4.1.1 Custom Data
To build an accurate model, good data is essential. The
forestat package includes a cleaned sample data
set that can be loaded and viewed using the following command:
# Load the forestData sample data included in the package
data("forestData")
# Or read the forestat.csv sample data included in the package
forestData <- read.csv(system.file("extdata", "forestData.csv", package = "forestat"))
# Select the ID, code, AGE, H, S, BA, and Bio fields from the forestData sample data
# and view the first 6 rows of data
head(dplyr::select(forestData, ID, code, AGE, H, S, BA, Bio))
# Output
ID code AGE H S BA Bio
1 6100005337 1 45 11.9 1508.468 50.13462 474.4957
2 410001607 1 42 16.7 1490.493 47.22381 444.5069
3 6100005337 1 35 11.0 1401.944 46.64877 435.8741
4 6100005337 1 40 12.8 1303.489 44.15220 415.9098
5 410001607 1 38 15.2 1350.941 42.37152 400.3925
6 6220002848 1 88 11.2 1631.235 50.43886 395.2503
Of course, you can also choose to load custom data:
# Load custom data
forestData <- read.csv("/path/to/your/folder/your_file.csv")
The custom data is required to be in csv format, and the
ID (sample plot ID),
code (forest type code of sample plot),
AGE (the average age of the stand), and
H (the average height of the stand) fields are required to
build the H Model and plot the relevant example graphs.
The S (stand density index),
BA (stand basal area), and Bio (stand biomass)
fields are optional and are used to build the BA Model and
Bio Model.
In the subsequent calculation of potential productivity and actual
productivity, the BA Model and Bio Model are
required. That is, if the custom data lacks the S,
BA, and Bio fields, potential productivity and
actual productivity cannot be calculated.
Figure 2. Custom data format requirements
4.1.2 Build Stand Growth Model
After the data is loaded, forestat will use the
class.plot() function to build a stand growth model. If the
custom data contains the ID, code, AGE, H, S, BA, Bio
fields, the H Model, BA Model, and
Bio Model will be built simultaneously. If only the
ID, code, AGE, H fields are included, only the
H Model will be built.
# Use the Richards model to build a stand growth model
# interval = 5 indicates that the initial stand age interval for tree height classification is set to 5, and number = 5 indicates that the maximum number of initial tree height classifications is 5
# The initial parameters for fitting the model are a = 19, b = 0.1, and c = 0.8
forestData <- class.plot(forestData, model = "Richards",
interval = 5, number = 5, a = 19, b = 0.1, c = 0.8)
The model is the model used to build the
H Model and can be selected from the
"Logistic", "Richards", "Korf",
"Gompertz", "Weibull", and
"Schumacher" models. The BA Model and
Bio Model are built using the Richard model by default.
interval is the initial stand age interval for tree height
classification, and number is the maximum number of initial
tree height classifications. a, b, c are the initial
parameters for fitting the model. When fitting errors occur, try
multiple initial parameters as attempts.
The result returned by the class.plot() function is the
forestData object, which includes
Input (input data and tree height classification results),
H model (tree height model),
BA model (stand basal area growth model),
Bio model (stand biomass growth model), and
output (model parameters).
Figure 3. Structure of the forestData object
4.1.3 Obtaining Summary Data
To understand the establishment of the model, you can use the
summary(forestData) function to obtain the summary data of
the forestData object. The function returns the
summary.forestData object and outputs the relevant data to
the screen.
The first paragraph of the output is the summary of the input data,
and the second, third, and fourth paragraphs are the parameters and
concise reports of the H model (tree height model),
BA model (basal area growth model), and
Bio model (stocking growth model), respectively.
# Output
# First paragraph
H S BA Bio
Min. : 2.00 Min. : 15.94 Min. : 0.3017 Min. : 1.224
1st Qu.: 7.70 1st Qu.: 360.38 1st Qu.: 9.3241 1st Qu.: 53.233
Median : 9.40 Median : 557.25 Median :14.9777 Median : 95.002
Mean : 9.83 Mean : 583.88 Mean :16.2648 Mean :109.322
3rd Qu.:11.90 3rd Qu.: 764.38 3rd Qu.:21.6455 3rd Qu.:147.737
Max. :17.70 Max. :1772.26 Max. :52.6455 Max. :474.496
# Second paragraph
Hmodel Parameters:
Nonlinear mixed-effects model fit by maximum likelihood
Model: H ~ 1.3 + a * (1 - exp(-b * AGE))^c
Data: data
AIC BIC logLik
2720.209 2746.159 -1355.105
Random effects:
Formula: a ~ 1 | LASTGROUP
a Residual
StdDev: 3.6513 0.6616545
Fixed effects: a + b + c ~ 1
Value Std.Error DF t-value p-value
a 11.226213 1.6509803 1319 6.799726 0
b 0.020457 0.0029541 1319 6.924853 0
c 0.370395 0.0228807 1319 16.188147 0
Correlation:
a b
b -0.137
c -0.122 0.949
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-4.13170023 -0.75823758 -0.03968202 0.74727148 4.97834758
Number of Observations: 1326
Number of Groups: 5
Concise Parameter Report:
Model Coefficients:
a1 a2 a3 a4 a5 b c
6.331338 8.578689 10.91438 13.61481 16.69184 0.0204566 0.3703953
Model Evaluations:
pe RMSE R2 Var TRE AIC BIC
-0.001864527 0.6604061 0.9455896 0.4364619 0.4185215 2720.209 2746.159
logLik
-1355.105
Model Formulas:
Func Spe
model1:H ~ 1.3 + a * (1 - exp(-b * AGE))^c model1:pdDiag(a ~ 1)
# Third paragraph (similar data format to the second paragraph)
BAmodel Parameters:
# Omitted here
......
# Fourth paragraph (similar data format to the second paragraph)
Biomodel Parameters:
# Omitted here
......
