Introduction to spatstat
0 likes944 views
This document introduces the R package spatstat for analyzing spatial point pattern data. Spatstat allows users to create, manipulate, and analyze point patterns. It supports exploratory data analysis, simulation of point process models, parametric model fitting, and hypothesis testing. The document demonstrates how to access spatstat in R, plot and summarize point patterns, analyze local spatial variations in intensity, and work with multiple point patterns including marking patterns with categorical variables.
Introduction to spatstat
- 1. INTRODUCTION TO SPATSTAT Richard W Wamalwa1 - MSc., MBA(Finance), BSc. 1Directorate of Academic Quality Assurance, JKUAT RM 610-Environmental, Spatial, GIS, 2011
- 2. OUTLINE Introduction to Spatstat Exploratory data analysis Multiple point patterns Reference
- 3. INTRODUCTION TO SPATSTAT Spatstat is a contributed R package for analysing spatial data which supports: a. creation, manipulation and plotting of point patterns b. exploratory data analysis c. simulation of point process models d. parametric model-fitting e. hypothesis tests, residual plots, diagnostics
- 4. INTRODUCTION – CONT’D # Getting help in R >help(spatstat) #Accessing spatstat in R >library(spatstat) >data(swedishpines) x <- swedishpines >plot(x) >summary(x) #To get an impression of local spatial variations in intensity >plot(density(x, 10)) > plot(density(swedishpines, sigma = 10))
- 5. INTRODUCTION – CONT’D #10 - chosen value for the standard deviation of the Gaussian smoothing kernel. >contour(density(x,10),axes=FALSE) #To see a list of all methods available in R for a particular generic function such as plot: >methods(plot) #To see a list of all methods that are available for a particular class such as ppp: >methods(class = "ppp")
- 6. EXPLORATORY DATA ANALYSIS Spatstat is designed to support exploratory data analysis for point patterns e.g. quadrat counting. The study region is divided into rectangles (‘quadrats’) of equal size, and the number of points in each rectangle is counted. Q <- quadratcount(X, nx = 4, ny = 3); <Q plot(x) plot(Q, add = TRUE, cex = 2) plot(Q, cex = 0.5, pch = "+") den <- density(Q, sigma = 70) plot(den) plot(Q, add = TRUE, cex = 0.5) K <- Kest(x) plot(K)
- 7. MULTIPLE POINT PATTERNS Multilpe point patterns is a marked point pattern in which the marks are a categorical variable. data(lansing) lansing summary(lansing) plot(lansing) plot(split(lansing)) The result of split(lansing) is a list of point patterns. To extract one of these patterns, e.g. the hickories, hick <- split(lansing)$hickory plot(hick)
- 8. MULTIPLE POINT PATTERNS a <- plot(lansing) legend(-0.25, 0.5, names(a), pch = a) Lansing Woods data Is the study region divided into domains where a single tree species dominates, or are they different species randomly interspersed?
- 9. REFERENCE Adrian Baddeley (2008). Analysing spatial point patterns in R. Adrian.Baddeley@csiro.au; adrian@maths.uwa.edu.au.