merge.data.BCI.R

### MERGE BCI DATA
### Edited by FH
rm(list = ls()); source("./R/format.function.R"); library(reshape)

#########################
## READ DATA
####################
### read individuals tree data
## Requires careful formatting of 7 census datasets
## The raw data is such that, once a tree dies in census X, then it no longer exists in census X+1, X+2 etc...
data.bci1 <- read.table("./data/raw/DataBCI/census1/PlotsDataReport.txt",header=TRUE,stringsAsFactors=FALSE,sep = "\t")
big.bci <- NULL
for(k in 2:7) {
	new.directory <- paste("./data/raw/DataBCI/census",k,"/PlotsDataReport.txt",sep="")
	data.bci2 <- read.table(new.directory,header=TRUE,stringsAsFactors=FALSE,sep = "\t")
	if(!is.null(big.bci)) {
		sub.bci <- merge(data.bci1[,c(2:7,11,13)], data.frame(TreeID = data.bci2[["TreeID"]], DBH2 = data.bci2[["DBH"]], Date2 = data.bci2[["Date"]], dead = as.numeric(data.bci2[["Status"]] == "dead")),sort = T, by = "TreeID")
		big.bci <- rbind(big.bci, sub.bci) }
	if(is.null(big.bci)) {
		big.bci <- merge(data.bci1[,c(2:7,11,13)], data.frame(TreeID = data.bci2[["TreeID"]], DBH2 = data.bci2[["DBH"]], Date2 = data.bci2[["Date"]], dead = as.numeric(data.bci2[["Status"]] == "dead")), 
			sort = T, by = "TreeID") }
	data.bci1 <- data.bci2 
	cat("Census", k, "now included\n") }
rm(data.bci1, data.bci2, data.bci3, sub.bci)
big.bci <- big.bci[order(big.bci$TreeID),]; colnames(big.bci)[c(7:8)] <- c("DBH1","Date1")
data.bci <- big.bci; rm(big.bci)

### read species names
species.clean <- read.table("./data/raw/DataBCI/TaxonomyDataReport.txt",stringsAsFactors=FALSE, header = T, sep = "\t")

## Try to relate SpeciesID in species.clean species names in data.bci
data.bci$sp <- rep(NA,nrow(data.bci))
for(k in 1:nrow(species.clean)) {
	latin.name <- apply(species.clean[k,2:3],1,paste,collapse=" ")[1]
	data.bci$sp[which(data.bci$Latin == latin.name)] <- species.clean$SpeciesID[k] }
	
length(unique(data.bci$sp))

######################################
## MASSAGE TRAIT DATA
############################
## THERE ARE HEIGHT VARAIBLES IN THE TRAITS DATA, BUT WE WILL NEED TO DISCUSS WHICH ONES TO USE

##########################################
## FORMAT INDIVIDUAL TREE DATA
#############
data.bci$Date1 <- as.Date(data.bci$Date1); data.bci$Date2 <- as.Date(data.bci$Date2)
#data.bci$yr1 <- format(strptime(data.bci$Date1, format = "%Y-%m-%d"),"%Y")
#data.bci$yr2 <- format(strptime(data.bci$Date2, format = "%Y-%m-%d"),"%Y")
data.bci$year <- as.numeric(difftime(data.bci$Date2, data.bci$Date1, units = "weeks")/52) ## Not rounded 

## change unit and names of variables to be the same in all data for the tree 
data.bci$G <- 10*(data.bci$DBH1-data.bci$DBH1)/data.bci$year ## diameter growth in mm per year - BASED ON UNROUNDED YEARS
data.bci$D <- data.bci[["DBH1"]]; 
data.bci$plot <- data.bci[["Quadrat"]] ## plot code?
data.bci$htot <- rep(NA,length(data.bci[["G"]])) ## DOES IT COME FROM THE TRAITS DATA?
data.bci$sp.name <- data.bci$Latin

######################
## ECOREGION
###################
## bci has only 1 eco-region?

######################
## PERCENT DEAD
###################
## variable percent dead/cannot do with since dead variable is missing
## compute numer of dead per plot to remove plot with disturbance

function.perc.dead2 <- function(dead) { out <- sum(dead,na.rm=T)/length(dead[!is.na(dead)]); if(!is.finite(out)) out <- NA; return(out) }