This tutorial covers how to improve plotting output using the
R. Specifically it covers using
levelplot() and adding meaningful custom
names to bands within a
R Skill Level: Intermediate - you’ve got the basics of
Goals / Objectives
After completing this activity, you will:
- Be able to assign custom names to bands in a RasterStack for prettier plotting.
- Understand advanced plotting of rasters using the
Things You’ll Need To Complete This Tutorial
You will need the most current version of
R and, preferably,
on your computer to complete this tutorial.
Install R Packages
- More on Packages in R - Adapted from Software Carpentry.
Data to Download
The imagery data used to create this raster teaching data subset were
collected over the
National Ecological Observatory Network’s
San Joaquin Experimental Range
The imagery was created by the U.S. Geological Survey (USGS) using a multispectral scanner on a Landsat Satellite. The data files are Geographic Tagged Image-File Format (GeoTIFF).
Set Working Directory: This lesson assumes that you have set your working
directory to the location of the downloaded and unzipped data subsets. An overview
of setting the working directory in
R can be found here.
R Script & Challenge Code: NEON data lessons often contain challenges that reinforce
learned skills. If available, the code for challenge solutions is found in the
R script of the entire lesson, available in the footer of each lesson page.
In this tutorial, we are working with the same set of rasters used in the
Raster Time Series Data in R
tutorial. This data is derived from the Landsat satellite and stored in
GeoTIFF format. Each raster covers the
NEON Harvard Forest field site.
If you have not already created the RasterStack, originally created in Raster Time Series Data in R , please create it now.
library(raster) library(rgdal) library(rasterVis) ## Loading required package: lattice ## Loading required package: latticeExtra ## Loading required package: RColorBrewer ## ## Attaching package: 'latticeExtra' ## The following object is masked from 'package:ggplot2': ## ## layer # Create list of NDVI file paths all_NDVI_HARV <- list.files("NEON-DS-Landsat-NDVI/HARV/2011/NDVI", full.names = TRUE, pattern = ".tif$") # Create a time series raster stack NDVI_HARV_stack <- stack(all_NDVI_HARV) # apply scale factor NDVI_HARV_stack <- NDVI_HARV_stack/10000
Plot Raster Time Series Data
We can use the
plot function to plot our raster time series data.
# view a histogram of all of the rasters # nc specifies number of columns plot(NDVI_HARV_stack, zlim = c(.15, 1), nc = 4)
Data Tip: The range of values for NDVI is 0-1. However, the data stored in our raster ranges from 0 - 10,000. If we view the metadata for the original .tif files, we will see a scale factor of 10,000 is defined. Multiplying values with decimal places by a factor of 10, allows the data to be stored in integer format (no decimals) rather than a floating point format (containing decimals). This keeps the file size smaller.
Our plot is nice however, it’s missing some key elements including, easily
readable titles. It also contains a legend that is repeated for each image. We
levelplot from the
rasterVis package to make our plot prettier!
The syntax for the
levelplot() function is similar to that for the
function. We use
main="TITLE" to add a title to the entire plot series.
# create a `levelplot` plot levelplot(NDVI_HARV_stack, main="Landsat NDVI\nNEON Harvard Forest")
Adjust the Color Ramp
Next, let’s adjust the color ramp used to render the rasters. First, we
can change the red color ramp to a green one that is more visually suited to our
NDVI (greenness) data using the
colorRampPalette() function in combination with
# use colorbrewer which loads with the rasterVis package to generate # a color ramp of yellow to green cols <- colorRampPalette(brewer.pal(9,"YlGn")) # create a level plot - plot levelplot(NDVI_HARV_stack, main="Landsat NDVI -- Improved Colors \nNEON Harvard Forest Field Site", col.regions=cols)
The yellow to green color ramp visually represents NDVI well given it’s a measure of greenness. Someone looking at the plot can quickly understand that pixels that are more green, have a higher NDVI value.
- For all of the
brewer.palramp names see the brewerpal page.
Data Tip: Cynthia Brewer, the creater of ColorBrewer, offers an online tool to help choose suitable color ramps, or to create your own. ColorBrewer 2.0; Color Advise for Cartography
Refine Plot & Tile Labels
Next, let’s label each raster in our plot with the Julian day that the raster
represents. The current names come from the band (layer names) stored in the
RasterStack and first part each name is the Julian day.
To create a more meaningful label we can remove the “x” and replace it with
“day” using the
gsub() function in
R. The syntax is as follows:
First let’s remove “_HARV_NDVI_crop” from each label.
# view names for each raster layer names(NDVI_HARV_stack) ##  "X005_HARV_ndvi_crop" "X037_HARV_ndvi_crop" "X085_HARV_ndvi_crop" ##  "X133_HARV_ndvi_crop" "X181_HARV_ndvi_crop" "X197_HARV_ndvi_crop" ##  "X213_HARV_ndvi_crop" "X229_HARV_ndvi_crop" "X245_HARV_ndvi_crop" ##  "X261_HARV_ndvi_crop" "X277_HARV_ndvi_crop" "X293_HARV_ndvi_crop" ##  "X309_HARV_ndvi_crop" # use gsub to modify label names.that we'll use for the plot rasterNames <- gsub("X","Day ", names(NDVI_HARV_stack)) # view Names rasterNames ##  "Day 005_HARV_ndvi_crop" "Day 037_HARV_ndvi_crop" ##  "Day 085_HARV_ndvi_crop" "Day 133_HARV_ndvi_crop" ##  "Day 181_HARV_ndvi_crop" "Day 197_HARV_ndvi_crop" ##  "Day 213_HARV_ndvi_crop" "Day 229_HARV_ndvi_crop" ##  "Day 245_HARV_ndvi_crop" "Day 261_HARV_ndvi_crop" ##  "Day 277_HARV_ndvi_crop" "Day 293_HARV_ndvi_crop" ##  "Day 309_HARV_ndvi_crop" # Remove HARV_NDVI_crop from the second part of the string rasterNames <- gsub("_HARV_ndvi_crop","",rasterNames) # view names for each raster layer rasterNames ##  "Day 005" "Day 037" "Day 085" "Day 133" "Day 181" "Day 197" "Day 213" ##  "Day 229" "Day 245" "Day 261" "Day 277" "Day 293" "Day 309"
Data Tip: Instead of substituting “x” and
“_HARV_NDVI_crop” separately, we could have used use the vertical bar character
( | ) to replace more than one element.
For example “X|_HARV” tells
R to replace all instances of both “X” and “_HARV”
in the string. Example code to remove “x” an “_HARV…”:
gsub("X|_HARV_NDVI_crop"," | ","rasterNames")
Once the names for each band have been reassigned, we can render our plot with
the new labels using
# use level plot to create a nice plot with one legend and a 4x4 layout. levelplot(NDVI_HARV_stack, layout=c(4, 4), # create a 4x4 layout for the data col.regions=cols, # add a color ramp main="Landsat NDVI - Julian Days \nHarvard Forest 2011", names.attr=rasterNames)
We can adjust the columns of our plot too using
we adjust the layout to be a matrix of 5 columns and 3 rows.
# use level plot to create a nice plot with one legend and a 4x4 layout. levelplot(NDVI_HARV_stack, layout=c(5, 3), # create a 5x3 layout for the data col.regions=cols, # add a color ramp main="Landsat NDVI - Julian Days \nHarvard Forest 2011", names.attr=rasterNames)
scales allows us to modify the x and y-axis scale. Let’s simply
remove the axis ticks from the plot with
# use level plot to create a nice plot with one legend and a 4x4 layout. levelplot(NDVI_HARV_stack, layout=c(5, 3), # create a 5x3 layout for the data col.regions=cols, # add a color ramp main="Landsat NDVI - Julian Days \nHarvard Forest 2011", names.attr=rasterNames, scales=list(draw=FALSE )) # remove axes labels & ticks
Challenge: Divergent Color Ramps
When we used
gsub to modify the tile labels we replaced the beginning of each
tile title with “Day”. A more descriptive name could be “Julian Day”.
- Create a plot and label each tile “Julian Day” with the julian day value following.
- Change the colorramp to a divergent brown to green color ramp to represent the data. Hint: Use the brewerpal page to help choose a color ramp.
Questions: Does having a divergent color ramp represent the data better than a sequential color ramp (like “YlGn”)? Can you think of other data sets where a divergent color ramp may be best?