Awesome

bfastApp

Loïc Dutrieux

An application to interactively select function parameters when segmenting irregular time-series

The application particularly focuses on irregular time-series, extracted from multi-temporal remote sensing data such as Landsat time-series. An integration with the bfastSpatial package makes it easy to extract the objects that need to be uploaded to the app. Below is a quick tutorial on how to use the app.

First if you have not yet installed the bfastSpatial package, you should do so by running the following command.

devtools::install_github('dutri001/bfastSpatial')

The bfastSpatial contains a built in dataset from which we will extract the time-series objects which will be later fed to the app. The object is named tura, it is a rasterBrick object with time stored in the Z dimension. Each layer correspond to NDVI calculated from a Landsat scene acquired at a different date.

# Load and visualize tura object
library(bfastSpatial)
data(tura)
plot(tura, 3)

# In case an object does not have time written to the z dimension, it can be done as follows
setZ(tura, getSceneinfo(names(tura))$date)

We will extract sample time-series using the zooExtract() function of the bfastSpatial package. The function works by extracting the time-series of individual pixels at locations specified by a SpatialPoints object. Therefore we need to generate that object since we do have one yet. However, such file could come from a shapefile for instance containing reference in-situ data. A regular sample can be achieved using the sampleRegular() function from the raster package.

# Generate SpatialPoints object and visualize
sp <- sampleRegular(tura, size = 40, sp=TRUE)
plot(tura, 3)
plot(sp, add=TRUE)

# Extract samples and prepare rds file
zooExtract(x = tura, sample = sp, file = 'turaZoo.rds')

Now the file is stored in your working directory, ready to be opened in the application.

Running the App

The app requires a set of packages that need to be installed prior to running it. In case you do not have installed you, run the following command.

install.packages('shiny', 'zoo', 'bfast', 'strucchange', 'ggplot2', 'lubridate')

The app can be ran more or less directly from github, using the runGitHub function.

library(shiny)
runGitHub(repo = 'bfastApp', username = 'dutri001')

Point to the turaZoo.rds file that you have saved in an earlier step, and you're ready to explore your time-series objects.