which also downloads the file forestpro.bib, and at the end of every report. The citation carries the Zenodo concept DOI, 10.5281/zenodo.23876374, which always leads to the latest version; the code is at https://github.com/luisangelbg/ForestPro.

The footer of the home page

Below the citation there are three links, which open in a new tab:

A last line says that ForestPro is free software (GPL-3.0) of the suite and that it is built with its own code, with no third-party libraries.

The block introduces itself like this: Five tools used together or on their own: the first measures standing trees; the second scales felled trees; with those volumes the third fits volume equations and the fourth, taper curves; the fifth estimates the missing heights. Cards 2.2, 2.3 and 2.4 form a chain —the sections of the felled trees feed the equations and the taper—; the other two work on their own.

The example of this chapter is the pine plantation: Block 1›Example data›Open in Mensuration loads the 40 felled trees with their 536 sections into the scaling card. Section 2.1 also uses the shade trees of the coffee farm example and Section 2.5 the pine plot with missing heights, which are loaded from the example list of their own data card (Section I.6).

2.1Standing trees: diameter, basal area and volume

The first card has two parts: a calculator for a single tree and a table of trees that is summarized by species.

At the top, four fields. Form factor (0.5 by default) is the fraction of the cylinder of basal area g and height H filled by the stem; both the calculator and the table use it. Calculator: DBH (cm) and or girth (cm) exclude each other: typing in one empties the other, and if you give the girth the DBH is computed from it. Height (m) is the total height. The tiles respond at once: the DBH with its girth, the basal area, the volume with the chosen factor and the slenderness.

The table of trees

The data card takes one row per tree with the columns Species, DBH (cm), Girth at 1.3 m (cm), Total height (m), Merchantable height (m), Crown diameter (m) and Status. None is required, but each tree needs its DBH or its girth; if it has both, DBH wins. It comes with two examples, Coffee farm: shade trees (25 plots) and Cloud forest (24 plots), and the Use the loaded inventory button takes the trees of Block 3.

With data loaded, five tiles appear (Trees, Mean DBH with the quadratic mean diameter below it, Mean height, Total basal area and Total volume); if some trees lack a height, the first tile says so, and the volume is added up only over those that have one. Two figures follow, Height and diameter by species (the eight species with the most trees in color, the rest in gray as others) and Basal area by species (the twelve with the largest basal area), and two tables: Summary by species, sorted by basal area, and Tree by tree, with the basal area, volume, merchantable volume (if there is a merchantable height), slenderness and crown area of each tree; 60 rows are shown on screen and the downloads carry them all.

Changing the form factor recalculates the calculator and the table at the same time. Bear in mind that a single factor for all species is a rough approximation —bananas, citrus and timber trees live together in the coffee farm—; for the volume of an inventory, Block 3 can use the equation fitted in Section 2.3.

2.2Scaling: logs and trees by sections

Scaling means measuring the true volume of a felled stem, section by section. The card does it at two scales: a single log, with three formulas, and whole trees from a table of sections.

One log: Huber, Smalian and Newton

The fields Large-end diameter (cm), Mid diameter (cm), Small-end diameter (cm) and Length (m) describe the log (34, 30.5 and 27 cm by 2.5 m, by default). The tiles give the volume by each formula and, below Huber and Smalian, their difference from Newton, which the app takes as the reference.

Newton is exact for a paraboloid, a cone or a neiloid; Smalian, which uses only the ends, overestimates in the neiloid butt of the tree, and Huber, which uses only the middle, underestimates. With the default log, Huber gives 0.1827 m³ (−0.4% against Newton), Smalian 0.1851 m³ (0.9%) and Newton 0.1835 m³. The difference grows the more the log departs from a truncated cone, above all in butt logs.

Whole trees

Below the calculator, Minimum top diameter (cm) (10 by default) sets how far the merchantable volume goes, and the Add the stump (cylinder) checkbox adds the stump to the total, scaled as a cylinder with the diameter of the first section. The data card takes one row per measured section: Tree*, DBH (cm), Total height (m)*, Height of the section (m)* and Diameter of the section (cm)*, with at least 4 complete rows. The example Pine: 40 felled trees by sections has the diameter over bark at 0.3 m, at 1.3 m and every 2.5 m up to near the tip.

Each tree is scaled by adding Smalian logs between consecutive sections and a cone from the last section to the tip (g·L/3). The merchantable volume goes up to the height at which the diameter drops to the minimum top diameter, interpolated linearly between the two sections around it. Each tree also yields its form factor, V/(g·H), and its form quotient, the diameter at half height over DBH.

The results are four tiles (Trees scaled, Total volume, which says whether it includes the stump, Mean form factor and Mean form quotient), two figures and the table Volume of each tree, which gives for each tree the stem volume, the stump volume (even if it is not added), the total, the merchantable volume with its height and the two form indices. Tree D01 of the example, for instance, with 34.6 cm and 24.89 m, has 1.1789 m³ in the stem, 1.1681 m³ merchantable up to 21.53 m and a form factor of 0.512.

The interpretation sums up: “The mean form factor of these trees is 0.450: smaller than that of the paraboloid.” That is the value to type in Section 2.1 for standing trees of the same species and condition, although, as the app itself says, the equations of the next section do better, because they let the shape change with size.

2.3Volume equations

A volume equation predicts the volume of a standing tree from its DBH and height. It is fitted with scaled trees and then applied to the whole inventory.

The Use the trees scaled above button takes the DBH, height and stem volume of the trees of Section 2.2 (plus the stump if that checkbox is ticked); if there are no scaled trees, a notice asks you to scale them first. You can also load your own trees into the data card, with DBH (cm)*, Total height (m)* and Volume (m³)*, at least 6. Its example, Pine: volumes of the 40 felled trees, holds the same Smalian volumes, already computed.

The app fits four models:

The logarithmic forms are brought back to the volume scale by multiplying by Meyer’s factor, exp(σ²/2), which corrects the bias of undoing the logarithm. The table Comparison of models (volume scale) gives for each model n, R², RMSE in m³ and as a percentage of the mean volume, the bias and Furnival’s index, and highlights the model with the smallest index. Below it, Show the model chooses which one is shown in detail (by default, the highlighted one): its coefficient table with standard error, t and p, Meyer’s factor if it is logarithmic, and two figures, Observed against predicted and Residuals.

The interpretation says it with the numbers of the example: “The model with the smallest Furnival index is Spurr logarithmic (I = 0.09031; R² = 0.9469; RMSE = 17.4 % of the mean volume). The spread of the residuals grows with tree size (heteroscedasticity); the logarithmic forms stabilize it, which is why they usually win.” The second sentence appears only when the spread of the larger half of the trees is more than 1.8 times that of the smaller half.

2.4Taper

A taper function describes the diameter of the stem at any height. With it you compute how far the merchantable diameter goes and how many logs of a given length come out of a standing tree, with the volume of each.

The Fit with the sections loaded above button fits two models to the relative profile —the diameter over DBH, d/D, against the height over total height, z = h/H— of all the sections of Section 2.2; if no sections are loaded, it first loads those of the pine example.

The table Comparison (on diameter) gives, for each model, the sections used, R² and RMSE of the diameter, the bias and the AIC; the coefficient table of each follows, and the figure Relative stem profile, with all the sections as gray points and the two curves.

The coefficients of the example are b₁ = −1.8810 and b₂ = 0.73497 for Kozak, and b₁ = −2.8202, b₂ = 1.2679, b₃ = −1.0119, b₄ = 32.658, a₁ = 0.71088 and a₂ = 0.082897 for Max and Burkhart: the paraboloid starts turning into a cone at about 71% of the height and the neiloid of the butt takes up the lowest 8%. In Max and Burkhart, b₁, b₂ and b₃ have large standard errors (p values of 0.015, 0.054 and 0.099): that is normal in a segmented model, whose coefficients offset one another; what matters is the profile they draw together.

The logs of a tree

With the profile fitted, the fields DBH of the tree (cm), Total height (m), Top diameter (cm), Log length (m), Stump height (m) and Model (Max–Burkhart or Kozak 1969) describe a standing tree and how it will be bucked. The app gives the total stem volume, the height at which the diameter drops to the minimum top diameter and the whole logs that fit from the stump up to that height, with their merchantable volume; the table Logs of the tree gives for each log where it starts and ends, its large-end and small-end diameters and its volume, and the figure The stem and its logs draws the stem to scale with a dashed cut per log and, in gold, the height to the minimum diameter.

Look at the From (m) column: each log starts 5 cm after the previous one ended (2.80 → 2.85 m). The app leaves that trim allowance between logs; there is no field to change it. That is why, in the example, an eighth log no longer fits: it would start at 18.15 m and end at 20.65 m, above the 20.03 m of the minimum diameter. The volume between two heights is integrated numerically over the profile (Simpson’s rule, 400 intervals), and the height to the minimum diameter is solved by Brent’s method. Changing any of these fields recalculates when you leave the control.

2.5Height–diameter relationship

Measuring the height of every tree is costly; measuring the diameter, cheap. The usual practice is to measure the height of some of the trees, fit a height–diameter curve and estimate the height of the rest with it.

The data card takes Tree, Species, DBH (cm)* and Total height (m) — empty if missing, with at least 6 trees with a measured height (above 1.3 m). Its examples are Pine: plot with missing heights (85 trees; 28 lack their height) and Cloud forest: Liquidambar (the trees of a single species, for a curve of its own); Use the loaded inventory takes the trees of Block 3. Below the card, Species fits the curve with all the trees (All) or with a single species, and Curve to estimate with chooses which curve estimates the missing heights: Lowest AIC, the default option, or a specific one.

The three curves pass through H = 1.3 m as the diameter tends to zero:

They are fitted by nonlinear least squares; Näslund starts from its linearized form, D/√(H − 1.3) = a + bD. The table Comparison of curves gives n, RMSE, R², bias, AIC and BIC, and highlights the curve used for estimating; its coefficients follow, and the figure Height–diameter curves, with the measured trees in gray, the three curves (the chosen one with a thick solid line) and the estimated heights as orange points on it.

The table Estimated heights gives the height of each tree that lacked one (tree 3, with 34.1 cm, 24.7 m), and the ⤓ Full table with filled heights (CSV) button downloads the 85 trees with their measured or estimated height and a column altura_estimada that is 1 for the estimated ones. The interpretation ends with a warning worth taking seriously: “Outside the range of measured diameters (6.0–47.1 cm) the curves may part ways: do not extrapolate far from it.” You can see it in the figure: above 45 cm, Näslund already moves away from the other two.

The block introduces itself like this: Load the trees and, if you have them, the plots (with their shape, size, stratum and coordinates). The inventory then serves biomass, diversity, agroforestry and the map. Everything else is computed on its own as soon as there are trees: there is no compute button, and every change to a control redoes the calculations that depend on it.

The chapter uses the two example inventories. The shaded coffee farm (Start with an example → on the home page) has 25 equal circular plots grouped in farms, with no area per stratum: it serves for simple random, systematic and cluster sampling. The stratified cloud forest (Example data›Open in Inventory on its card) has nested plots in three strata with their area: it serves for stratified sampling, point sampling and the structure of an uneven-aged forest.

3.1The inventory: trees and plots

The card holds two data cards, one below the other. Its subtitle sums it up: One row per tree with its plot; one row per plot with its design. If you do not load plots, all take the design below.

Trees

One row per tree. Plot* and DBH (cm)* are required; the rest are optional: Stratum, Species, Family, Total height (m), Crown diameter (m), Status (alive / dead) and the coordinates x (m) and y (m) of each tree from the center of its plot. A tree counts as standing dead if its status contains “muert”, “seco” or “tocón”, with or without the accent, or “dead”, “stump” or “snag”; any other value, or none, leaves it alive. The examples are Shaded coffee farm: trees and Cloud forest: trees; loading either one also loads its plots.

Plots (optional)

One row per plot, with Plot* (the same identifier as in the trees) and, as the case may be, Stratum, Area of the stratum (ha), Cluster, the dimensions of the plot and its coordinates Latitude and Longitude. The design of each plot is read from its columns in this order:

The collapsible box Default design and area of the property holds the design that plots without one of their own take —Shape (Circular, Square, Rectangular or Nested circular) with its dimensions: Radius (m) 17.84, Side (m) 31.62, Length (m) 50, Width (m) 20, Subplot radius (m) 5.64 and Cut DBH (cm) 10, which correspond to 1,000 m² and, for the subplot, 100 m²— and the Total area (ha) of the property (100 by default), which gives the total of the property in the estimation.

The tiles count the plots (with the number of strata), the trees (with the standing dead), the species and, if there are any, the clusters and the estimated heights. The coffee farm example gives 25 plots in 4 strata —the four management types—, 719 trees with 25 standing dead, 10 species and 5 clusters, and the design line reads circular of 999.9 m² (r = 17.84 m). That line is the first check you should make: if the area is not the one you measured in the field, everything that follows comes out wrong.

3.2Expansion to the hectare

Each measured tree stands for several trees per hectare: Each tree stands for 10,000/a trees per hectare, with a the area of the plot (or of the subplot its diameter belongs to).

In a 1,000 m² plot each tree stands for 10 trees per hectare; in the 100 m² subplot of the cloud forest, for 100. Three controls decide how the volume of each tree is computed: Volume of each tree chooses between g · H · form factor, with the Form factor next to it (0.5), and Equation fitted in Block 2, which uses the winning model of Section 2.3 if there is one (if not, it keeps the factor). The Estimate the missing heights (Näslund) checkbox, ticked by default, fits a Näslund curve with the trees of the same inventory that do have a height (at least 6 are needed) and estimates the others with it; the Heights estimated tile of Section 3.1 says how many. Without height there is no volume: if a tree lacks one, the volume of its plot cannot be computed and the volume estimate warns about it.

Below the figure, the table Values per hectare of each plot gives, for each plot, its stratum, the trees measured, N, G and V per hectare, the quadratic mean diameter, the dominant height and the standing dead per hectare. Plot P01 of the coffee farm, for instance, with 38 trees, has 380 trees/ha, 18.97 m²/ha, 141.2 m³/ha, a Dq of 25.2 cm and a dominant height of 16.5 m. The standing dead are counted apart and do not enter N, G or V: P01 has 0, P04 has 40 dead trees per hectare.

3.3Estimation and sampling error

This is the section that answers the main question of an inventory: how much there is per hectare on the property and how precisely it is known.

Four controls: Variable (Density (trees/ha), Basal area (m²/ha) or Volume (m³/ha)), Design, Confidence (90, 95 or 99 %) and the Finite-population correction checkbox. Each time you load a table of trees or of plots, the app chooses the design: Stratified if there are two or more strata and every plot carries the area of its stratum, and Simple random sampling in any other case; so the cloud forest example opens as stratified and the coffee farm as simple random. As long as you do not load another table, the design is the one you choose.

The tiles give the Mean with the half-width of the interval, the Standard error with its degrees of freedom, the Sampling error as a percentage —t·SE/mean·100, in green if it is 10% or less, in amber up to 20% and in red above 20%—, the Confidence interval and the Total of the property with its interval. The table The same variable under each design follows, with one row for each design the data allow and that of the chosen one highlighted; its Note column gives the standard error under Neyman allocation (stratified), the one the grid would have if treated as random (systematic) or the design effect (cluster). With the stratified design, the table By stratum also appears, and a figure with the mean and the interval of each stratum and that of the stratified estimate.

The total of the property under the stratified design is computed with the sum of the stratum areas (1,000 ha in the cloud forest: 2,346,797 trees), not with the Total area (ha) of the default design, which the other three designs use. If the two figures do not agree, check which one is right.

The coffee farm: simple random, systematic and cluster

With the coffee farm example, the density is 278 trees/ha, with a standard error of 25 and 24 degrees of freedom: a sampling error of 18.3% at 95% and an interval from 227 to 328 (Figure I.5 of the introduction shows this card). Over the default 100 ha, the total is 27,764 trees. The table by design gives two more readings of the same inventory:

The three means are equal; what changes is how much is known. As a systematic grid, neighboring plots resemble each other and the successive-differences variance brings the error down to 10.2%. As clusters —the five farms—, the error rises to 52.7%: the plots of one farm are so alike that, according to the design effect of 4.61, they are worth less than a quarter as many independent plots, and only 4 degrees of freedom remain. Basal area behaves the same way: 17.45 m²/ha with 22.4% as simple random, 18.0% as systematic and 60.0% as cluster; volume, 145.5 m³/ha with 27.3, 23.7 and 71.7%.

The finite-population correction uses as the number of possible plots the total area over the mean plot area (in the coffee farm, 100 ha / 0.1 ha = 1,000 plots). With 25 of 1,000 plots measured, hardly anything changes: the standard error drops from 25 to 24 and the sampling error from 18.3 to 18.0%. It matters when a large fraction of the property is measured. When the error exceeds 10%, the interpretation says how many plots would be needed to bring it down to that value with the coefficient of variation of the inventory: in the coffee farm, about 78 (CV of 44%).

3.4Sample size

Before going to the field, or to plan the second phase of an inventory, the question is the reverse: How many plots are needed for a given sampling error?

The fields are Coefficient of variation (%) (40 by default), Permissible error (%) (10), Possible plots N (empty = infinite), Confidence and Allocation by stratum (Optimum (Neyman) or Proportional). The Take the CV of the loaded inventory button writes the coefficient of variation of the variable chosen in Section 3.3, computed as simple random sampling. The results are recalculated as you type and when an inventory is loaded; with the default values (CV of 40% and error of 10%) they call for 64 plots (t = 1.998 with 63 df).

Since t depends on the degrees of freedom, which depend on n, the app starts with the t of 30 degrees of freedom and repeats with n − 1 until n stops changing. The tiles give the Plots needed with the final t, those needed Without finite correction (if you gave N) and the Iterations.

If the loaded inventory has strata with an area, the table Allocation by stratum computes the size of a stratified sample with the standard deviation and the mean of each stratum and allocates it: in proportion to W_hs_h with Neyman’s optimum allocation, or to W_h with the proportional one. The plots are allocated in whole numbers that add up to the n of the title: no stratum gets fewer than 2, and the rest are assigned by the largest remainders; if the minimum of 2 pushes the sum above the computed n, the title gives that sum. In the cloud forest, the riparian stratum, with 5% of the area, would get less than one plot and gets 2: hence n = 10 (5 + 3 + 2). The table uses the variable chosen in Section 3.3 and is redone when the inventory is loaded or a field of this card is touched. The comparison shows what stratifying is worth: 10 plots instead of 27 for the same error.

3.5Point sampling (relascope)

In point sampling there is no plot: from each point the trees are sighted with a relascope or a prism, and those that are “in”, that is, those that look wider than the angle of the instrument, are counted. Each counted tree stands for BAF/g trees per hectare.

The fields are Basal-area factor (m²/ha) (2 by default) and Form factor (0.5), for volume. The data card takes one row per counted tree: Point*, Species, DBH (cm)* and Height (m). Its example is Cloud forest: 15 points, BAF = 2, with 235 trees; it is not loaded with the cloud forest inventory, but from this same card.

A tree of diameter d (cm) is in if it is closer than R = d/(2√BAF) meters to the point: with BAF = 2, a 30 cm tree is in up to 10.6 m away. Thick trees are seen from farther than thin ones, and so the sample is weighted by basal area without measuring any plot. Across points, the estimate is that of simple random sampling, at 95%.

The table By point gives the trees counted and G, N and V per hectare at each point: at R01, 21 trees were counted, which with BAF = 2 are 42.0 m²/ha. The interpretation compares the average with the recommended range of 7 to 15 trees per point. In the example, with 15.7, it says they are “many; a larger BAF would save time without losing much precision”.

3.6Stand structure

The last card describes how the stand is built: how the diameters are distributed, which species dominate, how dense it is and how much competition each tree faces. It uses only live trees, and at least 10 are needed.

The controls are Class width (cm) (5), Weibull (2 parameters or 3 parameters), Smallest DBH measured (cm), Stratum (All or one) and Hegyi radius (m) (6). Smallest DBH measured (cm) is the diameter from which the trees were measured; when an inventory is loaded, the app proposes the smallest DBH, rounded down to the half centimeter (2.5 cm in the cloud forest, 5 cm in the coffee farm), and empty means no threshold.

Diameter distribution and Weibull model

The figure gives the trees per hectare in each diameter class, already expanded, with the fitted Weibull curve on top. The Weibull is fitted by maximum likelihood with each tree weighted by its expansion factor, so that the trees of the small subplot count for what they represent; with 3 parameters, the location a is searched by profile likelihood below the smallest diameter. If there is a smallest DBH measured t, the distribution is left-truncated —the density is divided by the probability of exceeding t—, because thinner trees exist but were not measured; the curve is then labeled Weibull 2p truncated and the classes start at t, so the first bar is a whole class like the rest.

The table Parameters and goodness of fit gives the location, the scale b, the shape c, the Smallest DBH measured, truncation (cm), the model mean DBH, Trees / effective size (Kish) and two goodness-of-fit tests with the same weights as the fit: Kolmogorov–Smirnov D (weighted), with an asymptotic p value (conservative, because the parameters come from the same sample), and χ² by classes with expected counts of 5 or more. Both use Kish’s effective size, (Σw)²/Σw², instead of the number of trees: with nested plots, the subplot trees weigh ten times more and the sample is worth as much as a much smaller one.

The interpretation reads the shape c: below 1.2, “a reverse J” (many thin trees and few thick ones, typical of an uneven-aged or regenerating forest); from 1.2 to 3, a right-skewed distribution; 3 or more, an almost symmetric bell, typical of an even-aged stand. The cloud forest, with c = 0.79, is a reverse J; the coffee farm, truncated at 5 cm, with c = 2.09 (b = 28.203 cm, mean DBH of 25.57 cm), is a right-skewed distribution. If the Kolmogorov–Smirnov test rejects the fit, the interpretation says so and suggests a mix of cohorts or strata. In the coffee farm it rejects it with D = 0.065 and p = 0.006 (χ² = 51.91 with 12 classes and 9 df, with 694 trees whose effective size is also 694, because all plots weigh the same); in the cloud forest, with D = 0.119 and p < 0.001.

Stand table and importance value index

The Stand table: trees per hectare by species and class crosses the ten most important species (plus an Others row and the Total) with the diameter classes. The classes on the right close with an open class (“≥ 65” in the coffee farm, “≥ 62.5” in the cloud forest, whose classes start at 2.5 cm) once 99% of the trees have been counted, so that the table does not have dozens of nearly empty columns; a dot (·) marks a cell without trees.

The table Importance value index gives for each species its density (N per hectare and relative), its dominance (basal area per hectare and relative) and its frequency (the share of plots in which it appears, absolute and relative), and the IVI, the sum of the three relative values: from 0 to 300, or as a percentage (IVI/3). The figure Components of the IVI (top 10 species) stacks the three parts.

In the cloud forest, the most important species is Liquidambar styraciflua, with an IVI of 31.40, and the nine that follow fall between 11 and 20: a forest without a clearly dominant species, as expected in a cloud forest.

Density and competition

The table Density indices by plot (on screen, the first 30 plots) gives for each plot its N, its Dq, Reineke’s stand density index —SDI = N·(Dq/25)^1.605, the equivalent number of 25 cm trees per hectare—, the dominant height and the Hart–Becking index —the mean spacing between trees, √(10,000/N), as a percentage of the dominant height—. The interpretation gives the averages and reads the Hart–Becking one: below 17%, dense stand, a thinning candidate; above 25%, open stand.

Reineke’s index was devised for even-aged stands and the Hart–Becking index for plantations; in a coffee farm or a cloud forest they serve to compare plots with each other, not as a thinning rule.

If the trees carry coordinates x (m) and y (m) —measured from the center of the plot— and a plot has at least 5 trees with them, the app computes the Hegyi competition index of each tree: the sum, over the neighbors within the chosen radius, of the diameter of the neighbor over that of the tree, divided by the distance. Trees closer than that radius to the edge of the plot are left out of the figure, because they lack neighbors. In nested plots only the trees of the large plot enter.

Block 4 carries the subtitle “Allometric equations, wood density, above- and belowground biomass, necromass, carbon and CO₂e with their uncertainty.” It works on the inventory of Block 3: the trees with their plot, their diameter, their height and their expansion factor. There is no need to load it again; if you already opened an inventory there, just enter the block and click Use the Block 3 inventory. The block’s own four data cards —downed wood, litter, soil and change in stocks— are independent of the inventory, and each is loaded on its own.

The examples of this chapter are the example shaded coffee farm (25 plots of 1,000 m², 719 shade trees), opened with the Open in Biomass button of its card in Block 1›Example data; the example cloud forest for downed wood, and the example pine plantation for the change between two measurements (Section I.7). All are simulated.

4.1Data and equations

The first card decides where the trees come from and which equation computes each one. Everything else in the block depends on these choices: when you change any of them, the app recomputes the biomass, the tables and the pools.

1 2 3 4 5 6 7 8 9 10

• 1Source of the trees. The list chooses between Shaded coffee farm and Cloud forest; Load example opens that inventory (and its settings) and Use the Block 3 inventory takes the one already loaded. Below, the status line gives the name, the plots and the trees.
• 2Default equation. The one applied to every tree without a rule of its own.
• 3E (environmental stress, if the equation asks for it). Only eq. 7 of Chave et al. (2014), without height, uses it.
• 4Root biomass. The IPCC (2006) ecological zone, Cairns et al. (1997) or a constant ratio.
• 5Root:shoot ratio (if constant). 0.24 by default; it only counts with Constant ratio.
• 6Carbon fraction. 0.47 by default.
• 7Fraction left in standing dead. From 0 to 1; 1 by default.
• 8Rules: equation by species or genus. An editable table of exceptions.
• 9Your equations. Local equations with their form and coefficients.
• 10See the catalog of equations (31). The equations included, with their source.

Where the trees come from

The block has no data card for the trees: it takes them from Block 3, with everything computed there —the expansion factor of each tree according to the size of its plot, the heights estimated with Näslund’s curve if requested, the stratum and the area of the property—. If there is no inventory yet, Use the Block 3 inventory says so: “There is no inventory yet: load one in Block 3 or an example here.” Load example opens the example inventory as Block 3 would and, along the way, sets the equation, the root zone and the rules that example needs; thus, for the coffee farm, you get the three rules of Figure 4.1.

From the inventory, each equation uses what it asks for: the diameter at breast height (DBH, in cm) always; total height (m) the equations “with height”; basic wood density (ρ, g/cm³) almost all the pantropical ones, which the app looks up in the table of Section 4.2, and the diameter 15 cm above the ground (d₁₅) the coffee-shrub equation. A tree that lacks what its equation needs adds no biomass, and the app says so with a warning that names what was missing.

The equation of each tree

The Default equation list groups the 31 equations of the catalog by where they were fitted (Table 4.1). With Rules: equation by species or genus you give certain trees another equation: in the Species or genus column you type a species (“Cordia alliodora”) or a genus (“Inga”) and in Equation you choose its equation; + Add row adds a rule and × removes it. The lookup ignores case and, as the card itself says, “A species rule wins over a genus rule.”

See the catalog of equations (31) unfolds the full table: the name, the Form (AGB in kg), the Variables it needs, the residual error σ (ln) when the source publishes it, the Valid DBH (cm) and the Source; ⤓ CSV or ⤓ XLSX downloads it. The number in parentheses grows when you add equations of your own.

Roots, carbon and dead trees

Root biomass (BGB) is not measured: it is estimated from the aboveground biomass of each plot. Root biomass offers three routes:

• The root:shoot ratios of Table 4.4 of IPCC (2006), one option per ecological zone, from IPCC 2006: Tropical rain forest to IPCC 2006: Boreal. In some zones the ratio depends on the aboveground biomass of the plot, and the app chooses it plot by plot. The coffee farm and the cloud forest use IPCC 2006: Tropical mountain systems, with a ratio of 0.27 in all their plots.
• Cairns et al. 1997 (per hectare): eq. 3 of those authors, applied to the aboveground biomass per hectare of each plot. With the example coffee farm it gives 92.9 Mg/ha of total biomass, against 97.8 Mg/ha with the IPCC ratio.
• Constant ratio: that of Root:shoot ratio (if constant), the same for all plots.

Carbon is biomass times the Carbon fraction: 0.47 by default, the default value of Table 4.3 of IPCC (2006). CO₂ equivalent is carbon times 44/12, the ratio of the molar masses of CO₂ and carbon. Trees the inventory marks as standing dead are computed with their same equation, multiplied by the Fraction left in standing dead (with 1 the tree keeps all its biomass; with 0.5, half), and do not add to the live trees: they are reported separately as necromass.

Your equations

A local equation fitted in your region is usually better than a general one. In Your equations, + Add row opens a row with Name, Form (the list of functional forms the app recognizes), the coefficients a, b, c and d, its residual error σ (ln) if you know it, the range D min. – D max. where it holds and the Source. The units are those of the catalog: the result in kg, D in cm, H in m and ρ in g/cm³. The equation appears at once in the Your equations group of the lists above and in the catalog; if you give it no name, it is called “My equation 1”, “My equation 2”…

4.2Wood density

Basic wood density —dry mass over green volume, in g/cm³— multiplies almost every pantropical equation: at equal diameter and height, a hardwood tree has more biomass than a lightwood one. This card tells you which density each species of the inventory receives and where it came from.

At the top is the editable reference table: Species, Density (g/cm³), Level and Source. It holds 42 taxa, those of the examples, with values from the global wood density database (Zanne et al. 2009); Level says whether the value is the species’ own or the mean of its genus. You can correct a value, add your species with + Add row or remove rows with ×; each change recomputes the biomass. The reference table also opens with Open in Biomass on the Densities and equations card of Block 1›Example data, which goes straight to it.

If you have the complete file of the global database, Load the global database (GWDD file)… reads it: CSV, text or a workbook (.xlsx, .xlsm or .ods; in a workbook, the Data sheet). An old-format .xls workbook is not read, and the app asks you to save it as .xlsx or .csv. The app looks for the species and density columns —and, if present, family and region— and reports how many records it loaded. With the database loaded, the Region list appears, which restricts the values to one region of the database or leaves them at World.

The search order

For each tree, the app looks for its density in this order and keeps the first one it finds:

1. Your table, by the exact species name. The level is whatever the table says.
2. The global database, if you loaded it: the mean of the records of the species; if missing, the mean of the species means of the genus; if missing and the inventory carries the family, the mean of the genus means of the family.
3. The plot mean: the average density of the other trees of the same plot.
4. The inventory mean, if not even that exists.

Without the global database, step 2 runs on your own table: a tree of a species that is missing, but whose genus is present, receives the mean of its genus. The result table, Density given to each species of the inventory, shows for each species the Trees, the Density (g/cm³), the Level at which it was found (species, genus, family, plot mean or inventory mean) and the SD for the Monte Carlo: 0.071 g/cm³ if the value is the species’, 0.094 if it is the genus’ and 0.123 if it is the family’s, as in the reference implementation, and one of its own at the plot and inventory levels (0.110 for Musa sp. in Figure 4.2).

4.3Biomass and carbon of the inventory

“Per tree, per plot and for the whole property, with the sampling interval.” This card has no controls: it fills in as soon as there is an inventory and is redone with each change in the two previous ones.

Each live tree contributes its aboveground biomass times its expansion factor; the sum of a plot, divided by a thousand, is its megagrams per hectare (1 Mg = 1 t). Roots are added to that aboveground biomass, and the sum is total biomass, which is converted to carbon and CO₂e. The mean of the property and its interval are computed with the plots as sampling units, with the same estimator as Block 3: stratified when the plots carry their stratum with an area, simple random otherwise.

The tiles read, from left to right:

• Aboveground biomass, Total biomass (with roots), Carbon and CO₂ equivalent, per hectare, each with the half-width of its 95% interval and the sampling error as a percentage. Since carbon and CO₂e are biomass times constants, all four share the same relative error.
• Carbon of the property: carbon per hectare times the property area given in Block 3, with its interval. Without an area, the tile does not appear.
• Outside the range of the equation, in amber: how many live trees have a diameter outside the one their equation holds for, and what percentage they are. In the coffee farm there are 35 (5%): 16 Cordia alliodora, 10 Inga vera, 8 Inga jinicuil and one banana, nearly all trees thicker than the 44 cm up to which the equations of Segura et al. (2006) hold.

Below come three charts —Biomass by plot, stacked into aboveground and roots and sorted from largest to smallest; Aboveground biomass by species, with the first ten and the rest as “Others”, and Biomass of each tree by its equation (Figure 4.4)— and two tables. By plot gives for each plot its stratum, the Aboveground (Mg/ha) and Roots (Mg/ha) biomass, the Root:shoot used, the Carbon (Mg C/ha), the CO₂e (Mg/ha) and the Standing dead (Mg/ha); in the coffee farm, plot P01 has 97.2 Mg/ha of aboveground biomass, 26.2 of roots and 58.0 Mg C/ha. Tree by tree shows the first 40 trees with their DBH (cm), Height (m), ρ (g/cm³), Equation, AGB (kg), Expansion factor and whether it is Alive; the download carries them all.

Figure 4.4 lets you check the equations at a glance: a group of points far from the rest at the same diameter points to an equation that does not suit those species, or to a wrong unit.

4.4Uncertainty by Monte Carlo

The interval of the previous section only says how much the mean could change if other plots had been measured. It says nothing about the errors in measuring diameters and heights, about a density borrowed from another species, or about how wrong the equation is tree by tree. This card propagates them by simulation.

The method repeats the whole biomass computation hundreds of times; in each repetition it perturbs the diameter, height and density of each tree by its error, and the biomass by the residual error of its equation. The spread of the results is the uncertainty of the measurements and the model. The controls:

• Iterations: 500, 1000 (default), 2000 or 5000.
• Diameter error: Chave et al. 2004 (95 % small, 5 % large), those authors’ model of measurement errors —a standard deviation of 0.0062·D + 0.0904 cm for 95% of the trees and of 4.64 cm for a random 5%, which represents gross recording errors—; or a fixed coefficient of variation, CV 1 % or CV 2 %; or No error.
• Error of measured height (CV %), 5 by default, and Error of estimated height (CV %), 15 by default, for the heights Block 3 filled in with a height–diameter curve.
• Residual error of the equation, checked by default: adds the error of each equation that publishes its σ on the log scale.

The computation starts with Run the Monte Carlo, which first needs the biomass of the inventory. The app uses a fixed seed, so with the same data and options, repeating the run gives the same numbers.

The four tiles are the Aboveground biomass (Monte Carlo) with its 95% interval (the 2.5 and 97.5 percentiles of the iterations); Measurements and equation, the half-width of that interval; Sampling, that of Section 4.3, and Combined total, the two added in quadrature —the square root of the sum of their squares—, as in approach 1 of propagation of IPCC (2006). The border of the last one is green if the combined uncertainty is below 10% of the mean, amber between 10 and 20% and red from 20% up. The histogram shows the distribution of the mean biomass of the property over the iterations, with the interval in dashed lines and the direct computation in a solid line, and the table By plot (Monte Carlo) gives for each plot the Direct (Mg/ha) value, the MC mean (Mg/ha), its SD, the 2.5 % and 97.5 % percentiles and the CV (%).

The interpretation ends by saying which source of error dominates. In the coffee farm, “Sampling dominates: more plots would cut the error more than better measurements.” When measurements and equation dominate, the app recommends a local equation or better heights instead.

4.5Necromass, litter and soil

“The other carbon pools, each with its own table.” Live trees are not all the carbon of a forest or a coffee farm: deadwood on the ground, litter and, above all, the soil hold as much again or more. Each is sampled in its own way, and that is why each has its own data card.

Downed wood (line intersect)

Downed wood is measured by line intersect: transects of known length are laid out and the diameter of each piece is measured where the line crosses it. The card asks for Transect*, Length (m)* and Diameter at the crossing (cm)*, and, optionally, the Decay class (1–3); a piece without a class counts as intermediate. Above, three fields give the density of each class: Density class 1, sound (g/cm³), 0.45 by default; Class 2, intermediate, 0.34, and Class 3, rotten, 0.23. They are only a starting point: the right thing is to measure them at your site.

The volume per hectare of each transect comes from Van Wagner’s (1968) formula, V = π²·Σd²/(8L), which with d in cm and L in m gives m³/ha directly; mass is the volume of each piece times the density of its class. The tiles give the Volume of downed wood, the Mass with its interval (the transects are the sampling units) and the Carbon, with the carbon fraction of Section 4.1; the table By transect gives for each one its length, the Pieces, V (m³/ha) and Mass (Mg/ha). In the example cloud forest, transect T01 crosses 20 pieces and gives 199.6 m³/ha.

Litter and soil organic carbon

Litter is collected in frames of known area and weighed dry. The card asks for Plot*, Dry weight (g)* and Frame area (m²)*; the mass in Mg/ha is the dry weight in grams over the frame area in m², times 0.01. The tiles give the Litter with its interval and its Carbon.

Soil organic carbon is computed layer by layer: Mg C/ha = C % × bulk density (g/cm³) × thickness (cm) × (1 − coarse fragments/100), and the layers of each profile are added. The card asks for Plot or profile*, Organic carbon (%)* and Bulk density (g/cm³)*, and the limits of each layer in one of three ways: Upper depth (cm) and Lower depth (cm); or the Layer (e.g. “0-10”), if there are no depths; or the Thickness (cm), if there are no depths or layer, and then the layers of each profile are stacked from the surface in the order they come. Optional: Coarse fragments (%), System or group (optional) and Year (optional).

Always compare soil pools at the same depth: the tile states it (“0–30 cm”), and a soil sampled to 0–20 cm can be neither added to nor compared with another to 0–30 cm. The table By profile gives the carbon of each plot or profile.

4.6Pools and change between inventories

“The carbon of every pool with its error, and how much it changed between two measurements.” The last card gathers what the block computed into a single balance and, with a separate table, measures the change in stocks over time.

The pool balance

The summary brings together, in Mg C/ha, whatever pools there are: Live trees: aboveground and Live trees: roots (from Section 4.3), Standing dead if the inventory has them, Downed wood, Litter and Soil if you loaded their tables, and Shrubs and others if you type a value in Shrubs and others (Mg C/ha) with Its standard error, for what was measured outside the app (shrubs, epiphytes, palms). With nothing computed, it says “Compute the biomass or load a pool to see the summary.”

The Total carbon of the system tile gives the total with the half-width of its 95% interval and its CO₂ equivalent; the chart Carbon by pool stacks them in a single bar, and the table Carbon pools (Figure 4.8) gives each one with its Std. error, its Lower 95 % and Upper 95 % limits, the % of total, the Mg CO₂e/ha and, in the last column, In the error of the total, how that pool enters the error of the total: summed by plot or independent.

The interpretation compares the two calculations. In the example coffee farm: “The error of the total adds aboveground biomass, roots, standing dead, litter and soil plot by plot (25 plots): ±10.7 Mg C/ha. Adding the pools of the same plots in quadrature, as if they were independent, would give ±8.9.” Quadrature would have understated the error.

The balance uses the data of a single site. If a card has the example or the file of a site other than the inventory loaded, that pool is left out with a warning. With the coffee farm as the inventory and the cloud-forest downed wood loaded, the app says: “Left out of the balance because they come from a site other than the inventory (Shaded coffee farm: trees): downed wood (Cloud forest: 12 transects of 30 m). Load the data of the same site in that card.” With the cloud forest as the inventory, in contrast, its downed wood does enter, and the balance of trees and downed wood comes to 159.3 ± 22.6 Mg C/ha.

Change in stocks

To know whether a property gains or loses carbon, it must be measured twice. The Change in stocks card asks for Plot*, Carbon at the 1st measurement (Mg/ha)* and Carbon at the 2nd measurement (Mg/ha)*, and, optionally, Year or age 1 and Year or age 2. If the year columns are complete, the time between measurements is their mean difference; if not, Years between measurements (if there are no year columns) is used, 5 by default. With The same plots (paired) checked, the change is the mean of the plot-by-plot differences, with the paired t test; unchecked, the two measurements are treated as independent samples and the difference of means carries Welch’s (1947) error.

The tiles give the Mean change with its 95% interval —green border if the interval lies above zero, red if below and amber if it includes it—, the Annual rate in Mg C/ha/yr with its interval and its CO₂e equivalent, and the Test with its p value and degrees of freedom. The interpretation for the example pine plantation says that “the 95 % interval excludes zero: the change stands out from the sampling error”, and adds that remeasuring the same plots removes the permanent differences between them from the variance. That is the reason to remeasure permanent plots instead of drawing two new samples: with the same plots, a small change is detected with far fewer plots.

Block 5 carries the subtitle “Current and mean annual increments, growth models, site index and stand projection.” It has two data cards: the tree card (Section 5.1), which also feeds the models of Section 5.2, and the card of remeasured plots (Section 5.3), which also feeds the projection of Section 5.4. It does not depend on the inventory of Block 3.

The example of the whole chapter is the example pine plantation, opened with Open in Growth on its card in Block 1›Example data: the stem analysis of six trees and thirty permanent plots measured six times (Section I.7). The data are simulated.

5.1Increments and biological rotation

“With the stem analysis or the remeasurements of a tree: the current annual increment (CAI), the mean annual increment (MAI) and the age at which MAI peaks.” It is the oldest computation in forest mensuration, and still the one that decides when it pays to harvest.

The data card asks for Tree or series* and Age (years)*, and at least one of the variables Diameter (cm), Height (m) and Volume (m³). A “series” can be a tree rebuilt by stem analysis —its rings give the diameter and height at each age— or a tree measured several times. The example Pine: stem analysis of 6 trees brings 203 rows: trees T1 to T6, with series of 30 to 36 years. Below the card there are three controls:

• Variable: Diameter (cm), Height (m), Volume (m³) (default) or Basal area (m²), computed from the diameter.
• Form factor (if volume is missing): 0.45 by default. If the series has no volume but does have diameter and height, the app computes it as basal area × height × form factor. It does so with the example pine plantation, which carries no volume.
• Tree for the models: the series fitted in Section 5.2 and shown in full in its series table.

The chart CAI (dashed) and MAI (solid): volume (m³) draws both curves for all the trees, one color per tree. Below, the table Biological rotation of each tree gives the Age of maximum MAI, the Maximum MAI and the CAI = MAI crossing (periodic CAI), and the table Series of tree T1 (or of the chosen tree) gives the variable, CAI and MAI year by year. In the example pine plantation, tree T1 reaches 2.2965 m³ at 35 years, with an MAI of 0.06561 m³/yr.

The interpretation gives the mean age of maximum MAI, 33.3 years in the example, and warns: “In 4 trees MAI is still rising at the last measured age: their rotation is beyond the series.” They are T1, T2, T3 and T6, whose maximum falls exactly at their last age; for them, the age in the table is only a lower bound of the rotation.

5.2Growth models

“Four sigmoid curves fitted to the chosen series, with the age of the largest current and of the largest mean increment of each.” A fitted curve smooths out the noise of the rings and lets you compute the rotation even if the series has not reached it.

The card has no controls of its own: it fits the series of the Tree for the models in the Variable chosen in Section 5.1, and is redone when you change them. It needs at least five ages (“At least 5 ages are needed.”). The four models, with A as the asymptote —the maximum size the model attributes to the tree—, are:

Each model is fitted by nonlinear least squares from several starting points. The table Comparison of models (tree T1) gives for each its Equation, the RMSE, the R², the AIC, the Asymptote A, the Age of max CAI —the inflection point, where the curve grows fastest— and the Age of max MAI; the shaded row is the one with the lowest AIC, the criterion the app uses to choose. Gompertz and the logistic do not pass through the origin, so their maximum MAI is searched only from the youngest observed age; if the maximum falls at an end of the search range, the column shows “— (at the edge)”.

Below, Coefficients: Chapman–Richards (or of the winning model) gives each parameter with its standard error, its t and its p: for T1, A = 4.6891, k = 0.050364 and p = 3.9500. The chart Fitted curves shows the observed data and the four curves, extended beyond the last age; the one on the right, the CAI and MAI of the winning model, which cross exactly at the age of maximum MAI.

The interpretation sums it up: “Chapman–Richards has the lowest AIC. By it, the current increment peaks at 27.3 years and the mean increment at 46.0 years.” That 46-year rotation lies beyond the 35 years measured: the curve sets it, not the data. That is why the app adds that the asymptote (4.689 m³) “is an extrapolation: take it as an order of magnitude, not as a fact”. The four models fit almost equally within the data (R² from 0.9948 to 0.9986) and part ways afterwards: the age of maximum MAI ranges from 34.1 years with the logistic to 47.6 with Schumacher.

5.3Site index

“The dominant height each plot reaches at the base age: anamorphic or polymorphic curves fitted to the remeasurements, and the site class of each plot.” Dominant height hardly depends on stand density or thinnings, and that is why it is the usual measure of site quality.

The data card asks for Plot*, Age (years)* and Dominant height (m)*, one row per plot and measurement. The columns Basal area (m²/ha), Volume (m³/ha) and Trees per hectare are optional here, but the projection of Section 5.4 needs the first two. The example Pine: 30 permanent plots, 6 remeasurements brings 180 rows, with ages from 5 to 30 years. The controls:

• Base age (years): the age at which the index is expressed; 20 by default.
• Family of curves: Schumacher anamorphic (guide curve ln H = b₀ + b₁/t), Chapman–Richards anamorphic (guide curve) or Chapman–Richards polymorphic (algebraic differences on the rate), the default.
• Site classes: 3, 4 (default) or 5.

The index of a plot is the mean of the indices of its measurements. The table Site classes splits the range of indices into equal parts —class I is the best— with their limits SI from (m) and SI to (m) and the Plots in each: in the example, class I runs from 25.0 to 28.8 m with 7 plots; II, from 21.2 to 25.0 m with 8; III, from 17.4 to 21.2 m with 11, and IV, from 13.7 to 17.4 m with 4. Site index of each plot gives for each plot its SI at 20 years (m), the SD among measurements, the Measurements and its Class; plot R01, for instance, has SI = 19.68 m (class III).

The SD among measurements is the test that the family of curves works: if a plot grows as its curve says, all its measurements give the same index. The interpretation gives the mean of that spread and judges it: below 0.6 m, “the curves follow well the way the plots grow”; above, the app suggests trying another family.

5.4Stand projection and yield table

“Clutter's compatible system fitted to the same remeasurements projects basal area and volume; with survival and the site curve it builds the yield table, which you can edit.” It is the step from the tree to the stand: how much volume there will be per hectare at each age, and when the maximum mean increment is reached.

The card uses the remeasurements of Section 5.3, with their basal area and volume per hectare, and the site index of each plot. If they are missing, it says so: “The projection needs the basal area and volume per hectare of the remeasurements and the site index of each plot.” At least 6 remeasurement pairs are needed.

The seven fields of the projection are Site index, Initial age (years), Initial basal area (m²/ha), Initial trees per hectare, Annual mortality (%), Final age (years) and Step (years). If you type nothing in the first three, the app proposes them from the data: as the index, the median of the plot indices (21.4 m in the example); as the initial age, the youngest age measured (5 years), and as the initial basal area, the mean of the youngest plots (18.0 m²/ha). The others come with 1200 trees per hectare, 1.5% annual mortality, a final age of 40 years and a 2-year step. Change them to project the stand you are interested in: each change redoes the table, and what you type in those three fields stays even if you fit the site again. If you clear one, the app proposes it again.

The table Clutter system (150 remeasurement pairs) gives the six coefficients with their standard error —in the example, α₀ = 4.3462 and β₃ = 1.9477— and, below, the residual error of each equation (σ of ln G₂ = 0.0472; σ of ln V₂ = 0.0627). The tiles give the Rotation of maximum MAI, with a green border, or amber with the note “MAI still rising at the end of the table” if the maximum is not reached before the final age; the Maximum MAI, and the Volume at rotation. The charts Projected volume and CAI and MAI of volume show the table as curves.

The yield table

The table is shown twice. The first is editable: Age, Dominant height (m), N (trees/ha), G (m²/ha) and V (m³/ha). Its title explains it: “change a value and Dq, CAI, MAI and the rotation are recomputed”. It serves to correct the projection with what you know —a thinning that lowers N and G at a certain age, a measured volume— or to remove with × the ages you do not want; + Add row adds an age. The second, Full yield table, adds the quadratic mean diameter Dq (cm), the CAI (m³/ha/yr) and the MAI (m³/ha/yr), with the rotation row shaded, and it is the one that is downloaded.

Each column comes from a different piece: dominant height, from the site curve of Section 5.3 with the chosen index; N, from survival at a constant annual rate, N = N₀(1 − m)^{t − t₀}; G and V, from the Clutter system, and Dq = √(40,000·G/(π·N)). In the example, MAI reaches 23.74 m³/ha/yr at 19 years and drops to 23.69 at 21: the maximum is flat, and a few years before or after the rotation the mean yield barely changes. The interpretation puts it like this: “With site index 21.4 m, the mean annual volume increment peaks at 19.0 years (23.7 m³/ha/yr). The system was fitted with plots from 5.0 to 30.0 years: beyond, the table is extrapolation.”

Open the block with Block 6›Diversity and ecology in the bar on the left. Its title on screen is Forest diversity and ecology, with the line “Richness, indices and Hill numbers, rarefaction, similarity among sites, vertical structure and spatial pattern.” It has six cards. Two of them read data: The community, on which the next four depend, and Spatial pattern of the trees, which needs the coordinates of every stem and works on its own. The example of this chapter is the cloud forest (Section I.7): the Open in Diversity button on the home page, or the example of each card, loads both the community of its 24 plots and the mapped 50 × 50 m plot.

6.1The community

The whole block starts from a long table: one row per site and species with its abundance, or one row per individual. The app turns it into the sites-by-species matrix used by Sections 6.2 to 6.5.

The data card is the usual one (Section I.6): example, file, pasted table, table typed by hand, Use the loaded inventory (the one in Block 3) or ⤓ Template. It asks for these columns:

With Use the loaded inventory, the app takes the live trees of Block 3, each one as an individual, with its plot as the site, its stratum (or its cluster) as the group, and its height. The diversity thus comes from the same inventory that gave the growing stock, without preparing another table.

Below come three tiles—Sites (with the number of groups), Species and Individuals—and the Sites by species matrix, which shows up to 30 rows on screen and downloads in full with ⤓ CSV or ⤓ XLSX.

6.2Diversity indices

The card asks for nothing: as soon as there is a community, it computes the classic indices for each site and for all sites together.

The six tiles refer to the whole set: Observed richness (with Chao1 and ACE below), Shannon H′ (with its exponential, in “effective species”), Simpson 1 − λ (with 1/λ), Pielou J, Fisher's α and Sample coverage, which is marked in green when it exceeds 95%. The Indices by site table repeats everything by plot—N, S, H′, 1 − λ, 1/λ, J, α, Margalef, Chao1, ACE and the coverage—and closes with the All sites row. The chart Observed and estimated richness by site shows, for each plot sorted from most to fewest species, the observed richness and the two estimates.

The app’s interpretation applies three rules: if Chao1 exceeds the observed richness by less than half a species, “sampling captured practically all of them”; the coverage says what share of the individuals belongs to species not yet seen; and if J exceeds 0.8, “individuals are evenly spread,” otherwise “a few species dominate.” For the cloud forest example it says that practically all species were captured and that the spread is even (J = 0.93); the coffee farm example gives 10 species, Chao1 = 10.0, H′ = 2.057 and J = 0.893.

6.3Hill numbers, rarefaction and accumulation

Hill numbers put all the indices on a single scale, that of effective species: how many equally abundant species would give the same diversity. Rarefaction and accumulation tell how richness grows with sampling effort.

This card asks for nothing either. If the community has between 2 and 8 groups, each chart carries one curve per group besides that of All (in black). There are three charts and one table:

• Hill diversity profile: the effective number of species ^qD for order q from 0 to 3. At q = 0 it is richness; at q = 1, exp(H′); at q = 2, the inverse Simpson.
• Rarefaction (solid) and extrapolation (dashed): the expected species in a subsample of m individuals, up to the observed size (the dot) and extrapolated up to twice that size.
• Species accumulation by plot (± 1 SD): the expected species as 1, 2, … plots are pooled, with its band.
• Observed and asymptotic diversity: for each set, the individuals, the observed and asymptotic diversity at q = 0, 1 and 2, and the coverage.

The Hill profile is the most honest way to compare communities: if two profiles cross, one community is richer but the other more even, and no single index would tell you so. In the example, the disturbed and riparian strata almost coincide along the whole profile, although the disturbed stratum has 593 individuals and the riparian one 391.

The interpretation looks at the last stretch of the accumulation curve: if the last plot added less than half a species, “the accumulation curve is already leveling off; more plots would hardly add species”; otherwise, “the curve is still rising.”

6.4Similarity among sites and ordination

How much do the sites share species? The card computes the dissimilarity between every pair of sites and summarizes it in a heat map, a dendrogram and a two-dimensional ordination. At least 3 sites are needed.

The only control is Index: Bray–Curtis (the default, with abundances), Jaccard or Sørensen (both with presence and absence). The card draws the Dissimilarity among sites as a heat map (0 = identical, 1 = nothing in common), the UPGMA dendrogram with its cophenetic correlation in the title, and the NMDS ordination with its stress; the sites are colored by group.

The buttons Send the matrix to ClusteringPro and Send the matrix to PCAPro download the sites-by-species matrix as CSV (with the site and group columns first) and open that app of the suite in a new tab, where you load it with its Open file. Opening the apps requires a connection; the file is downloaded anyway.

Changing the index changes the reading. With Jaccard the mean dissimilarity of the cloud forest is 0.327 within strata and 0.398 between them, and with Sørensen, 0.198 and 0.251; with both the stress rises to 0.242. Since almost every plot has almost every species, presence and absence tell little apart: what separates the strata is how much of each species there is, and only Bray–Curtis sees that. (Jaccard and Sørensen give the same NMDS because one is an increasing function of the other, and NMDS only uses the order.)

The interpretation compares the mean dissimilarity between groups with that within groups: if the former is more than 1.15 times the latter, “the groups have different compositions, and the ordination separates them.” The stress reads as a good representation below 0.10, fair up to 0.20 and poor above that.

6.5Vertical structure

Do all species occupy all stories, or does each story have its own? The card summarizes this with Pretzsch’s species profile index and needs the height of at least 10 individuals; without it, it says so.

The app divides the stand into three zones by maximum height: zone I, above 80% of the maximum height; zone II, 50 to 80%; zone III, up to 50%. The tiles give Pretzsch's A (with its maximum possible value), Relative A and Maximum height. The Pretzsch's index by group table repeats them for all individuals and for each group, with the individuals in each zone; the chart Individuals by height zone (10 most abundant species) stacks the three zones of each species; and the table Strata from the mean and SD of height gives another, simpler division into lower, middle and upper strata.

Each group uses its own maximum height, so the zones of a group are not those of the whole set: that is why the conserved stratum has 21 individuals in its zone I although the whole set has only 11. The interpretation calls the vertical structure complex if relative A exceeds 70%, and simple otherwise; in the example, 77%.

6.6Spatial pattern of the trees

With the coordinates of every stem in a rectangular plot, the card tells you whether the trees are clustered, randomly distributed or regularly spaced, and at what distances. At least 10 trees are needed.

The data card asks for x (m) * and y (m) * of each stem, measured from one corner of the plot, and, if you have them, Species, DBH (cm) (for the size of the points on the map) and Group (storey, cohort…). The controls are above: Plot width (m) and Plot length (m) (50 and 50), Maximum distance for K (m) (12), Simulations for the envelope (99; the app accepts 19 to 999) and Trees, which selects All or one group. Trees that fall outside the rectangle are not used, and a warning says how many there are.

The tiles give Clark–Evans R with its z and p, R with Donnelly correction with its p, and the Density. On the left is the Stem map (size by DBH), colored by species; on the right, L(r) − r with an envelope of 99 simulations: the line is what was observed and the gray band, what randomness would give with the same number of trees.

The interpretation uses the corrected R: with p < 0.05 it says whether the trees are more clustered or more regularly spaced than at random, and then gives the distance ranges at which L(r) − r leaves the envelope from above (clustering) or from below (regularity or competition).

Open the block with Block 7›Agroforestry systems. Its line on screen reads “Shade of coffee farms (by inventory or by hemispherical photograph), shade and yield, planting layouts and carbon budget.” It is designed for coffee and other shaded crops, and its five cards are independent: each has its own example button. The Open in Agroforestry button of the coffee farm on the home page (Section I.7) fills them all at once: it loads the coffee farm inventory in Block 3, computes the shade, reads the 25 plots of shade and yield, analyzes the synthetic photo and builds the carbon budget. The figures in this chapter are those of that example.

7.1Shade from the inventory

If the inventory has the crown diameter of each shade tree, the app computes for each plot the density, the basal area, the crown cover, the richness of shade species and a complexity index, and summarizes them by management type.

The card has no data card of its own: it reads the Block 3 inventory. Load the example coffee farm loads the coffee farm in Block 3 and comes back here; Use the Block 3 inventory takes the one already loaded. Live trees count, with the area of each plot according to its design, and the group of each plot (its stratum or its cluster) serves as the management type. If no tree has a crown diameter, the app warns that cover cannot be computed and suggests the photograph (Section 7.2).

The five tiles are plot averages. If there is more than one management type, the table By management type gives for each one the plots, trees per hectare, basal area, cover with its standard deviation, species and Holdridge index, sorted from most to least cover, and the chart Crown cover by management (mean ± SD) draws the mean cover of each management type with its standard deviation. Below, the table By plot also gives for each plot the sum of the crowns (Σ crowns (%)) and the mean height.

7.2Shade from a photograph of the canopy

When the inventory has no crowns, or to measure the shade at one point, a photo of the canopy taken from below is enough: the app separates sky from foliage and measures the openness.

Open a photo… reads an image from your computer (the photo does not leave it; if it is large, it is reduced to 640 pixels per side for the computation). Example photo (synthetic) draws a test hemispherical photo under a shade canopy. The controls are:

• Kind of photo: Hemispherical (fisheye), taken with a fisheye lens toward the zenith, or Ordinary upward, a common photo.
• Channel: Blue (recommended) or Luminance. In blue the sky is much brighter than the leaves, and they separate better.
• Threshold (empty = Otsu): the gray level (0 to 255) that separates sky and canopy. Left empty, the app computes it.
• Circle radius (% of the short side): the useful circle of the lens (100 by default); it only counts for the hemispherical photo.

The tiles give the Canopy cover, the Canopy openness (only with the hemispherical photo), the Gap fraction with the pixels counted, and the Threshold, with “Otsu” or “manual.” With the ordinary photo, cover is 1 − the sky fraction. Below come the Gray histogram and threshold and, for the hemispherical photo, the table Openness by zenith-angle rings, with the visible sky in 15° rings.

7.3Shade and crop yield

At what shade does the crop yield most? The card fits the regression of yield on shade, linear and quadratic, picks the better one and, if the curve has a maximum, gives the shade of maximum yield with its interval.

The data card asks for Shade (%) * and Yield * of each plot, and optionally Plot and Group (management), which colors the points. At least 5 plots are needed. The example, Coffee farm: 25 plots, has the measured shade and the yield of parchment coffee (kg/ha) of each plot. The shade can come from Section 7.1, from the photos of Section 7.2 or from any other measurement.

The tiles give the Model chosen, with the F test of the quadratic term; the Shade of maximum yield with its interval; the Estimated maximum yield; and the R² (quadratic) next to the linear one. The chart Yield against shade draws both curves—the quadratic with its 95% confidence band for the mean—and a vertical line at the shade of the maximum. The table Coefficients of the quadratic model gives b0, b1 and b2 with their standard error, t and p: in the example, b1 = 28.648 and b2 = −0.36872.

The interpretation says the curvature is significant if the quadratic term has p < 0.05; otherwise it warns that “the data do not tell an optimum from a linear trend.” If the curve has no maximum (b₂ ≥ 0), it says so and gives no optimal shade.

7.4Planting layout

For a new planting: how many trees fit in a rectangular field with a given layout and spacing, how many per hectare, how much they cost and how they share the field with the crop.

The card reads no data; it draws the plan as soon as the block opens, with the default values, and redraws it with each change. Layout offers Square, Triangular, Rectangular, Alley cropping, Live fence (perimeter) and Boundary rows (two sides). The other fields are Field width (m) and Field length (m) (100 and 100), Spacing between trees (m) (6), Spacing between rows (m) (8, for the rectangle), Alley width (m) (12, for alleys), Crop: spacing (m) and Crop: between rows (m) (2 and 1.5) and Cost per tree planted (25, in whatever currency you use).

The tiles give the Trees in the field (with the area in hectares), the Density per hectare with the theoretical one, the Area per tree, the Crop plants and, if there is a cost, the Cost of the trees. The plan is to scale: the trees in dark green and the crop in light green, on its own grid. The crop plants are counted over the whole field, as if the trees took no room from them.

7.5Carbon budget of the system

How much carbon does the shaded coffee farm hold against a full-sun monoculture? The card adds up that of the shade trees, the coffee bushes, the litter and the soil of the coffee farm example and compares it with an assumed monoculture.

The card works with the tables of the coffee farm example, not with your own data: Compute with the example coffee farm runs the computation, and afterwards each change in the fields repeats it. The fields are Coffee bushes per hectare (system) (left empty, the app enters the mean density of the example plots, 3447), Coffee bushes per hectare (monoculture) (5000), Soil carbon of the monoculture (Mg C/ha) (60), Litter of the monoculture (Mg/ha) (3), Carbon fraction (0.47) and Root:shoot of the crop (0.27).

The interpretation says how much more the shaded coffee farm holds and in which pools most of the difference lies: in the shade trees and, if the soil of the coffee farm exceeds the assumed one, also in the soil.

Open the block with Block 8›Maps; its title on screen is Maps and light geoprocessing, with the line “Plots and trees on the map, polygon areas, sampling grid, GPX/KML/GeoJSON export and interpolation.” The five cards share the same layers: what you load, import or draw in one appears on the map and in the others. The main example is the cloud forest (Open in Maps on the home page, Section I.7): its 24 plots with coordinates and the polygon of the property, with 9 vertices. The interpolation also uses the 25 plots of the coffee farm.

8.1The map

A map of the app’s own, which pans and zooms like web maps, with the plots drawn at their true radius and the trees in their place. Below it there is a static copy in UTM coordinates, the one that goes into the report.

1 2 3 4 5 6 7 8

• 1The example. The list chooses between Cloud forest: plots and property and Coffee farm: plots of 5 farms; Load example loads it in Block 3 and brings its plots here (and the property, for the cloud forest).
• 2Use the Block 3 inventory. Takes the plots of the loaded inventory, if they have latitude and longitude, with their radius and stratum, and the trees that have x and y.
• 3Import GPX, KML or GeoJSON… Adds the polygons and points of a file; a message says how many were imported.
• 4The layers. Plots, Trees, Polygons, Grid and Interpolation are switched on and off separately.
• 5OpenStreetMap background (requests tiles from the internet). Off by default. When on, the browser requests the tiles of the visible area; your data are not sent.
• 6The map buttons. + and − zoom in and out; ⤢ fits the map to the data.
• 7The scale and, on the right, the background note: with no background, “No background: latitude and longitude graticule.”
• 8Draw a polygon. One click per vertex; a double click or Finish the polygon closes it and Esc cancels it.

The map pans by dragging and zooms with the wheel, the buttons or the + and − keys; the arrows move it. Plots are circles at their true radius; those without a radius (imported points) are drawn as points. Trees appear as you zoom in, from zoom level 15, with a size according to their DBH: the app places them by adding their local coordinates (x to the east, y to the north, in meters) to the center of their plot in UTM.

Below the map, the Static map in UTM (for the report) repeats the polygons, the plots (colored by stratum if there is more than one), the grid and the interpolated surface in the UTM zone of the center of the data, with a north arrow and a scale. It is the figure that goes into the Block 10 report and the figure studio.

8.2Polygons and areas

The area and perimeter of each polygon—the property, the stands, the harvest areas—computed in two ways: in the UTM zone of its center and on the sphere.

The data card asks for the vertices of each polygon in order: Polygon * (the name, which groups the vertices), Latitude * and Longitude *, in decimal degrees. A polygon needs at least three vertices; there is no need to repeat the first one at the end. The example is the Cloud forest property (9 vertices). The polygons imported and those drawn on the map also reach the Polygons table; Remove the last polygon undoes the last one.

The interpretation compares the two areas—in the example they differ by 0.27%—and recommends using the UTM area for the inventory: it is the one used in cadastre and the one that matches the distances measured in the field.

8.3Sampling grid

The plot centers of a new inventory: a systematic grid or random points inside a polygon, with their geographic and UTM coordinates, ready for the GPS.

The controls are Polygon (any of those loaded), Design (Systematic (grid) or Random), Spacing between points (m) (500) and Rotation (°) (0) for the grid, Random points (30) for the random design, Plot radius (m) (17.84, that of 1,000 m²) to compute the sampling intensity, and the Random start of the grid check box. Generate the points computes them and puts them on the map.

The tiles give the Points (with those expected, for the grid), the Area of the polygon and the Sampling intensity: the area of the plots over that of the polygon. The Sampling points table shows the first 25—number, latitude, longitude, UTM easting and northing—and its downloads carry all of them; the ⤓ GPX, ⤓ KML and ⤓ GeoJSON buttons of this card export only the grid (forestpro_grid.gpx…), with the points named G1, G2…

8.4Export

The map layers in the formats read by GPS receivers, virtual globes and mapping programs.

The four buttons—⤓ GPX, ⤓ KML, ⤓ GeoJSON and ⤓ CSV—download in a single file (forestpro_map.gpx, .kml, .geojson or .csv) the loaded plots, the grid points and the polygons. Each plot carries its name, its stratum and, if it has them, its basal area and density; each grid point, its UTM coordinates. The CSV has the columns name, lat, lon and description, and does not carry the polygons. If nothing is loaded, a message says so.

8.5Interpolation among plots

A continuous map of a variable measured in the plots—basal area, carbon, yield, shade—by inverse distance or by kriging, with the variogram that tells whether the variable has spatial structure and the cross-validation that tells which method predicts better.

The data come from two sources. If the plots on the map come from the inventory, Variable offers theirs: Basal area (m²/ha), Density (trees/ha), Volume (m³/ha) and, if they were computed in Block 4 before bringing in the inventory, Aboveground biomass (Mg/ha) and Carbon (Mg C/ha). Or else the data card of this section reads a table of your own with Latitude *, Longitude *, Value * and Name; its two examples are the yield of parchment coffee (kg/ha) and the shade (%) of the 25 plots of the coffee farm, and when they are loaded the points replace the map layers.

The controls are Method (Ordinary kriging, Simple kriging or Inverse distance (IDW)), Variogram model (Spherical, Exponential or Gaussian), IDW power (2), Variogram cutoff (m, empty = automatic), Cells on the long side (60; from 10 to 200) and Clip to the first polygon. Interpolate runs the computation; at least 5 points with a value are needed.

The tiles give the Points and the method, the Range of the map and two cross-validations, each with its root mean square error (RMSE) and its mean error: Cross-validation: ordinary kriging (or Cross-validation: simple kriging, if that kriging was chosen; with IDW chosen, the ordinary one) and Cross-validation: IDW. Both are always computed, so they can be compared. Then come the chart Experimental variogram and fitted model (the size of each point according to the pairs of plots that form it) and the Variogram model table, with the nugget, the partial sill, the range, the nugget proportion and the cutoff with the width of the lag classes. The surface is painted on the interactive map (Interpolation layer) and on the static map, with its color scale.

The interpretation compares the RMSE of the two methods; reads the nugget proportion—if it exceeds 60%, “most of the variation is short-range or error,” otherwise “there is clear spatial dependence” up to the practical range—; and closes by reminding you that with few plots the map serves to locate zones, not to read point values. The pure nugget effect is declared when the range is smaller than the first lag class of the variogram or the partial sill is less than 5% of the total sill.

This block does not use the data card of the other blocks (Section I.6): its data are few and go into editable tables, with one row per log, per kind of piece or per money item. Each row has its × button to remove it and, below the table, + Add row. Cells are filled one by one: these tables read neither files nor pasted blocks. Results appear when you press the button of each section (Scale, Tally, Compute the yield, Compute, Evaluate); the unit conversion answers by itself as you type.

The pine plantation example opens the four sections that have it already computed: from Home›Example data, the Open in Utilities button of the pine card loads the 12 logs, the lumber sawn from them and the 25-year plantation cash flow, and computes the yield. Every figure in this chapter comes from that example, which is simulated.

9.1Log scaling

A log is measured in the yard with a tape and a caliper: its length and the diameters of its ends. With that, the section gives its volume in cubic meters by four formulas and the sawn lumber expected from it, in board feet, by the three log rules most used in the timber trade.

At the top are Load example: 12 pine logs, Empty table (leaves a single blank row) and the Volume under bark checkbox. The table asks, per log:

The note under the table sums it up: “Length in meters; diameters over bark in centimeters (small end, large end and, if you have it, the middle); bark thickness in millimeters, on one side.” A row without a length or without one of the two end diameters is not scaled.

The Volume by log table gives, for each log and in total, the diameter inside bark in inches (D (in)) and the length in feet (L (ft)) the rules work with, the four volumes and the board feet of the three rules.

With Volume under bark, the cubic-meter volumes are recomputed with the diameters minus the bark: the 12 logs drop to 2.828 m³ by Smalian, 2.793 by Newton and 2.775 by Huber. The board feet do not change (599, 380 and 539), because the rules always work with the diameter inside bark. If in some log the small end is thicker than the large end, the app warns: “In … logs the small end is thicker than the large end; check the order of the columns.”

The Board feet by log by rule figure puts the three rules side by side, log by log, and the interpretation sums up what the differences mean.

9.2Sawn lumber

The other end of the sawmill: what came out of the logs, counted by kind of piece. The section adds up the board feet and cubic meters of a list of pieces and says what share of the volume each kind is.

The Units menu chooses Inches and feet (thickness and width in inches, length in feet, with the nominal sizes used in the timber trade) or Centimeters and meters (thickness and width in centimeters, length in meters). The table asks for Piece (the name), Thickness, Width, Length and Pieces, the number of identical pieces (1 if left empty). Load example: the lumber from the 12 logs enters seven kinds of 8¼-foot pieces, and Tally does the sum.

A board foot is a piece of 1 inch × 12 inches × 1 foot, 2.359737216 dm³ (NIST 2008). So an 8¼-foot 1″ × 6″ board is 4.125 board feet, and the 50 of the example, 206.25. When you switch units, the table changes its column labels but does not convert the numbers you already typed.

9.3Sawmill yield

How much sawn lumber comes out of the roundwood. It is the figure a sawmill measures, and the one that tells whether a log rule falls short for it.

Three fields: Roundwood (m³), Sawn lumber (m³) and Reference rule (board feet). “They fill themselves with the totals of the two sections above (Smalian and International ¼″); you may type others.” The rule is optional; without it there is no overrun.

• Lumber recovery = m³ sawn / m³ of roundwood. The rest is sawdust, slabs, trimmings and bark.
• Board feet per m³ of roundwood: the same ratio in board feet (0.521 × 423.776 ≈ 221).
• Overrun = board feet sawn / board feet of the rule − 1. Positive if the mill gets more than the rule predicts (Steele 1984).

The interpretation grades the recovery: below 45% it calls it low (“small or crooked logs, thick-kerf saws or a cutting pattern that wastes the log”), above 65% high (“typical of large straight logs or thin saws”), and in between, common for band or circular saw mills with medium logs. If more sawn lumber than roundwood comes out, the app flags it in red: “More sawn lumber than roundwood comes out: check the units or the data.”

9.4Form factor

The fraction of the cylinder (basal area × height) the stem fills. With it, the volume of a standing tree is estimated from its diameter and height when there is no volume equation for the species.

Four fields, with sample values: Diameter at breast height (cm) (32), Height (m) (24), Volume (m³) (0.85) and Form factor (empty). “Leave empty what you want to compute: the factor (from volume) or the volume (from factor).” Compute gives the factor, the volume and the reference cylinder, and the Reference solids table says what volume the tree would have with the shape of each solid.

This is the artificial form factor, referred to breast height (Prodan et al. 1997). The reference solids, measured at their base, have f = 1 (cylinder), 1/2 (paraboloid), 1/3 (cone) and 1/4 (neiloid). In conifers it usually lies between 0.40 and 0.55. With a factor of 0.45 instead of the volume, the same tree would hold 0.8686 m³. The interpretation warns if the factor is above 1 (“points to an error in the volume or the sizes”) or below 0.3 (“rare in a commercial stem: check the volume”). If you have scaled trees, the volume equations of Block 2 (Chapter 2) are a better substitute for a fixed form factor.

9.5Unit conversion

From imperial to metric and back, with the exact definitions: what you need to read an inventory, a yield table or a log rule from another country.

Quantity chooses what is converted; Value and From, the number and its unit. The … equal table (“1 meter equals,” with the value 1) gives the value in each of the other units of that quantity, and is recomputed as you type:

The definitions are the exact ones: 1 inch = 2.54 cm, 1 foot = 0.3048 m, 1 acre = 4,046.8564224 m², 1 pound = 0.45359237 kg, 1 board foot = 144 cubic inches and 1 cord = 128 ft³ of stacked wood (NIST 2008). Per-area quantities combine the factor of the unit and that of the area: 120 ft²/acre of basal area are 27.5482 m²/ha. Carbon becomes CO₂ equivalent when multiplied by 44/12: 46 Mg C/ha are 168.67 Mg CO₂e/ha.

Stacked wood

Below the table, Stacked wood (stacked m³) and Stacking factor (solid/stacked) give the Solid wood. A stacked cubic meter is the space taken by piled firewood, gaps included; the stacking factor, the part that is wood. It depends on the diameter and straightness of the pieces: 0.6 to 0.7 is usual for firewood, and the app opens with 0.65.

9.6Cost and profitability of a plantation

A forest plantation spends almost everything at the start and earns almost everything at the end, twenty or thirty years later. To know whether it pays, every dollar has to be brought to the present with a discount rate. The section does that sum per hectare with the measures of forest economics.

Three fields: Discount rate (% per year) (8), Fixed annual cost (protection, management) (900), charged every year from 1 to the rotation, and Rotation (years; empty = last year of the flow). The table has one row per item: Year, Item, Cost and Revenue, per hectare and in the currency you use; “several rows may share a year.” Load example: pine at 25 years brings eleven items, from site preparation to the final harvest, and Evaluate does the sum.

Below the tiles come two figures. Cash flow by year puts revenues upward and costs downward; Net present value by discount rate draws the NPV between 0 and 15%, with the chosen rate and the IRR, which is where the curve crosses zero.

The Discounted cash flow table gives, by year, the costs (annual cost included), the revenues, the net, the discounted net and the cumulative discounted flow, which in the pine goes from −$30,227 in year 24 to $3,956 in year 25. It only shows the years with an item of their own (in the pine, 0, 1, 2, 3, 5, 8, 12, 18 and 25); its note says so: “Years with nothing but the annual cost are not shown; they are counted.”

The figures of this chapter were made with the example coffee farm open in Blocks 3, 4 and 7 (the Open in Inventory, Open in Biomass and Open in Agroforestry buttons of its card in Home›Example data) and the pine plantation in Block 9 (Open in Utilities). With that, four blocks have results, and their numbers in the left-hand bar have turned into check marks.

10.1Report

The report is an HTML file that carries everything inside: the cover, a summary, the figures, the tables, the interpretations and, if you wish, the methods with their sources. It depends on neither the app nor the internet: it opens with a double click in any browser, travels by email and prints to PDF.

At the top go the data of the cover: Title (it opens with “Forest report”), Property or project, Prepared by, Date (today’s) and Introduction or notes (optional), a free text the report puts in its own section. Below, the Blocks to include box has one checkbox per results block, all checked, and two more: With the methods and their sources (the “Method and sources” box of each card) and With the warnings (the warning messages of the blocks).

The checkboxes carry the number and name of their block, from “2. Tree mensuration” to “9. Forestry tools,” and the report uses the same numbers in its contents and in the titles of its sections (“3. Inventory and sampling”).

The note of the card says what goes in: “What each block shows right now goes in; a block with no results does not appear. Figures go in the colors of the light theme and with whatever you changed in the editor ✎.” In other words, the report is a snapshot of the screen: if you change a control in a block and come back, the report changes. Three buttons:

What the report contains

The figures go inside the file as vector drawings, with their styles written in and the colors of the light theme, even if you work in the dark one; the canopy photograph of Block 7 goes as an image. The tables carry the rows shown on screen: a very long table, which says “Showing … of … rows; the downloads carry them all.” on its card, arrives just as short, and the workbook of Section 10.2 brings it in full. When printed, the page is A4 with 16 and 14 mm margins, each block starts on a new page, and no figure, tile or interpretation is split between two pages.

10.2All the tables

Every result table of Blocks 2 to 9 has its ⤓ CSV and ⤓ XLSX buttons; this card downloads them all at once, with their full numbers, not rounded as on screen.

• ⤓ Excel workbook (XLSX), one sheet per table downloads forestpro_tables.xlsx. Each sheet is named with the block number and the table title (“B9 Volume by log”), cut to 31 characters, the limit of the format.
• ⤓ CSV, one file per table (ZIP) downloads forestpro_tables_csv.zip, with one CSV per table, under the file name the table uses on its own button (forestpro_logs.csv, forestpro_cash_flow.csv…). They are comma-separated, with a decimal point and in UTF-8 with a byte-order mark, so a spreadsheet reads accented characters correctly.

The message counts the tables; with no results it says “There are no result tables yet.”. The tables of the data cards (what you loaded) do not go in: only the result tables.

The figures: the studio and the editor ✎

The figures do not go into the workbook: “Figures are exported one by one from their card, as SVG, PNG at 300 dpi or PDF, with the figure studio.” Every figure of the app has two buttons in its corner: the pencil and ruler opens the figure studio, and ✎ opens the editor of that figure. Both are the same in every app of the suite.

The studio puts the figure on the left, to scale and with the outline of its output size, and the inspector on the right. The top bar undoes and redoes, compares before and now with a curtain, zooms in and out (Fit makes it the size of the screen), shows a ruler, applies one-click Presets and, with Export, downloads the figure as PNG, SVG, vector PDF or TIFF, in millimeters or inches, at 300, 600 or 1,200 dpi, with a transparent background if you ask. Other buttons move the inspector to the left or let it float, and Close studio returns to the app. In the inspector, Search a control… finds any setting by its name, and the sections are Figure settings (the controls the figure has in its block; ForestPro’s have none, and the studio says “This figure has no options of its own: use the studio’s (palette, text, size) or the app’s editor if it has one.”), Fine editing of this figure (the app’s ✎), Palette and color, Text and lines, Legend, Size and export, Presets and History. The footer line recalls the shortcuts: “Click the figure: its section · Double-click a text: change it · Ctrl + wheel: zoom · B: before and now · ?: shortcuts”.

The ✎ button opens the same fine editor without leaving the page, in a panel floating over that figure, with six tabs.

At the foot of the panel, Undo everything on this figure returns it to how the block draws it, and Copy the style to every figure carries the same settings to the others. The edits are kept in this browser, figure by figure, and put back every time the figure is redrawn—when the data, the language or the theme change; as they are written on the figure itself, what you see is what is exported and what goes into the report.

10.3Project

Save the work to a file and open it later, here or on another computer. The project does not store the results but the way to reach them, so it weighs little and always gives the same as the app computes.

With the coffee farm and the pine opened from Home, the project has four steps, one per example. A project with your own data carries those data inside, row by row, because the app needs them to repeat the computation.

10.4How to cite

If ForestPro helped in a piece of work, cite it. The card gives the citation in APA, to copy into the reference list, and the BibTeX entry, for a reference manager.

Copy APA and Copy BibTeX copy it to the clipboard (“Copied.”; if the browser does not allow it, “Could not copy; select it by hand.”), and ⤓ forestpro.bib downloads the entry as a file. The same citation is on the app’s home page (Chapter 1) and at the end of every report. The address it carries is that of its concept DOI on Zenodo (10.5281/zenodo.23876374); Appendix E gives the citation of the manual.

Cite also the methods you used: the allometric equations, the sampling design, the site index curves. Each card names them in its “Method and sources” box, and the report carries them if you leave With the methods and their sources checked; Appendix F gathers them all.

AThe files the app reads and writes

ForestPro runs in the browser and sends nothing to any server: every file comes in from your computer and goes out to the downloads folder. The app does not save your work by itself: if you close the tab without saving the project (Section 10.3), the work of that session is lost.

A.1 · What the app reads

How text files are read

• Encoding. UTF-8, with or without a byte-order mark; UTF-16 with a mark; and, if the file is not valid UTF-8, Latin-1 (Windows-1252).
• Separator. Guessed from the first rows, among tab, semicolon, comma and vertical bar: the one that gives the same number of columns in more rows wins. Quoted fields are respected, line breaks included; a quote that is never closed stops the reading.
• Decimals. If the separator is not the comma and numbers with a decimal comma outnumber those with a point, the comma is read as the decimal mark (2,35 = 2.35); then 1.234,5 reads as 1234.5. With a comma separator, the decimal is always the point, and 1,234,567 reads as a number with thousands commas. Inf and -Inf read as infinities.
• Numbers or text. A column is taken as numeric if more than half of its non-empty cells are numbers; in it, whatever is not a number is left empty.
• Empty rows. They are skipped. Empty or repeated column names become V1, V2… or name.1.
• Workbooks. Each cell is read with its stored value, not the displayed one; dates as YYYY-MM-DD, and error cells (#N/A, #DIV/0!) as empty. Old-format workbooks (.xls) do not open: save them as .xlsx or .csv.

How columns are matched

Each field has its key (the one the ⤓ Template writes in the header row), its label in Spanish and in English, and several synonyms. The name of each column of your table is compared ignoring case, accents, spaces and signs, and whatever is in parentheses or brackets (so DAP (cm), dap and D.A.P. are the same). First an exact match with the key, the label or a synonym is sought; if there is none, a column whose name starts with one of them, of three letters or more. What is matched appears already chosen in each field’s list, and the required ones carry an asterisk.

A.2 · The columns of each data card

The first column is the key, the name the app recognizes without asking; the asterisk marks the required fields. The synonyms are compared as explained above, and the label and the key count too. The keys and many synonyms are Spanish words, because the app was written in Spanish; the English labels are those the app shows in English.

The Smallest DBH measured (cm) of Block 3›Stand structure is not a column but a field: the diameter from which trees were measured (2.5 cm in the example cloud forest, 5 cm in the coffee farm), which truncates the Weibull and sets the start of the classes (B.2). Do not confuse it with dap_corte_cm, the diameter that separates the large plot from the small one in nested plots.

The example CSV files of the data/ folder use these names, with a decimal point and in UTF-8, so they load without matching anything. The soil card also reads the layer format of soil analyses, with its system, year, layer written as “0-10” and thickness; with several systems or years, it computes them one at a time. A template for any card, with the keys and three rows of the first example, comes from its ⤓ Template button as forestpro_<card>_template.csv (for example, forestpro_b3t_template.csv; with the interface in Spanish, …_plantilla.csv):

parcela,estrato,especie,familia,dap,altura,copa,estado,x,y
P01,,Inga vera,Fabaceae,29.7,14.6,6.1,alive,-7.1,2.6
P01,,Inga vera,Fabaceae,38.2,12.8,7.6,alive,-0.2,7.9
P01,,Inga vera,Fabaceae,36.6,14,7.3,alive,-6.6,12.7

The template of the Block 3 trees, with the first three rows of the example coffee farm, with the interface in English (the status of the example trees comes in the interface language: alive, vivo in Spanish; the keys of the header row do not change). Its stratum column is empty because the coffee farm takes its strata from the plot table.

A.3 · What the app writes

All the CSV files use the comma as separator, the decimal point and rows ending in carriage return and line feed, and go in UTF-8 with a byte-order mark, so a spreadsheet reads accented characters without asking. Numbers go with 12 significant digits, not rounded as on screen, and the same in the XLSX workbooks.

A.4 · The project file

It is JSON text, with these fields:

Values that are not finite numbers are written as "__NaN__", "__Inf__" and "__-Inf__", and become so again when the file is opened.

BThe rules, gathered

The default values, limits and rules the app applies to compute, classify or warn, with the block where they act. They are the same ones each chapter explains; here they are together, to look up at a glance.

B.1 · Conversions and constants of all the blocks

B.2 · Inventory and sampling (Block 3)

B.3 · Biomass and carbon (Block 4)

B.4 · Mensuration, growth and diversity (Blocks 2, 5 and 6)

B.5 · Agroforestry, maps and tools (Blocks 7, 8 and 9)

All are guides for reading the results, not laws: what you know of your forest, of how it was measured and of what your management plan or carbon standard asks for always has the last word.

CGlossary

The forestry terms the manual uses, in the sense in which the app uses them, in alphabetical order, with the Spanish term the app shows when its interface is in Spanish.

DWhen something goes wrong

The messages of the app, as they appear in English, with what they usually mean and what to do. Ellipses stand for the part of the message that changes (a name, a number).

EThe tool: license and how to cite it

E.1 · What it is and what it is not

ForestPro is written for this program: the numerical routines, the nonlinear model fitting, the sampling estimators, biomass and its uncertainty, the diversity indices, the variogram and kriging, the CSV, XLSX and ODS reader, the XLSX and ZIP writers, the figures, the map, the report and the illustrations are in its own code, which can be read. It uses no third-party libraries, has no minified code or build steps, brings no fonts (it uses the computer’s) and loads nothing from the network: that is why it opens with a double click and without a connection. The example data sets of the coffee farm, the cloud forest and the pine plantation are simulated with a script of its own; they are not measurements from any site.

It implements published methods—allometric equations, root:shoot ratios, volume, taper, height–diameter and growth models, sampling estimators, diversity and structure indices, log rules, financial formulas—each one cited to its source in the “Method and sources” box of its card and in Appendix F. What comes from third parties is little, and the file LICENSES-TERCEROS.md declares it with the text of its licenses:

PROCEDENCIA.md says where each part of the program comes from, so that anyone can check it. The common modules of the suite (the bar, the navigator, the figure studio, the ✎ editor and the effects) are by the same author.

E.2 · License

ForestPro is distributed under the GNU General Public License, version 3 or later (GPL-3.0-or-later). The full text is in the LICENSE file, in the program folder.

E.3 · How to cite it

The How to cite section of the app’s home page and that of Block 10 give the citation ready to copy, as text and as a BibTeX entry (Section 10.4).

@software{BarreraGuzman2026ForestPro,
  author  = {Barrera-Guzmán, Luis Ángel},
  title   = {{ForestPro}: a browser-based platform for forest inventory, mensuration, biomass, carbon and agroforestry},
  year    = {2026},
  version = {1.0.0},
  license = {GPL-3.0-or-later},
  doi     = {10.5281/zenodo.23876374},
  url     = {https://doi.org/10.5281/zenodo.23876374}
}

Concept DOI, which always leads to the latest version: 10.5281/zenodo.23876374; each version also has its own on Zenodo (version 1.0.0: 10.5281/zenodo.23876375). Code: https://github.com/luisangelbg/ForestPro · online app: https://luisangelbg.github.io/ForestPro/.

Cite also, in your methods, the works behind the methods you used: the allometric equations and the density database, the sampling design and its estimators, the site index curves, the diversity indices. The “Method and sources” box of each card names them, and the report gathers them if you leave With the methods and their sources checked; if you used the density table, cite Zanne et al. (2009).

FReferences

The 103 works the app cites in the “Method and sources” boxes of Blocks 2 to 9 and in the sources of its catalog of allometric equations, grouped by block in the order of the app and, within each block, in alphabetical order. A work cited in several blocks appears only in the first one that cites it.

F.1 · Block 2 · Tree mensuration

• Avery, T. E. and Burkhart, H. E. (2002). Forest Measurements, 5th ed. McGraw-Hill.
• Furnival, G. M. (1961). An index for comparing equations used in constructing volume tables. Forest Science 7:337–341.
• Husch, B., Beers, T. W. and Kershaw, J. A. (2003). Forest Mensuration, ch. 12–13. Wiley.
• Kozak, A., Munro, D. D. and Smith, J. H. G. (1969). Taper functions and their application in forest inventory. For. Chron. 45:278–283.
• Marquardt, D. W. (1963). An algorithm for least-squares estimation of nonlinear parameters. J. SIAM 11:431–441.
• Max, T. A. and Burkhart, H. E. (1976). Segmented polynomial regression applied to taper equations. Forest Science 22:283–289.
• Mehtätalo, L., de-Miguel, S. and Gregoire, T. G. (2015). Modeling height-diameter curves for prediction. Can. J. For. Res. 45:826–837.
• Meyer, H. A. (1941). A correction for a systematic error occurring in the application of the logarithmic volume equation. Pennsylvania State Forest School Res. Paper 7.
• Näslund, M. (1937). Skogsförsöksanstaltens gallringsförsök i tallskog. Medd. Statens Skogsförsöksanstalt 29.
• Richards, F. J. (1959). A flexible growth function for empirical use. J. Exp. Bot. 10:290–300.
• Schumacher, F. X. and Hall, F. S. (1933). Logarithmic expression of timber-tree volume. J. Agric. Res. 47:719–734.
• Smalian, H. L. (1837). Beitrag zur Holzmeßkunst. Stralsund.
• Spurr, S. H. (1952). Forest Inventory. Ronald Press.
• Yang, R. C., Kozak, A. and Smith, J. H. G. (1978). The potential of Weibull-type functions as flexible growth curves. Can. J. For. Res. 8:424–431.

F.2 · Block 3 · Inventory and sampling

• Assmann, E. (1970). The Principles of Forest Yield Study. Pergamon.
• Bailey, R. L. and Dell, T. R. (1973). Quantifying diameter distributions with the Weibull function. Forest Science 19:97–104.
• Bitterlich, W. (1948). Die Winkelzählprobe. Allgemeine Forst- und Holzwirtschaftliche Zeitung 59:4–5.
• Cochran, W. G. (1977). Sampling Techniques, §4.6 and §5.5. Wiley.
• Curtis, J. T. and McIntosh, R. P. (1951). An upland forest continuum in the prairie-forest border region of Wisconsin. Ecology 32:476–496.
• Grosenbaugh, L. R. (1952). Plotless timber estimates — new, fast, easy. Journal of Forestry 50:32–37.
• Hegyi, F. (1974). A simulation model for managing jack-pine stands. In Growth Models for Tree and Stand Simulation, pp. 74–90. Royal College of Forestry.
• Reineke, L. H. (1933). Perfecting a stand-density index for even-aged forests. J. Agric. Res. 46:627–638.
• Satterthwaite, F. E. (1946). An approximate distribution of estimates of variance components. Biometrics Bulletin 2:110–114.
• Shiver, B. D. and Borders, B. E. (1996). Sampling Techniques for Forest Resource Inventory. Wiley.
• Wilson, F. G. (1946). Numerical expression of stocking in terms of height. Journal of Forestry 44:758–761.

F.3 · Block 4 · Biomass and carbon

• Cairns, M. A., Brown, S., Helmer, E. H. and Baumgardner, G. A. (1997). Root biomass allocation in the world's upland forests. Oecologia 111:1–11.
• Chave, J. et al. (2004). Error propagation and scaling for tropical forest biomass estimates. Phil. Trans. R. Soc. B 359:409–420.
• Chave, J. et al. (2005). Tree allometry and improved estimation of carbon stocks and balance in tropical forests. Oecologia 145:87–99. doi:10.1007/s00442-005-0100-x
• Chave, J. et al. (2009). Towards a worldwide wood economics spectrum. Ecology Letters 12:351–366.
• Chave, J. et al. (2014). Improved allometric models to estimate the aboveground biomass of tropical trees. Global Change Biology 20:3177–3190. doi:10.1111/gcb.12629
• IPCC (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories, vol. 4, ch. 4 (tables 4.3 and 4.4).
• Pearson, T. R. H., Brown, S. L. and Birdsey, R. A. (2007). Measurement Guidelines for the Sequestration of Forest Carbon. USDA Forest Service, GTR NRS-18.
• Réjou-Méchain, M. et al. (2017). BIOMASS: an R package for estimating above-ground biomass and its uncertainty in tropical forests. Methods in Ecology and Evolution 8:1163–1167.
• Van Wagner, C. E. (1968). The line intersect method in forest fuel sampling. Forest Science 14:20–26.
• Welch, B. L. (1947). The generalization of “Student’s” problem when several different population variances are involved. Biometrika 34:28–35.
• Zanne, A. E. et al. (2009). Data from: Towards a worldwide wood economics spectrum. Dryad, doi:10.5061/dryad.234.

F.4 · Block 5 · Growth and yield

• Bailey, R. L. and Clutter, J. L. (1974). Base-age invariant polymorphic site curves. Forest Science 20:155–159.
• Campos, J. C. C. and Leite, H. G. (2017). Mensuração Florestal, 5th ed. UFV.
• Cieszewski, C. J. and Bailey, R. L. (2000). Generalized algebraic difference approach. Forest Science 46:116–126.
• Clutter, J. L. (1963). Compatible growth and yield models for loblolly pine. Forest Science 9:354–371.
• Clutter, J. L., Fortson, J. C., Pienaar, L. V., Brister, G. H. and Bailey, R. L. (1983). Timber Management: a Quantitative Approach. Wiley.
• Davis, L. S., Johnson, K. N., Bettinger, P. S. and Howard, T. E. (2001). Forest Management, 4th ed. McGraw-Hill.
• Gompertz, B. (1825). On the nature of the function expressive of the law of human mortality. Phil. Trans. R. Soc. 115:513–583.
• Schumacher, F. X. (1939). A new growth curve and its application to timber yield studies. Journal of Forestry 37:819–820.
• Theil, H. (1953). Repeated least squares applied to complete equation systems. Central Planning Bureau, The Hague.
• Zeide, B. (1993). Analysis of growth equations. Forest Science 39:594–616.

F.5 · Block 6 · Diversity and ecology

• Baddeley, A., Rubak, E. and Turner, R. (2015). Spatial Point Patterns: Methodology and Applications with R. CRC Press.
• Chao, A. (1987). Estimating the population size for capture-recapture data with unequal catchability. Biometrics 43:783–791.
• Chao, A. and Jost, L. (2012). Coverage-based rarefaction and extrapolation. Ecology 93:2533–2547.
• Chao, A. and Lee, S.-M. (1992). Estimating the number of classes via sample coverage. JASA 87:210–217.
• Chao, A. et al. (2014). Rarefaction and extrapolation with Hill numbers. Ecological Monographs 84:45–67.
• Chao, A., Wang, Y. T. and Jost, L. (2013). Entropy and the species accumulation curve. Methods Ecol. Evol. 4:1091–1100.
• Clark, P. J. and Evans, F. C. (1954). Distance to nearest neighbor as a measure of spatial relationships in populations. Ecology 35:445–453.
• Colwell, R. K., Mao, C. X. and Chang, J. (2004). Interpolating, extrapolating, and comparing incidence-based species accumulation curves. Ecology 85:2717–2727.
• de Leeuw, J. and Mair, P. (2009). Multidimensional scaling using majorization: SMACOF in R. J. Stat. Softw. 31(3).
• Donnelly, K. P. (1978). Simulations to determine the variance and edge effect of total nearest neighbor distance. In Simulation Methods in Archaeology, pp. 91–95. Cambridge University Press.
• Fisher, R. A., Corbet, A. S. and Williams, C. B. (1943). The relation between the number of species and the number of individuals. J. Anim. Ecol. 12:42–58.
• Hill, M. O. (1973). Diversity and evenness: a unifying notation and its consequences. Ecology 54:427–432.
• Hurlbert, S. H. (1971). The nonconcept of species diversity. Ecology 52:577–586.
• Kruskal, J. B. (1964). Nonmetric multidimensional scaling: a numerical method. Psychometrika 29:115–129.
• Legendre, P. and Legendre, L. (2012). Numerical Ecology, 3rd ed. Elsevier.
• Magurran, A. E. (2004). Measuring Biological Diversity. Blackwell.
• Ohser, J. (1983). On estimators for the reduced second moment measure of point processes. Math. Operationsforsch. Statist. 14:63–71.
• Pretzsch, H. (1996). Strukturvielfalt als Ergebnis waldbaulichen Handelns. DVFFA, Sektion Ertragskunde, pp. 134–154.
• Pretzsch, H. (2009). Forest Dynamics, Growth and Yield. Springer.
• Ripley, B. D. (1977). Modeling spatial patterns. J. R. Stat. Soc. B 39:172–212.
• Sokal, R. R. and Rohlf, F. J. (1962). The comparison of dendrograms by objective methods. Taxon 11:33–40.
• Souza, D. R. et al. (2003). Análise estrutural em floresta ombrófila densa de terra firme. Revista Árvore 27:75–83.

F.6 · Block 7 · Agroforestry systems

• Crookston, N. L. and Stage, A. R. (1999). Percent Canopy Cover and Stand Structure Statistics from the Forest Vegetation Simulator. USDA Forest Service, GTR RMRS-24.
• Evans, J. and Turnbull, J. W. (2004). Plantation Forestry in the Tropics, 3rd ed. Oxford University Press.
• Frazer, G. W., Canham, C. D. and Lertzman, K. P. (1999). Gap Light Analyzer (GLA) 2.0: users manual. Simon Fraser University.
• Holdridge, L. R., Grenke, W. C., Hatheway, W. H., Liang, T. and Tosi, J. A. (1971). Forest Environments in Tropical Life Zones: a Pilot Study. Pergamon.
• Jonckheere, I. et al. (2004). Review of methods for in situ leaf area index determination. Part I. Agric. For. Meteorol. 121:19–35.
• Moguel, P. and Toledo, V. M. (1999). Biodiversity conservation in traditional coffee systems of Mexico. Conservation Biology 13:11–21.
• Nair, P. K. R. (1993). An Introduction to Agroforestry. Kluwer.
• Otsu, N. (1979). A threshold selection method from gray-level histograms. IEEE Trans. Syst. Man Cybern. 9:62–66.
• Segura, M., Kanninen, M. and Suárez, D. (2006). Allometric models for estimating aboveground biomass of shade trees and coffee bushes grown together. Agroforestry Systems 68:143–150. doi:10.1007/s10457-006-9005-x
• Soto-Pinto, L., Anzueto, M., Mendoza, J., Jiménez-Ferrer, G. and de Jong, B. (2010). Carbon sequestration through agroforestry in indigenous communities of Chiapas, Mexico. Agroforestry Systems 78:39–51.
• Soto-Pinto, L., Perfecto, I., Castillo-Hernández, J. and Caballero-Nieto, J. (2000). Shade effect on coffee production at the northern Tzeltal zone of the state of Chiapas, Mexico. Agric. Ecosyst. Environ. 80:61–69.
• Weisberg, S. (2014). Applied Linear Regression, 4th ed., sec. 7.6 (delta method). Wiley.

F.7 · Block 8 · Maps

• Cressie, N. (1985). Fitting variogram models by weighted least squares. Mathematical Geology 17:563–586.
• Gregoire, T. G. and Valentine, H. T. (2008). Sampling Strategies for Natural Resources and the Environment. Chapman & Hall/CRC.
• Isaaks, E. H. and Srivastava, R. M. (1989). An Introduction to Applied Geostatistics. Oxford University Press.
• Karney, C. F. F. (2011). Transverse Mercator with an accuracy of a few nanometers. Journal of Geodesy 85:475–485.
• Matheron, G. (1963). Principles of geostatistics. Economic Geology 58:1246–1266.
• OpenStreetMap Foundation. Tile usage policy and attribution: openstreetmap.org/copyright.
• Shepard, D. (1968). A two-dimensional interpolation function for irregularly-spaced data. Proc. ACM National Conference, pp. 517–524.
• Snyder, J. P. (1987). Map Projections — A Working Manual. USGS Professional Paper 1395.
• Todhunter, I. (1886). Spherical Trigonometry, 5th ed. Macmillan (L'Huilier's theorem).

F.8 · Block 9 · Forestry tools

• Bettinger, P., Boston, K., Siry, J. P. and Grebner, D. L. (2017). Forest Management and Planning, 2nd ed. Academic Press.
• Bruce, D. (1925). A formula for the Scribner rule. Journal of Forestry 23:432–433.
• Clark, J. F. (1906). The measurement of saw logs. Forestry Quarterly 4:79–93.
• Faustmann, M. (1849). Berechnung des Werthes, welchen Waldboden sowie noch nicht haubare Holzbestände für die Waldwirthschaft besitzen. Allgemeine Forst- und Jagd-Zeitung 25:441–455.
• Klemperer, W. D. (1996). Forest Resource Economics and Finance. McGraw-Hill.
• NIST (2008). Guide for the Use of the International System of Units (SI). Special Publication 811, appendix B.
• Prodan, M., Peters, R., Cox, F. and Real, P. (1997). Mensura Forestal. IICA and GTZ, San José.
• Steele, P. H. (1984). Factors Determining Lumber Recovery in Sawmilling. USDA Forest Service, Gen. Tech. Rep. FPL-39.

F.9 · The allometric equations of the catalogue (Block 4)

• Brown, S. (1997). Estimating Biomass and Biomass Change of Tropical Forests: a Primer. FAO Forestry Paper 134, eq. 3.2.1.
• Díaz-Franco, R. et al. (2007). Determinación de ecuaciones alométricas para estimar biomasa y carbono en Pinus patula. Madera y Bosques 13(1):25–34. doi:10.21829/myb.2007.1311233
• Figueroa-Navarro, C. M. et al. (2010). Estimación de la biomasa en un bosque bajo manejo de Pinus patula. Rev. Mex. Cienc. For. 1(1):105–112.
• Jenkins, J. C. et al. (2003). National-scale biomass estimators for United States tree species. Forest Science 49:12–35. doi:10.1093/forestscience/49.1.12
• Segura, M. et al. (2006). Agroforestry Systems 68:143–150 (d₁₅: stem diameter 15 cm above the ground).
• van Noordwijk, M. et al. (2002). Carbon stock assessment for a forest-to-coffee conversion landscape in Sumber-Jaya. Science in China C 45 (suppl.):75–86. The source does not say at what height D is measured.