Introduction to Spatial Plotting of Breakpoints in bfast Lite

In the world of time-series analysis, identifying breakpoints is a crucial step in understanding and modeling complex patterns in data. The bfast package in R provides an efficient method for detecting these breaks using the bfastlite function. However, one common question arises after successfully identifying breakpoints: how to visually represent these points on a spatial map? In this blog post, we will delve into the world of spatial plotting and explore ways to link breakpoints with pixels in your study site.

Understanding bfast Lite

Before diving into spatial plotting, it’s essential to understand the bfastlite function and its output. The bfast package uses a combination of wavelet analysis and machine learning techniques to detect breaks in time-series data. The bfastlite function takes a univariate time series as input and returns an object that contains information about the detected breakpoints.

library(bfast)

# Load sample dataset from bfast package
data(simts)
datats <- ts(rowSums(simts$time.series))
tsp(datats) <- tsp(simts$time.series)  # Assign correct time series attributes

# Detect breaks using bfastlite function
bp <- bfastlite(datats)

# Plot the original time series data
plot(datats)

Visualizing Breakpoints with Plotting Functions

After detecting breakpoints, it’s essential to visualize these points on a plot. The plot function in R is suitable for this purpose. However, since we are dealing with multiple breaks (in this case, two), we need to modify the approach slightly.

# Detect breaks using bfastlite function
bp <- bfastlite(datats)

# Plot breakpoints
plot(bp)

Spatial Plotting of Breakpoints

Now that we have identified and visualized the breakpoints, it’s time to link these points with pixels in our study site. This requires spatial plotting techniques.

One common approach is to use a library like maps or ggmap which provides tools for mapping data onto geographical locations. We will focus on using maps in this example.

# Install and load maps package
install.packages("maps")
library(maps)

# Get a list of US states (you can replace with your study site)
us_states <- map_data("state")

# Filter out the state where our data is located
data_state <- us_states[us_states$region == "YourStateName",]

# Create a spatial plot of the breakpoints
plot(datats, main = "Spatial Plot of Breakpoints")

Note: Replace "YourStateName" with the actual name of your study site.

Creating a Cross-Shaped Marker for Each Breakpoint

To create a cross-shaped marker at each breakpoint, we need to use a combination of abline and text. This will give us a clear visual representation of where these breaks occur.

# Detect breaks using bfastlite function
bp <- bfastlite(datats)

# Create spatial plot of the breakpoints with crosses
plot(datats)
abline(h = bp$season, v = bp$abrupt, col = "red")
text(bp$season, bp$abrupt, label = "+", pos = 4, cex = 2)

Note: pos = 4 is used to position the cross marker above and to the right of each break point.

Visualizing Spatially Interpolated Breakpoints

Another approach to spatially interpolate breakpoints is using a library like focal or sp. These packages provide functions for performing focal analysis, which can be useful when dealing with large datasets.

# Install and load sp package
install.packages("sp")
library(sp)

# Perform focal analysis on the dataset
data_sp <- spatstat(data = datats)

# Get the focal points (breakpoints) with their corresponding distances from each data point
focal_points <- spatstat_focalize(data_sp, c(x = bp$season, y = bp$abrupt))

# Create a spatial plot of the focal points
plot(focal_points)

Note: This example uses spatstat which is a more advanced package for spatial statistics in R.

Conclusion

In conclusion, while there may not be a single straightforward solution to plotting breakpoints on a map, we have explored various approaches and techniques that can be used depending on the specific requirements of your project. From using built-in R functions like plot and abline, to more specialized packages like spatstat, spatial plotting offers many exciting opportunities for visualizing complex patterns in time-series data.

Remember, with great power comes great responsibility - when working with geospatial data, make sure you have the necessary permissions and adhere to any relevant regulations. Always test your code thoroughly before deploying it on a large scale.

We hope this blog post has provided valuable insights into spatial plotting of breakpoints using R and bfast package. If you have any questions or would like further clarification, please don’t hesitate to ask in the comments below!

Last modified on 2024-01-21