In VincyaneBadouard/TreeData: Forest Inventories Harmonization & Correction

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>", echo = T, message = T, warning = F, cache = T
)

Load TreeData and datasets

Install TreeData

devtools::install_github("VincyaneBadouard/TreeData", build_vignettes = TRUE)

Load the packages

library(TreeData)
library(knitr)
library(kableExtra)
library(shiny)
library(shinydashboard)
library(shinyjs)
library(shinyWidgets)
library(data.table)
library(ggplot2)

Open the shiny app

# TreeData::RunApp() # doest' work, why?
shiny::runGitHub( "VincyaneBadouard/TreeData", subdir = "inst/app") # data.tree, stringdist

Load the example dataset stored in the package

data("TestData")

Metadata of the output columns of the TreeData package

• Comment (character): error type informations

Bota
- Family_TreeDataCor (character): corrected Family name
- FamilyCorSource (character): source of the Family correction
- Genus_TreeDataCor (character): corrected Genus name
- Species_TreeDataCor (character): corrected Species name
- BotanicalCorrectionSource (character): source of the Genus and Species correction
- ScientificName_TreeDataCor (character): corrected ScientificName
- VernName_TreeDataCor (character): completed if information available at IdTree level.

Life Status
- LifeStatus_TreeDataCor (logical): corrected stem life status.

Taper
- TaperDBH_TreeDataCor (numeric): corrected trees diameter at default HOM

Diameter
- Diameter_TreeDataCor (numeric): corrected trees diameter at default HOM
- DiameterCorrectionMeth (character): diameter correction methode = "taper"/"local linear regression"/"weighted mean" /phylogenetic hierarchical("species"/"genus"/"family"/"stand")/"shift realignment"/"Same value".
- POM_TreeDataCor (factor): POM value at which the corrected diameters are proposed. Corresponds to the 1st POM value at which the stem was measured
- HOM_TreeDataCor (numeric): HOM value at which the corrected diameters are proposed. Corresponds to the DefaultHOM if Taper correction applied. If not, corresponds to the 1st HOM value at which the stem was measured

Recruitment
- CorrectedRecruit (logical): TRUE: the row was added to represent the year when the stem was larger than the minimum diameter, but absent from the dataset. FALSE: original dataset row.

All the correction functions of the package

GeneralErrorsDetection()
BotanicalCorrection()
StatusCorrection()
TaperCorrection()
DiameterCorrection()
RecruitmentCorrection()

The corrections illustration functions of the package

StatusCorrectionPlot()
DiameterCorrectionPlot()
BotanicalCorrectionPlot()

Arguments presentation

Decomposed corrections

General errors detection

Detect errors - Remove duplicated rows - Check missing value in X-YTreeUTM/PlotArea/Plot/Subplot/Year/TreeFieldNum/ IdTree/IdStem/Diameter/POM/HOM/Family/Genus/Species/VernName - Check missing value (NA/0) in the measurement variables: "Diameter", "HOM", "TreeHeight", "StemHeight" - Check of the unique association of the IdTree with plot, subplot and TreeFieldNum (at the site scale) - Check duplicated IdTree/IdStem in a census (at the site scale) - Check for trees outside the subplot (not implemented yet) - Check invariant coordinates per IdTree/IdStem - Check fix Plot and Subplot number (not implemented yet)

Rslt <- GeneralErrorsDetection(TestData)

# The detected errors
unique(Rslt[Comment != "", Comment])

Rslt[Comment != "", ]

Botanical identification correction

• No special characters (typography)
• No family name in the Genus and Species columns (the suffix "aceae" is specific to the family name.
• Correct spelling of botanical names (Taxonstand or WorldFlora)
• Family & Scientific names match (BIOMASS::getTaxonomy or WorldFlora)
• Update the scientific botanical names with the current phylogenetic classification
• Check invariant botanical informations per IdTree (1 IdTree = 1 family, 1 scientific and 1 vernacular name)

Rslt <- BotanicalCorrection(TestData, Source = "TPL", WFOData = NULL)

Life status

• if UseSize argument chosen : if Diameter != NA -> Alive If (the value in bold is modified by the value given after the arrow): (the ">" gives the chronological order of the sequence)

• Dead > Alive -> NA

• add rows for the forgotten censuses between 2 'Alive' if chosen
• Alive > Dead/NA > Alive -> Alive
• Alive > NA > Dead -> NA
• Alive > Dead > NA -> Dead
• Alive > NA > NA: if DeathConfirmation > unseens -> NA if DeathConfirmation =< unseens -> Dead

Status_corr <- StatusCorrection(TestData,
                                InvariantColumns = c("Site",
                                                     "Genus",
                                                     "Species",
                                                     "Family",
                                                     "ScientificName"),
                                DeathConfirmation = 2,
                                UseSize = FALSE,
                                AddRowsForForgottenCensuses = TRUE,
                                DetectOnly = FALSE,
                                RemoveRBeforeAlive = TRUE,
                                RemoveRAfterDeath = TRUE)

Status_corr[LifeStatus!=LifeStatus_TreeDataCor]

Status_corr[IdStem == "101362_1_auto", .(IdStem, LifeStatus, LifeStatus_TreeDataCor, Diameter, Year, Comment)]
# Les années (Year) sont rangées par ordre croissant avant correction donc les NA se retrouvent à la fin. Le NA est entre deux vivants (TRUE) donc corrigé TRUE. La dernière ligne n'apparait pas sur le graphique car Year = NA donc pas plotable.

# pdf("LifeStatusCorrectionPlots.pdf", width = 25, height = 10)
LifeStatusCorrectionPlot(Status_corr, OnlyCorrected = T, SeveralWindows = F)
# dev.off()

Taper correction

Transform the tree diameter measured at a given height into the diameter corresponding to the default measurement height (HOM), using an allometry.

Data <- data.table(IdStem = "A",
                   ScientificName = "Tree",
                   Year = c(1998, 2008, 2016, 2017, 2018, 2019, 2021),
                   IdCensus = factor(c(1998, 2008, 2016, 2017, 2018, 2019, 2021), ordered = TRUE),
                   Diameter = c(19, 19, 21.4, 22.6, 23.1, 23.1, 23.6),
                   HOM = c(1.30, 3.25, 3.25, 3.25, 3.25, 3.25, 3.25))

Rslt <- TaperCorrection(Data, DefaultHOM = 1.3,
                        TaperParameter = function(DAB, HOM) 0.156 - 0.023 * log(DAB) - 0.021 * log(HOM),

                        TaperFormula = function(DAB, HOM, TaperParameter, DefaultHOM) DAB / (exp(- TaperParameter*(HOM - DefaultHOM))))

DiameterCorrectionPlot(Rslt, CorCol = "TaperDBH_TreeDataCor")

Rslt <- TaperCorrection(Status_corr, DefaultHOM = 1.3,
                        TaperParameter = function(DAB, HOM) 0.156 - 0.023 * log(DAB) - 0.021 * log(HOM),

                        TaperFormula = function(DAB, HOM, TaperParameter, DefaultHOM) DAB / (exp(- TaperParameter*(HOM - DefaultHOM))))

DiameterCorrectionPlot(Rslt, CorCol = "TaperDBH_TreeDataCor", OnlyCorrected = TRUE, SeveralWindows = T)

Diameter correction

Diam_corr <- DiameterCorrection(
  Status_corr,
  KeepMeas = c("MaxHOM", "MaxDate"),

  DefaultHOM = 1.3,
  MaxDBH = 500,
  PositiveGrowthThreshold = 5,
  NegativeGrowthThreshold = -2,

  Pioneers = NULL,
  PioneersGrowthThreshold = 7.5,

  WhatToCorrect = c("POM change", "punctual", "shift"),
  CorrectionType = "individual",

  Digits = 1L,

  DBHCorForDeadTrees = FALSE,

  coef = 0.9,

  DetectOnly = FALSE)

# pdf("DiameterCorrectionPlots_TestData2.pdf", width = 25, height = 10)
DiameterCorrectionPlot(Diam_corr, OnlyCorrected = TRUE, SeveralWindows = F)
# dev.off()

Recruitment

Rslt <- RecruitmentCorrection(Diam_corr,
                              KeepMeas = c("MaxHOM", "MaxDate"),
                              MinDBH = 10,
                              PositiveGrowthThreshold = 5,
                              InvariantColumns = c("Site",
                                                   "Genus_TreeDataCor",
                                                   "Species_TreeDataCor",
                                                   "Family_TreeDataCor",
                                                   "ScientificName_TreeDataCor"),
                              DetectOnly = FALSE)

IdCorr <- Rslt[CorrectedRecruit %in%  TRUE, IdStem]
TreesCorr <- Rslt[IdStem %in% IdCorr, .(IdStem,
                                        Year,
                                        Diameter,
                                        Diameter_TreeDataCor,
                                        CorrectedRecruit)]

ggplot(TreesCorr) +
  aes(x = Year, y = Diameter_TreeDataCor) +
  geom_line(linewidth = 0.5, colour = "#112446") +
  geom_point(shape = "circle", size = 1.5, mapping = aes(color = CorrectedRecruit)) +
  theme_minimal() +
  facet_wrap(vars(IdStem), scales = "free")

Interesting other variables

Mortality rate

Recruitment rate

Growth rate

Basal area

Wood density

Biomass

Tree height

Disturbance intensity

VincyaneBadouard/TreeData documentation built on Jan. 4, 2024, 2:56 a.m.