Vertical profile in r plot() - plot

Question

I am building a vertical profile plot of water columns. My issue is that the dots are connected on the x observations, and not the y observations. Under ggplot, I know geom_path can do this, but I can't use ggplot as I want to add several x axis. Therefore I am using plot().
So here is what I tried:
Storfjorden <- read.delim("C:/Users/carvi/Desktop/Storfjorden.txt")
smooF=smooth.spline(Storfjorden$Fluorescence,Storfjorden$Depth,spar=0.50)
plot(Storfjorden$Fluorescence,Storfjorden$Depth,ylim=c(80,0),type="n")
lines(smooF)
Resulting plot
As you see, the dots are connected through x observations. But to observe a vertical profile, I would like to see them connected through y observations. I tried ordering them by depth (using order()) and it didn't affect the result. Anyone has a clue?
If, as an alternative, someone would have an idea how to plot different lines with different axis on a single plot (Temperature, salinity, fluorescence), then I may use geom_path (). Thank you!
**An emerging question I have that you may answer, is there a way in ggplot to make a geom_smooth(), but with the observations connected in order they appear instead of x axis?
ggplot(melteddf,aes(y=Depth,x=value))+geom_path()+facet_wrap
+(~variable,nrow=1,scales="free‌​_x")+scale_y_reverse‌​()
+geom_smooth(span=‌​0.5,se=FALSE)
I tried using smooth.spline, but didn't recognize the object in geom_path. Thanks!

Answer 1

There is a reason that ggplot2 makes it difficult to plot multiple x-axes on a single plot -- it generally leads to difficult to read (or worse, misleading) graphs. If you have a motivating example for why your example will not fall into one of those categories, it might allow us to help you more to know more details. Below, however, are two workarounds that might help.
Here is a quick MWE to address the question -- it might be more helpful if you gave us something that looks like your actual data, but this at least gets things on very different scales (though, with no structure, the plots are rather messy).
Note that I am using dplyr for several manipulations and reshape2 to melt the data into a long format for easier plotting.
library(dplyr)
library(reshape2)
df <-
data.frame(
depth = runif(20, 0, 100) %>% round %>% sort
, measureA = rnorm(20, 10, 3)
, measureB = rnorm(20, 50, 10)
, measureC = rnorm(20, 1000, 30)
)
meltedDF <-
df %>%
melt(id.vars = "depth")
The first option is to simply use facets to plot the data next to each other:
meltedDF %>%
ggplot(aes(y = depth
, x = value)) +
geom_path() +
facet_wrap(~variable
, nrow = 1
, scales = "free_x") +
scale_y_reverse()
The second is to standardize the data, then plot that. Here, I am using the z-score, though if you have a reason to use something else (e.g. scaled to center at the "appropriate" amount of whatever variable you are using) you could change that formula:
meltedDF %>%
group_by(variable) %>%
mutate(Standardized = (value - mean(value)) / sd(value) ) %>%
ggplot(aes(y = depth
, x = Standardized
, col = variable)) +
geom_path() +
scale_y_reverse()
If you need to plot multiple sites, here is some sample data with sites:
df <-
data.frame(
depth = runif(60, 0, 100) %>% round %>% sort
, measureA = rnorm(60, 10, 3)
, measureB = rnorm(60, 50, 10)
, measureC = rnorm(60, 1000, 30)
, site = sample(LETTERS[1:3], 60, TRUE)
)
meltedDF <-
df %>%
melt(id.vars = c("site", "depth"))
You can either use facet_grid (my preference):
meltedDF %>%
ggplot(aes(y = depth
, x = value)) +
geom_path() +
facet_grid(site~variable
, scales = "free_x") +
scale_y_reverse()
or add facet_wrap to the standardized plot:
meltedDF %>%
ggplot(aes(y = depth
, x = value)) +
geom_path() +
facet_grid(site~variable
, scales = "free_x") +
scale_y_reverse()