title: "An introduction to SiTree" author: "Clara Antón Fernández" date: 03-08-2021 output: html_document: toc: true number_sections: true
The sitree package provides a framework to implement Single Tree forest growth models in a fast and memory efficient way. It keep tracks of all alive, dead, and removed trees in a robust and efficient way. SiTree is designed to run single tree simulations where trees can be defined by two time-dependent variables (such as diameter (or basal area), and height), and on time-independent variable, such as tree species. SiTree simulates birth, growth, and death of trees as well as management. Functions can also be defined that affect characteristics of the stand (external modifiers), such as climate change, or fertilization.
The easiest way to start with your own simulation is probably to modify the example functions provided (see the Test Equations vignette).
Two types of input are required by SiTree: tree level and stand level. Tree level information is passed in tree.df
, while stand level information is passed in stand.df
.
tree.df
should be a data frame with four columns named plot.id
, treeid
, dbh
, height
, and tree.sp
, which correspond to a stand/plot ID, a tree ID, diameter, height, and tree species.
Plot and stand data is passed in stand.df
, which should be a data
frame or a list, with at least a column or element named plot.id
which should contain all the plot IDs present in tree.df
. Typical
information provided in stand.df
are plot size, elevation, site
index, plot coordinates, distance to road, temperature or
precipitation.
An example of tree data and stand data are provided.
library(sitree) head(tr) head(fl)
Many of the main functions in a simulation use plot-level variables,
like competition indices (e.g. plot-level basal area). In order to
make the code more transparent, compact, and robust all variables
required in the sub-models that can be estimated from tree, stand and
plot variables are calculated in one place. The fn.prep.common.vars
function.
For example, if diameter increment is estimated as a function of initial diameter, stand basal area and number of trees per ha, both stand basal area variable and number of trees per ha should be calculated in fn.prep.common.vars
. Other typical examples of variables calculated in the fn.prep.common.vars
are top height (the mean height of the trees with the largest diameter in a stand), basal area of larger trees, or tree volume.
The fn.prep.common.vars
function should be provided by the user, so it fits the particular needs of the growth model selected and the data. Calculating the most common variables used in forestry should be straight forward as they are already provided in either the SiTree or the SiTreeE package.
An example of a fn.prep.common.vars
is provided in SiTree.
prep.common.vars.fun
trList
and trListDead
classesIn order to efficiently store the list of all individual live and dead (and removed) trees, two Reference classes (or refclasses) are defined in SiTree. Refclases is chosen instead of S3 or S4 classes because refclasses objects are mutable and the usual R copy on modify semantics do not apply. When simulating for long periods, or for large datasets (e.g. a whole national forest inventory) the risk of running out of memory is not negligible. Using refclasses aim at maintaining the memory needs to the minimum by using mutable objects for storing the larger objects such as represented by the tree lists.
There are two Reference Classes implemented in the sitree package, one for live trees (trList) and other for dead trees (trListDead).
trList This class has two fields, data and nperiods. Under data basic information for each tree is stored (a unique stand ID plot.id, a unique tree ID treeid, and dbh and height for each period, dbh.mm, and height.dm). The nperiods field is an integer that stores the number of periods to be simulated.
trListDead This class extends trList. DBH and heights measured while the tree was alive are stored under the data field. Also under this field information on how long the tree has been in the simulation can be found (yrs.sim). In this class the new field last.measurement stores the dimensions of the tree when it died or was removed. How these dimensions are calculated is defined on the dead.trees.growth function.
Reference Classes objects are mutable, they don't use R's usual copy-on-modify semantics, but are modified in place.
We have provided a function to convert the sitree()
output
containing trList and trListDead* objects to a data frame, the
sitree2dataframe
function. The resulting data frame follows the
usual R copy on modify semantics, and most users might be more
comfortable with it.
result.sitree <- sitree (tree.df = stand.west.tr, stand.df = stand.west.st, functions = list( fn.growth = 'grow.dbhinc.hgtinc', fn.mort = 'mort.B2007', fn.recr = 'recr.BBG2008', fn.management = NULL, fn.tree.removal = NULL, fn.modif = NULL, fn.prep.common.vars = 'prep.common.vars.fun' ), n.periods = 12, period.length = 5, mng.options = NA, print.comments = FALSE, fn.dbh.inc = 'dbhi.BN2009', fn.hgt.inc = 'height.korf' ) str(result.sitree$live) head(sitree2dataframe(result.sitree$live))
The sitree()
function is the core function of the SiTree
package. It is the function that runs the simulations. It requires
tree data (tree.df
), stand/plot data (stand.df
), and a list
of functions to be used in the simulation (functions
), the
number of periods for which to run the simulation (n.periods
),
and the period length (period.length
). Management options can
be passed through the mng.options
argument, and it is also
possible to print comments about the progress of the simulation
selecting print.comments = TRUE
. Additional arguments needed by
the selected functions go into the ellipsis ('...') and can be
retrieved by simply converting it to a list, e.g. arguments <-
list(...)
.
The functions
argument must be a list containing at
least 7 elements:
fn.growth
the name of the function that implements the growth sub-model fn.mort
the name of the function that implements the mortality sub-modelfn.recr
the name of the function that implements the recruitment sub-model: ingrowth, and natural and artificial regenerationfn.management
the name of the function that implements the management (e.g. harvest), can be NULL
fn.tree.removal
the name of the function that implements the management at the tree level, that is the selection of trees to be fell, can be NULL.fn.modif
the name of the function to calculate the effect of external modifiers, can be NULL
fn.prep.common.vars
the name of the function to calculate
auxiliary variables, such as basal area of the stand (see 'The fn.prep.common.vars' subsection above) .Further details on the requirements of the functions listed above can be found under the section "The user-defined functions".
The sitree()
function is a flexible framework for forest growth simulations. Any growth sub-model, mortality sub-model, management, etc. can be used. Some examples are provided in SiTree and in SiTreeE, but generally, the submodels functions need to be provided by the user. The examples provided in SiTree and in SiTreeE can be used as a template. To debug the user-defined functions we suggest to use the provided example as a starting point, set print.comments = TRUE
and switch the submodels functions one by one to test them.
An example of how the list provided in the functions
argument of sitree
should look
like is given below, and further details on each of the functions are provided next.
The fn.growth
function should return a data frame with two
columns giving diameter increment (dbh.inc.mm
) and height increment
(hgt.inc.dm
) of all live trees. This data frame should only contain
numerical data (no missing data allowed). Care must be taken to ensure
that the order matches that of the tree list. Examples of the growth
functions are provided as grow.dbhinc.hgtinc
, dbhi.BN2009
, and height.korf
.
The fn.mort
function should return a TRUE
/FALSE
vector of same
length as the number of trees in the tree list. TRUE
indicates a
tree that will die before the next period, and FALSE
indicates a
tree that will stay alive. An example of a fn.mort
function is
provided in mort.B2007
.
The fn.recr
function is the function that estimates recruitment, the
new trees for the next period. This function should return a list of
new trees (or an empty list if there are no new trees) with elements
plot.id
, treeid
, dbh.mm
, height.dm
, yrs.sim
(indicates when
are the trees incorporated to the plot, for example, in the middle of
the period), and tree.sp
.An example of a fn.recr
function is provided in recr.BBG2008
.
fn.management
is optional. It should return a list, with at least one element
called management
which should be a vector with length equal to
the number of plots in stand.df
. The example we provide uses a
simple code to define management (a five characters string indicating
with a binary code (1 = present, 0 = absent) the treatments to be
executed: harvest-thinning-fertilization-pruning-other), but any other
way to code management can be used, as far as fn.management
returns
a vector. When no management will take place during the simulation
fn.management
can be set to NULL
. An example of a fn.management
function is provided in management.prob
.
fn.tree.removal
is optional. It should return a TRUE
/FALSE
vector indicating
which trees are to be removed. The vector should have the same length
as the number of trees alive at the current period. When no tree
removal will take place during the simulation (no harvest is allowed)
fn.tree.removal
can be set to NULL. An example of a fn.tree.removal
function is provided in mng.tree.removal
.
fn.modif
is a function that can be used to modify characteristics of
the plot or stand, such as site index. This function is optional, and
no example is provided in the current version of the package. It
should return a list with names matching some of those in the
stand.df
data frame. After the external modifiers are calculated
with the function defined as fn.modif
, the elements in the plot data
that matches those of the results of fn.modif
are replaced before
the rest of the simulation continues. For example, if the plot has
been fertilized and we can assume that SI has increased by 2 meters,
the fn.modif
function needs to return a list with a SI element with
all SI as in the plot data frame except for those that have
changed.
fn.prep.common.vars
is the function used to calculate
everything needed for the fn.growth
, fn.mort
, etc to be
calculated. For example, the fn.prep.common.vars
function is
the place to calculate stand competition indices, volume, stand age,
etc. An example is given in the function prep.common.vars.fn
.
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.