#R code for creating ggplots of time series with smooth (GAM) and linear term


######################
#Import Libraries and set working directory
######################
library(zoo)
library(hydroTSM) #you need to install these packages first before you can load them here
library(lubridate)
library(mgcv)
library(ggplot2)
library(gridExtra)
library(scales)
options(scipen=999)
setwd("C:/Users/z5025317/OneDrive - UNSW/WRL_Postdoc_Manual_Backup/WRL_Postdoc/Projects/Paper#1/")
######################

######################
#Set inputs
######################
Case.Study <- "CASESTUDY2"
Estuary <- "HUNTER"
Climvar <- 'tasmean'
ggplotGAM.k <- 14
######################


######################
#Set input file paths
######################
AirT_CSV_Path <- "./Data/SILO/CASESTUDY2/HUNTER/SILO_climdata_HUNTER_Catchment_-32.162479_150.5335812.csv"
RivT_CSV_Path <- "./Data/River_Gauge_Data/HUNTER_210064_20180615/210064_Temp.csv"
SST_CSV_Path <- './Data/NARCLIM_Site_CSVs/CASESTUDY2/HUNTER_32.751_151.690/sstmean_NNRP_HUNTER_33.034_152.156_NARCliM_summary.csv'
######################


######################
#Analyse
######################

############tasmean
#Load a daily (no gaps) time series as a time serie baseline for other time series used here
AirT.df <- data.frame(read.csv(AirT_CSV_Path))
AirT.full.TS <- zoo((AirT.df$max_temp_Celsius + AirT.df$max_temp_Celsius)/2, order.by= as.Date(AirT.df[,"date"], format = "%Y-%m-%d")) #=daily time series of rainfall for creation of clean, daily TS of ET and Q
AirT.TS <- window(AirT.full.TS, start=as.Date("1990-01-01"), end=as.Date("2018-01-01"))
AirT.full.df <- data.frame(AirT.full.TS)
AirT.df <- data.frame(AirT.TS)
colnames(AirT.df) <- 'tasmean'
colnames(AirT.full.df) <- 'tasmean'
############
AirT.df$Julday1 <- seq(1,length(AirT.df[,1]),1)
linear.trend.model_EC_all <- lm(tasmean ~ Julday1, AirT.df)
AirT.pvalNCV_ECall <- summary(linear.trend.model_EC_all )$coefficients[2,4] 
AirT.lintrend <- summary(linear.trend.model_EC_all )$coefficients[2,1] * 356
############
AirT.full.df$Julday1 <- seq(1,length(AirT.full.df[,1]),1)
linear.trend.model_EC_all <- lm(tasmean ~ Julday1, AirT.full.df)
AirT.full.pvalNCV_ECall <- summary(linear.trend.model_EC_all )$coefficients[2,4] 
AirT.full.lintrend <- summary(linear.trend.model_EC_all )$coefficients[2,1] * 356
############tasmean


############River temp
#Load a daily (no gaps) time series as a time serie baseline for other time series used here
RivT.df <- data.frame(read.csv(RivT_CSV_Path))
RivT.full.TS <- zoo(RivT.df$Temp, order.by= as.Date(RivT.df[,"Date"], format = "%d/%m/%Y")) #=daily time series of rainfall for creation of clean, daily TS of ET and Q
RivT.TS <- window(RivT.full.TS, start=as.Date("1995-01-01"), end=as.Date("2018-01-01"))
RivT.full.df <- data.frame(RivT.TS)   ### This is only done because 
RivT.df <- data.frame(RivT.TS)
colnames(RivT.df) <- 'rivTmean'
colnames(RivT.full.df) <- 'rivTmean'
############
RivT.df$Julday1 <- seq(1,length(RivT.df[,1]),1)
linear.trend.model_EC_all <- lm(rivTmean ~ Julday1, RivT.df)
RivT.pvalNCV_ECall <- summary(linear.trend.model_EC_all )$coefficients[2,4] 
RivT.lintrend <- summary(linear.trend.model_EC_all )$coefficients[2,1] * 356
RivT.full.df$Julday1 <- seq(1,length(RivT.full.df[,1]),1)
linear.trend.model_EC_all <- lm(rivTmean ~ Julday1, RivT.full.df)
RivT.full.pvalNCV_ECall <- summary(linear.trend.model_EC_all )$coefficients[2,4] 
RivT.full.lintrend <- summary(linear.trend.model_EC_all )$coefficients[2,1] * 356
############River temp

############ SST
#Load a daily (no gaps) time series as a time serie baseline for other time series used here
SST.df <- data.frame(read.csv(SST_CSV_Path))
SST.full.TS <- zoo(SST.df$NNRP_R1_1950-273.15, order.by= as.Date(SST.df[,"X"], format = "%Y-%m-%d")) #=daily time series of rainfall for creation of clean, daily TS of ET and Q
SST.TS <- window(SST.full.TS, start=as.Date("1990-01-01"), end=as.Date("2018-01-01"))
SST.full.df <- data.frame(SST.full.TS)
SST.df <- data.frame(SST.TS)
str(SST.df)
colnames(SST.df) <- 'SSTmean'
colnames(SST.full.df) <- 'SSTmean'
############
SST.full.df$Julday1 <- seq(1,length(SST.full.df[,1]),1)
linear.trend.model_EC_all <- lm(SSTmean ~ Julday1, SST.full.df)
SST.full.pvalNCV_ECall <- summary(linear.trend.model_EC_all)$coefficients[2,4] 
SST.full.lintrend <- summary(linear.trend.model_EC_all)$coefficients[2,1] * 356
SST.df$Julday1 <- seq(1,length(SST.df[,1]),1)
linear.trend.model_EC_all2 <- lm(SSTmean ~ Julday1, SST.df)
SST.pvalNCV_ECall <- summary(linear.trend.model_EC_all2)$coefficients[2,4] 
SST.lintrend <- summary(linear.trend.model_EC_all2)$coefficients[2,1] * 356
############ SST



######################
#Plot
######################

##################################### Full Time Period
#Air temp Full period 
p1air <- ggplot(AirT.full.df, aes(y=tasmean, x=index(AirT.full.TS))) + geom_line(alpha=0.5) + 
  ggtitle(paste(Estuary, " - Linear and smooth trend in catchment airT (SILO) lin trend was ",
                round(AirT.full.lintrend,3), ' C°/year with p=', round(AirT.full.pvalNCV_ECall,10), sep=" ")) +
  theme(plot.title=element_text(face="bold", size=9)) +
  geom_smooth(method='lm',fill="green", formula=y~x, colour="darkgreen", size = 0.5) +
  stat_smooth(method=gam, formula=y~s(x, k=ggplotGAM.k), se=T, size=0.5) +
  ylab("Air Temperature [C°]") + xlab("Time")

#Riv temp Full period 
p1riv <- ggplot(RivT.full.df, aes(y=rivTmean, x=index(RivT.TS))) + geom_line(alpha=0.5) + 
  ggtitle(paste(Estuary, " - Linear and smooth trend in river temperature (GAUGE) lin trend was ",
                round(RivT.full.lintrend,3), ' C°/year with p=', round(RivT.full.pvalNCV_ECall,10), sep=" ")) +
  theme(plot.title=element_text(face="bold", size=9)) +
  geom_smooth(method='lm',fill="green", formula=y~x, colour="darkgreen", size = 0.5) +
  stat_smooth(method=gam, formula=y~s(x, k=ggplotGAM.k), se=T, size=0.5) +
  ylab("River Temperature [C°]") + xlab("Time")

#Sea temp Full period 
p1sst <- ggplot(SST.full.df, aes(y=SSTmean, x=index(SST.full.TS))) + geom_line(alpha=0.5) + 
  ggtitle(paste(Estuary, " - Linear and smooth trend in sea surface temperature (NNRP) lin trend was ",
                round(SST.full.lintrend,3), ' C°/year with p=', round(SST.full.pvalNCV_ECall,10), sep=" ")) +
  theme(plot.title=element_text(face="bold", size=9)) +
  geom_smooth(method='lm',fill="green", formula=y~x, colour="darkgreen", size = 0.5) +
  stat_smooth(method=gam, formula=y~s(x, k=ggplotGAM.k), se=T, size=0.5) +
  ylab("Sea Surface Temperature [C°]") + xlab("Time")

#export to png
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_tasmean_full_period_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(p1air,ncol=1)
dev.off()
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_rivTmean_full_period_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(p1riv,ncol=1)
dev.off()
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_SSTmean_full_period_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(p1sst,ncol=1)
dev.off()
##################################### Full Time Period




######################### 1990-present
combined.TS <- window(merge(AirT.full.TS, window(RivT.full.TS, start=as.Date("1995-01-01"), end=end(RivT.full.TS)), SST.full.TS, all=T), start=as.Date("1990-01-01"), end=end(AirT.full.TS))
combined.df <- data.frame(combined.TS)
colnames(combined.df) <- c('tasmean','rivTmean', 'SSTmean')

#Air temp 
p1air <- ggplot(combined.df, aes(y=tasmean, x=index(combined.TS))) + geom_line(alpha=0.5) + 
  ggtitle(paste(Estuary, " - Linear and smooth trend in catchment airT (SILO) lin trend was ",
                round(AirT.lintrend,3), ' C°/year with p=', round(AirT.pvalNCV_ECall,10), sep=" ")) +
  theme(plot.title=element_text(face="bold", size=9)) +
  geom_smooth(method='lm',fill="green", formula=y~x, colour="darkgreen", size = 0.5) +
  stat_smooth(method=gam, formula=y~s(x, k=ggplotGAM.k), se=T, size=0.5) +
  ylab("Air Temperature [C°]") + xlab("Time")

#Riv temp 
p1riv <- ggplot(combined.df, aes(y=rivTmean, x=index(combined.TS))) + geom_line(alpha=0.5) + 
  ggtitle(paste(Estuary, " - Linear and smooth trend in river temperature (GAUGE) lin trend was ",
                round(RivT.lintrend,3), ' C°/year with p=', round(RivT.pvalNCV_ECall,10), sep=" ")) +
  theme(plot.title=element_text(face="bold", size=9)) +
  geom_smooth(method='lm',fill="green", formula=y~x, colour="darkgreen", size = 0.5) +
  stat_smooth(method=gam, formula=y~s(x, k=ggplotGAM.k), se=T, size=0.5) +
  ylab("River Temperature [C°]") + xlab("Time")

#Sea temp 
p1sst <- ggplot(combined.df, aes(y=SSTmean, x=index(combined.TS))) + geom_line(alpha=0.5) + 
  ggtitle(paste(Estuary, " - Linear and smooth trend in sea surface temperature (NNRP) lin trend was ",
                round(SST.lintrend,3), ' C°/year with p=', round(SST.pvalNCV_ECall,10), sep=" ")) +
  theme(plot.title=element_text(face="bold", size=9)) +
  geom_smooth(method='lm',fill="green", formula=y~x, colour="darkgreen", size = 0.5) +
  stat_smooth(method=gam, formula=y~s(x, k=ggplotGAM.k), se=T, size=0.5) +
  ylab("Sea Surface Temperature [C°]") + xlab("Time")

#export to png
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_tasmean_1990-present_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(p1air,ncol=1)
dev.off()
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_rivTmean_1990-present_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(p1riv,ncol=1)
dev.off()
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_SSTmean_1990-present_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(p1sst,ncol=1)
dev.off()

#export an overview plot as png
gA <- ggplotGrob(p1air)
gB <- ggplotGrob(p1riv)
gC <- ggplotGrob(p1sst)
gA$widths <- gB$widths
gC$widths <- gB$widths
png.name <- paste('./Output/', Case.Study, '/', Estuary, '/Trends_Summary_1990-present_', Sys.Date(),".png", sep="")
png(file = png.name, width = 10.5, height = 7, units='in', res=500)
grid.arrange(gA,gB ,gC,ncol=1)
dev.off()