• Application of the continuous wavelet transform (CWT), cross wavelet transform (XWT), and wavelet coherence (WTC) to time series data.
  • More information about Wavelet analysis can be found in a paper by A. Grinsted et al. (2004) here
  • So called: Complete time-frequency space
  • Some applications of Wavelet analysis in Atmospheric science can be found in References part
  • Key notes (A. Grinsted et al., 2004):
    • The CWT expands a time series into a time frequency space where oscillations can be seen in a highly intuitive way
    • The XWT exposes regions with high common power and further reveals information about the phase relationship. If the two series are physically related we would expect a consistent or slowly varying phase lag that can be tested against mechanistic models of the physical process
    • WTC can be thought of as the local correlation between two CWTs. In this way locally phase locked behavior is uncovered. The more desirable features of the WTC come at the price of being slightly less localized in time frequency space. The significance level of the WTC has to be determined using Monte Carlo methods

1 Initializing

  • Set up environment
library(tidyverse)      # Core environment for all R related works
library(data.table)     # Data_table control
library(microbenchmark) # To estimate system time
library(RColorBrewer)   # Color control
library(gridExtra)      # Multiple panels plot

library(biwavelet)      # Core package in this analysis


theme_set(theme_bw(base_size = 13, base_family = "Helvetica") +
          theme(plot.title = element_text(size = 14, hjust = 0.0),
                panel.grid.minor = element_blank(),
                panel.grid.major = element_line(colour = "gray", linetype="dashed",size = 0.1),
                legend.position = "none", 
                legend.title = element_blank(),
                legend.text.align = 0))
  • Read input, including JJA 3-hourly 2m Temperature & 2m Specific humidity during a 10-year period
# Reading 2m Temperature
df.tas = fread("data-txt/wavelet_OBS_tas.text", header=FALSE, skip=1,
                col.names=c("date", "time", "tas"))
# Reading 2m Specific humidity 
df.hum = fread("data-txt/wavelet_OBS_huss.text", header=FALSE, skip=1,
                col.names=c("date", "time", "hum"))
  • Merging data into one is a good behaviour when working with “tidyverse”
df = inner_join(df.tas, df.hum) %>%
     mutate(year=substr(date, 1, 4),
            mon =substr(date, 6, 7),
            day =substr(date, 9, 10)) %>% 
     dplyr::select(date, year, mon, day, time, tas, hum)
  • Standardizing the data
df$tas = df$tas - mean(df$tas)
df$hum = df$hum - mean(df$hum)
  • Quick look at the data using ggplot
df %>% 
ggplot() +
  geom_line(aes(x=date, y=tas, group=year), color="orange") +
  geom_line(aes(x=date, y=hum*1000, group=year), color="blue") +
  scale_x_continuous(name="Date",
                     breaks=as.Date(paste0(seq(2000, 2010, 1), "-06-01"))) +
  scale_y_continuous(name="Temperature (°C)", 
                     sec.axis = sec_axis(name="Specific Humidity (g)", ~ . /1000)) +
  labs(title="3-hourly comparison between T and Q => UNCLEAR signal") +
  theme(legend.position="bottom",
        axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
        axis.text.y.left = element_text(color = "orange"),
        axis.title.y.left = element_text(color = "orange"),
        axis.text.y.right = element_text(color = "blue"),
        axis.title.y.right = element_text(color = "blue"))

df.daily =df %>% 
  group_by(date, year, mon, day) %>%
  summarise(tas=mean(tas),
            hum=mean(hum)) %>%
  ungroup()

df.daily %>% 
ggplot() +
  geom_line(aes(x=date, y=tas, group=year), color="orange") +
  geom_line(aes(x=date, y=hum*1000, group=year), color="blue") +
  scale_x_continuous(name="Date",
                     breaks=as.Date(paste0(seq(2000, 2010, 1), "-06-01"))) +
  scale_y_continuous(name="Temperature (°C)", 
                     sec.axis = sec_axis(name="Specific Humidity (g)", ~ . /1000)) +
  labs(title="Daily comparison between T and Q => still UNCLEAR signal") +
  theme(legend.position="bottom",
        axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
        axis.text.y.left = element_text(color = "orange"),
        axis.title.y.left = element_text(color = "orange"),
        axis.text.y.right = element_text(color = "blue"),
        axis.title.y.right = element_text(color = "blue"))

  • A common activity is to find a way to look at the distribution(s) of the data

    p1=df.daily %>%
    ggplot() +
      geom_boxplot(aes(x=year, y=tas), alpha=0., width=0.5) +
      geom_violin(aes(x=year, y=tas), fill="orange", alpha=0.6) +
      scale_x_discrete(breaks=seq(2000, 2010, 1)) +
      labs(title="(a) Boxplot of daily 2m Temperature for each year",
           x="Year", y="Temperature (°C)") +
      theme(legend.position="bottom")
    
    p2=df.daily %>%
    ggplot() +
      geom_boxplot(aes(x=year, y=hum), alpha=0., width=0.5) +
      geom_violin(aes(x=year, y=hum), fill="blue", alpha=0.6) +
      scale_x_discrete(breaks=seq(2000, 2010, 1)) +
      labs(title="(b) Boxplot of daily 2m Specific humidity for each year",
           x="Year", y="Specific humidity (g)") +
      theme(legend.position="bottom")
    
    grid.arrange(p1, p2, ncol=1)

2 Univariate wavelet: Continuous Wavelet Transform (CWT)

  • Used for feature extraction purposes, especially in extracting low s/n ratio signals in time series

  • Discrete Wavelet Transform (DWT) is also existed, but not under interest yet

  • Defining two sets of data (2 variables) with time stamps

df.w.tas = data.frame(
              time=1:((dim(df.daily)[1])),
              val =df.daily %>% dplyr::select(tas) %>% .$tas)
df.w.hum = data.frame(
              time=1:((dim(df.daily)[1])),
              val =df.daily %>% dplyr::select(hum) %>% .$hum)
  • Applying Wavelet Transform to estimate wavelet power spectrum
df.wt.tas = wt(df.w.tas)
df.wt.hum = wt(df.w.hum)
  • Defining a visualize function
plot.wavelet = function(df, type="power", plot.cb=TRUE, plot.phase=FALSE,
                        lty.coi=1, col.coi="grey", lwd.coi=2, lwd.sig=2, arrow.lwd=0.03, arrow.len=0.12,
                        xlab="Time (Date; JJA only", ylab="Period (day)",
                        main="", xaxt="n") {
  plot(df, type=type, plot.cb=plot.cb, plot.phase=plot.phase,
       lty.coi=lty.coi, col.coi=col.coi, lwd.coi=lwd.coi, lwd.sig=lwd.sig,
       xlab=xlab, ylab=ylab, main=main, xaxt=xaxt)
  axis(side=1, at=seq(1, 930, 93), labels=seq(2000, 2009, 1))
}
  • Visualizing Wavelet power. There are several options:
Type Meanings
power Raw power
power.corr Bias-corrected Power
power.norm Power normalized by the variance
power.corr.norm Bias-corrected power normalized by the variance
wavelet Wavelet coefficients
phase Phase
  • Ploting Raw power
png("figures/fig.wl-1.png", width=17, height=6, res=300, units="in")
  par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)

  plot.wavelet(df.wt.tas, type="power.norm", main="(a) Temperature Raw power")
  plot.wavelet(df.wt.hum, type="power.norm", main="(b) Specific humidity Raw power")
dev.off()
## svg 
##   2

  • Ploting Bias-corrected normalized power
png("figures/fig.wl-2.png", width=17, height=6, res=300, units="in")
  par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)

  plot.wavelet(df.wt.tas, type="power.corr.norm", main="(a) Temperature Bias-corrected power")
  plot.wavelet(df.wt.hum, type="power.corr.norm", main="(b) Specific humidity Bias-corrected power")
dev.off()
## svg 
##   2

3 Bivariate Wavelet: Cross Wavelet Transform (XWT)

  • Calculating XWT

    df.xwt=xwt(df.w.tas, df.w.hum)
  • Visualizing

    png("figures/fig.wl-3.png", width=17, height=6, res=300, units="in")
      par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)
    
      plot.wavelet(df.xwt, type="power", plot.phase=TRUE,
                   main="XWT of Temperature & Specific humidity")
      plot.wavelet(df.xwt, type="power.corr.norm", plot.phase=TRUE,
                   main="XWT of Temperature & Specific humidity")
    dev.off()
    ## svg 
    ##   2

  • This results are possibly caused by: 1) Specific humidity is not in form of Normal distribution?; or 2) Temperature and Specific humidity have different characteristics

    p1=df.w.tas %>% ggplot() +
      geom_histogram(aes(x=val, y=..count..), fill="orange", alpha=0.6) +
      labs(title="Temperature distribution", x="Temperature", y="Count")
    p2=df.w.hum %>% ggplot() +
      geom_histogram(aes(x=val, y=..count..), fill="blue", alpha=0.6) +
      labs(title="Specific humidity distribution", x="Specific humidity", y="Count")
    grid.arrange(p1, p2, ncol=2)

    • Solution 1: Try to Sampling a data following Normal distribution based on Specific Humidity data

      # Sample time-series
      n=dim(df.w.tas)[1]
      df.w.tas.sample.1 = data.frame(time=1:n,
                                     val=rnorm(n, mean=mean(df.w.hum$val),
                                                  sd=sd(df.w.hum$val)))
    • Solution 2: Try to simply Adjust Specific Humidity data closer to Temperature (by multiply with 1000)

      # Sample time-series
      df.w.tas.sample.2 = data.frame(time=df.w.hum$time,
                                     val =df.w.hum$val*1000)
    • Visualizing

      p1 = df.w.tas.sample.1 %>% ggplot() +
        geom_histogram(aes(x=val, y=..count..), fill="orange", alpha=0.6) +
        labs(title="Solution 1", x="Sampling data based on Specific Humidity", y="Count")
      
      p2 = df.w.tas.sample.2 %>% ggplot() +
        geom_histogram(aes(x=val, y=..count..), fill="orange", alpha=0.6) +
        labs(title="Solution 2", x="Adjust Specific humidity data", y="Count")
      
      grid.arrange(p1, p2, ncol=2)

  • Performing XWT again & visualizing

    # Cross-wavelet
    df.xwt.sample.1 = xwt(df.w.tas, df.w.tas.sample.1)
    df.xwt.sample.2 = xwt(df.w.tas, df.w.tas.sample.2)
    
    # Make room to the right for the color bar
    png("figures/fig.wl-3-2.png", width=17, height=6, res=300, units="in")
      par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)
      plot.wavelet(df.xwt.sample.1, type="power.corr.norm", plot.phase=TRUE,
                   main="XWT based on Solution 1")
      plot.wavelet(df.xwt.sample.2, type="power.corr.norm", plot.phase=TRUE,
                   main="XWT based on Solution 2")
    dev.off()
    ## svg 
    ##   2

4 Bivariate Wavelet: Wavelet Coherence (WTC)

  • Specifying the number of iterations. The more, the better, but more time-comsuming
nrands1 = 10
nrands2 = 100
  • Calculating wavelet coherence & Estimating performance with different nrands
system.time({df.wtc.1=wtc(df.w.tas, df.w.hum, nrands=nrands1, quiet=TRUE)})
##    user  system elapsed 
##   1.349   0.148   1.371
system.time({df.wtc.2=wtc(df.w.tas, df.w.hum, nrands=nrands2, quiet=TRUE)})
##    user  system elapsed 
##   9.859   0.218   9.946
  • Plot
png("figures/fig.wl-4.png", width=17, height=6, res=300, units="in")
  par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)

  plot.wavelet(df.wtc.1, type="power.corr.norm", plot.phase=TRUE,
               main="(a) Wavelet Coherence: T2m vs Q2m; nrands=10")
  plot.wavelet(df.wtc.2, type="power.corr.norm", plot.phase=TRUE,
               main="(b) Wavelet Coherence: T2m vs Q2m; nrands=100")
dev.off()
## svg 
##   2

  • Interpretation

    • Time is displayed on the horizontal axis, while the vertical axis shows the frequency (the lower the frequency, the higher the scale). Regions in time-frequency space where the two time series co-vary are located by the wavelet coherence.

    • Warmer colors (red) represent regions with significant interrelation, while colder colors (blue) signify lower dependence between the series. Cold regions beyond the significant areas represent time and frequencies with no dependence in the series.

    • An arrow in the wavelet coherence plots represents the lead/lag phase relations between the examined series. A zero phase difference means that the two time series move together on a particular scale. Arrows point to the right (left) when the time series are in phase (anti-phase).

    • When the two series are in phase, it indicates that they move in the same direction, and anti-phase means that they move in the opposite direction. Arrows pointing to the right-down or left-up indicate that the first variable is leading, while arrows pointing to the right-up or left-down show that the second variable is leading.

---
title: "Analysis: Wavelet Analysis"
author: "by NXTTNX"
output:
  html_document: 
    code_download: TRUE
    code_folding: show
    number_sections: TRUE
    theme:
        bootswatch: lumen
    toc: TRUE
    toc_float: TRUE
    dev: 'svg'
---

<style>
/* resize the widget container */
.plotly { 
  width: 60% !important;
  margin: auto;
}

/* center the widget */
div.svg-container {
  margin: auto !important;
}
</style>

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, fig.align = "center")
```

```{r klippy, echo=FALSE, include=TRUE}
klippy::klippy(c('r', 'python', 'json', 'linux'), position = c('top', 'right'),
               tooltip_message = 'Click to copy', tooltip_success = 'Copied')
```

* Application of the continuous wavelet transform (CWT), cross wavelet transform (XWT),
  and wavelet coherence (WTC) to time series data.
* More information about Wavelet analysis can be found in a paper by A. Grinsted et al. (2004) [here](./refs/Grinsted-npg2004-wavelet-coherence.pdf)
* So called: Complete time-frequency space
* Some applications of Wavelet analysis in Atmospheric science can be found in **References** part
* Key notes (A. Grinsted et al., 2004):
  * The CWT expands a time series into a time frequency space where oscillations can be seen in a highly intuitive way
  * The XWT exposes regions with high common power and further reveals information about the phase relationship. If the two series are physically related we would expect a consistent or slowly varying phase lag that can be tested against mechanistic models of the physical process
  * WTC can be thought of as the local correlation between two CWTs. In this way locally phase locked behavior is uncovered. The more desirable features of the WTC come at the price of being slightly less localized in time frequency space. The significance level of the WTC has to be determined using Monte Carlo methods


# Initializing

  * **Set up environment**

  ```{r message = FALSE, warning = FALSE}
  library(tidyverse)      # Core environment for all R related works
  library(data.table)     # Data_table control
  library(microbenchmark) # To estimate system time
  library(RColorBrewer)   # Color control
  library(gridExtra)      # Multiple panels plot

  library(biwavelet)      # Core package in this analysis
  

  theme_set(theme_bw(base_size = 13, base_family = "Helvetica") +
            theme(plot.title = element_text(size = 14, hjust = 0.0),
                  panel.grid.minor = element_blank(),
                  panel.grid.major = element_line(colour = "gray", linetype="dashed",size = 0.1),
                  legend.position = "none", 
                  legend.title = element_blank(),
                  legend.text.align = 0))
  ```

  * **Read input, including JJA 3-hourly 2m Temperature & 2m Specific humidity during a 10-year period**

  ```{r message = FALSE, warning = FALSE}
  # Reading 2m Temperature
  df.tas = fread("data-txt/wavelet_OBS_tas.text", header=FALSE, skip=1,
                  col.names=c("date", "time", "tas"))
  # Reading 2m Specific humidity 
  df.hum = fread("data-txt/wavelet_OBS_huss.text", header=FALSE, skip=1,
                  col.names=c("date", "time", "hum"))
  ```

  * **Merging data into one is a good behaviour when working with "tidyverse"**

  ```{r message = FALSE, warning = FALSE}
  df = inner_join(df.tas, df.hum) %>%
       mutate(year=substr(date, 1, 4),
              mon =substr(date, 6, 7),
              day =substr(date, 9, 10)) %>% 
       dplyr::select(date, year, mon, day, time, tas, hum)
  ```

  * **Standardizing the data**

  ```{r message = FALSE, warning = FALSE}
  df$tas = df$tas - mean(df$tas)
  df$hum = df$hum - mean(df$hum)
  ```

  * **Quick look at the data using ggplot**

  ```{r message = FALSE, warning = FALSE}
  df %>% 
  ggplot() +
    geom_line(aes(x=date, y=tas, group=year), color="orange") +
    geom_line(aes(x=date, y=hum*1000, group=year), color="blue") +
    scale_x_continuous(name="Date",
                       breaks=as.Date(paste0(seq(2000, 2010, 1), "-06-01"))) +
    scale_y_continuous(name="Temperature (°C)", 
                       sec.axis = sec_axis(name="Specific Humidity (g)", ~ . /1000)) +
    labs(title="3-hourly comparison between T and Q => UNCLEAR signal") +
    theme(legend.position="bottom",
          axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
          axis.text.y.left = element_text(color = "orange"),
          axis.title.y.left = element_text(color = "orange"),
          axis.text.y.right = element_text(color = "blue"),
          axis.title.y.right = element_text(color = "blue"))
  ```
  
  ```{r message = FALSE, warning = FALSE}
  df.daily =df %>% 
    group_by(date, year, mon, day) %>%
    summarise(tas=mean(tas),
              hum=mean(hum)) %>%
    ungroup()

  df.daily %>% 
  ggplot() +
    geom_line(aes(x=date, y=tas, group=year), color="orange") +
    geom_line(aes(x=date, y=hum*1000, group=year), color="blue") +
    scale_x_continuous(name="Date",
                       breaks=as.Date(paste0(seq(2000, 2010, 1), "-06-01"))) +
    scale_y_continuous(name="Temperature (°C)", 
                       sec.axis = sec_axis(name="Specific Humidity (g)", ~ . /1000)) +
    labs(title="Daily comparison between T and Q => still UNCLEAR signal") +
    theme(legend.position="bottom",
          axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
          axis.text.y.left = element_text(color = "orange"),
          axis.title.y.left = element_text(color = "orange"),
          axis.text.y.right = element_text(color = "blue"),
          axis.title.y.right = element_text(color = "blue"))
  ```

* **A common activity is to find a way to look at the distribution(s) of the data**

  ```{r message = FALSE, warning = FALSE}
  p1=df.daily %>%
  ggplot() +
    geom_boxplot(aes(x=year, y=tas), alpha=0., width=0.5) +
    geom_violin(aes(x=year, y=tas), fill="orange", alpha=0.6) +
    scale_x_discrete(breaks=seq(2000, 2010, 1)) +
    labs(title="(a) Boxplot of daily 2m Temperature for each year",
         x="Year", y="Temperature (°C)") +
    theme(legend.position="bottom")
  
  p2=df.daily %>%
  ggplot() +
    geom_boxplot(aes(x=year, y=hum), alpha=0., width=0.5) +
    geom_violin(aes(x=year, y=hum), fill="blue", alpha=0.6) +
    scale_x_discrete(breaks=seq(2000, 2010, 1)) +
    labs(title="(b) Boxplot of daily 2m Specific humidity for each year",
         x="Year", y="Specific humidity (g)") +
    theme(legend.position="bottom")
  
  grid.arrange(p1, p2, ncol=1)
  ```

# Univariate wavelet: Continuous Wavelet Transform (CWT)

  * **Used for feature extraction purposes, especially in extracting low s/n ratio signals in time series**

  * **Discrete Wavelet Transform (DWT) is also existed, but not under interest yet**

  * **Defining two sets of data (2 variables) with time stamps**
  
  ```{r message = FALSE, warning = FALSE}
  df.w.tas = data.frame(
                time=1:((dim(df.daily)[1])),
                val =df.daily %>% dplyr::select(tas) %>% .$tas)
  df.w.hum = data.frame(
                time=1:((dim(df.daily)[1])),
                val =df.daily %>% dplyr::select(hum) %>% .$hum)
  ```

  * **Applying Wavelet Transform to estimate wavelet power spectrum**
  ```{r message = FALSE, warning = FALSE}
  df.wt.tas = wt(df.w.tas)
  df.wt.hum = wt(df.w.hum)
  ```

  * **Defining a visualize function**

  ```{r message = FALSE, warning = FALSE}
  plot.wavelet = function(df, type="power", plot.cb=TRUE, plot.phase=FALSE,
                          lty.coi=1, col.coi="grey", lwd.coi=2, lwd.sig=2, arrow.lwd=0.03, arrow.len=0.12,
                          xlab="Time (Date; JJA only", ylab="Period (day)",
                          main="", xaxt="n") {
    plot(df, type=type, plot.cb=plot.cb, plot.phase=plot.phase,
         lty.coi=lty.coi, col.coi=col.coi, lwd.coi=lwd.coi, lwd.sig=lwd.sig,
         xlab=xlab, ylab=ylab, main=main, xaxt=xaxt)
    axis(side=1, at=seq(1, 930, 93), labels=seq(2000, 2009, 1))
  }
  ```



  * **Visualizing Wavelet power. There are several options:**

Type                | Meanings
------------------- | -------------------------------------------------
**power**           | Raw power
**power.corr**      | Bias-corrected Power
**power.norm**      | Power normalized by the variance
**power.corr.norm** | Bias-corrected power normalized by the variance
**wavelet**         | Wavelet coefficients
**phase**           | Phase

  * **Ploting Raw power**

  ```{r message = FALSE, warning = FALSE}
  png("figures/fig.wl-1.png", width=17, height=6, res=300, units="in")
    par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)

    plot.wavelet(df.wt.tas, type="power.norm", main="(a) Temperature Raw power")
    plot.wavelet(df.wt.hum, type="power.norm", main="(b) Specific humidity Raw power")
  dev.off()
  ```
![](figures/fig.wl-1.png)

  * **Ploting Bias-corrected normalized power**
  
  ```{r message = FALSE, warning = FALSE}
  png("figures/fig.wl-2.png", width=17, height=6, res=300, units="in")
    par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)

    plot.wavelet(df.wt.tas, type="power.corr.norm", main="(a) Temperature Bias-corrected power")
    plot.wavelet(df.wt.hum, type="power.corr.norm", main="(b) Specific humidity Bias-corrected power")
  dev.off()
  ```
![](figures/fig.wl-2.png)


# Bivariate Wavelet: Cross Wavelet Transform (XWT)

* **Calculating XWT**
  
  ```{r message = FALSE, warning = FALSE}
  df.xwt=xwt(df.w.tas, df.w.hum)
  ```

* **Visualizing**

  ```{r message = FALSE, warning = FALSE}
  png("figures/fig.wl-3.png", width=17, height=6, res=300, units="in")
    par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)
    
    plot.wavelet(df.xwt, type="power", plot.phase=TRUE,
                 main="XWT of Temperature & Specific humidity")
    plot.wavelet(df.xwt, type="power.corr.norm", plot.phase=TRUE,
                 main="XWT of Temperature & Specific humidity")
  dev.off()
  ```
![](figures/fig.wl-3.png)

* **This results are possibly caused by: 1) Specific humidity is not in form of Normal distribution?; or 2) Temperature and Specific humidity have different characteristics**

  ```{r message = FALSE, warning = FALSE, fig.width=10}
  p1=df.w.tas %>% ggplot() +
    geom_histogram(aes(x=val, y=..count..), fill="orange", alpha=0.6) +
    labs(title="Temperature distribution", x="Temperature", y="Count")
  p2=df.w.hum %>% ggplot() +
    geom_histogram(aes(x=val, y=..count..), fill="blue", alpha=0.6) +
    labs(title="Specific humidity distribution", x="Specific humidity", y="Count")
  grid.arrange(p1, p2, ncol=2)
  ```

  * **Solution 1**: Try to Sampling a data following Normal distribution based on Specific Humidity data

    ```{r message = FALSE, warning = FALSE, fig.width=10}
    # Sample time-series
    n=dim(df.w.tas)[1]
    df.w.tas.sample.1 = data.frame(time=1:n,
                                   val=rnorm(n, mean=mean(df.w.hum$val),
                                                sd=sd(df.w.hum$val)))
    ```

  * **Solution 2**: Try to simply Adjust Specific Humidity data closer to Temperature (by multiply with 1000)

    ```{r message = FALSE, warning = FALSE, fig.width=10}
    # Sample time-series
    df.w.tas.sample.2 = data.frame(time=df.w.hum$time,
                                   val =df.w.hum$val*1000)
    ```

  * Visualizing

    ```{r message = FALSE, warning = FALSE, fig.width=10}
    p1 = df.w.tas.sample.1 %>% ggplot() +
      geom_histogram(aes(x=val, y=..count..), fill="orange", alpha=0.6) +
      labs(title="Solution 1", x="Sampling data based on Specific Humidity", y="Count")

    p2 = df.w.tas.sample.2 %>% ggplot() +
      geom_histogram(aes(x=val, y=..count..), fill="orange", alpha=0.6) +
      labs(title="Solution 2", x="Adjust Specific humidity data", y="Count")

    grid.arrange(p1, p2, ncol=2)
    ```

* **Performing XWT again & visualizing**

  ```{r message = FALSE, warning = FALSE, fig.width=10}
  # Cross-wavelet
  df.xwt.sample.1 = xwt(df.w.tas, df.w.tas.sample.1)
  df.xwt.sample.2 = xwt(df.w.tas, df.w.tas.sample.2)

  # Make room to the right for the color bar
  png("figures/fig.wl-3-2.png", width=17, height=6, res=300, units="in")
    par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)
    plot.wavelet(df.xwt.sample.1, type="power.corr.norm", plot.phase=TRUE,
                 main="XWT based on Solution 1")
    plot.wavelet(df.xwt.sample.2, type="power.corr.norm", plot.phase=TRUE,
                 main="XWT based on Solution 2")
  dev.off()
  ```
![](figures/fig.wl-3-2.png)


# Bivariate Wavelet: Wavelet Coherence (WTC)

  * **Specifying the number of iterations. The more, the better, but more time-comsuming**

  ```{r message = FALSE, warning = FALSE}
  nrands1 = 10
  nrands2 = 100
  ```

  * **Calculating wavelet coherence & Estimating performance with different nrands**

  ```{r message = FALSE, warning = FALSE}
  system.time({df.wtc.1=wtc(df.w.tas, df.w.hum, nrands=nrands1, quiet=TRUE)})
  ```

  ```{r message = FALSE, warning = FALSE}
  system.time({df.wtc.2=wtc(df.w.tas, df.w.hum, nrands=nrands2, quiet=TRUE)})
  ```

  * **Plot**

  ```{r message = FALSE, warning = FALSE}
  png("figures/fig.wl-4.png", width=17, height=6, res=300, units="in")
    par(mfrow=c(1, 2), oma=c(1, 0, 0, 1.2), mar=c(5, 5, 3, 5) + 0.1)

    plot.wavelet(df.wtc.1, type="power.corr.norm", plot.phase=TRUE,
                 main="(a) Wavelet Coherence: T2m vs Q2m; nrands=10")
    plot.wavelet(df.wtc.2, type="power.corr.norm", plot.phase=TRUE,
                 main="(b) Wavelet Coherence: T2m vs Q2m; nrands=100")
  dev.off()
  ```
![](figures/fig.wl-4.png)

* **Interpretation**

  * Time is displayed on the horizontal axis, while the vertical axis shows the
    frequency (the lower the frequency, the higher the scale). Regions in 
    time-frequency space where the two time series co-vary are located by the 
    wavelet coherence.

  * Warmer colors (red) represent regions with significant interrelation, while 
    colder colors (blue) signify lower dependence between the series.
    Cold regions beyond the significant areas represent time and frequencies with 
    no dependence in the series.

  * An arrow in the wavelet coherence plots represents the lead/lag phase relations
    between the examined series. A zero phase difference means that the two time
    series move together on a particular scale. Arrows point to the right (left)
    when the time series are in phase (anti-phase).

  * When the two series are in phase, it indicates that they move in the same
    direction, and anti-phase means that they move in the opposite direction. 
    Arrows pointing to the right-down or left-up indicate that the first variable 
    is leading, while arrows pointing to the right-up or left-down show that the 
    second variable is leading.


# References
* https://atoc.colorado.edu/research/wavelets/montecarlo.html
* https://cran.r-project.org/web/packages/biwavelet/biwavelet.pdf
* https://www.glaciology.net/publication/2004-12-24-application-of-the-cross-wavelet-transform-and-wavelet-coherence-to-geophysical-time-series/
* https://rstudio-pubs-static.s3.amazonaws.com/152496_026c3ac97f7d40e5ba0cadf757730fce.html
* https://github.com/tgouhier/biwavelet
* http://dx.doi.org/10.1002/2015GL063968
* https://doi.org/10.1175/JCLI-D-18-0076.1
* https://doi.org/10.1029/2019GL084995
* https://doi.org/10.3389/fenrg.2020.591469

