SLIDE 1 Spatio-Temporal Statistics with R
Chapter Two: Exploring Spatio-Temporal Data
SLIDE 2
Spatio-Temporal Data
SLIDE 3 Spatio-Temporal Data
- Geostatistical: continuous spatial index
- Areal (lattice): defined on finite/countable subset in space
- Point process: randomly located spatial processes
SLIDE 4
NOAA Daily Weather
SLIDE 5
Sea surface temperature
SLIDE 6
Mediterranean winds
SLIDE 7 Spatio-Temporal Data … in R
- spacetime package in R extends definitions from sp and xts
- How to represent spatio-temporal data?
- Time-wide tables
- Space-wide tables
- Long format
SLIDE 8
Visualization
SLIDE 9
Spatial Plots
SLIDE 10
Time-Series Plots
SLIDE 11
Hovmöller Plots
SLIDE 12
Hovmöller Plots
SLIDE 13 Interactivity
- trelliscope package allows for exploration/visualization of large-
scale data sets
- Designed to visualize distributed date by processing in parallel
and then recombining.
SLIDE 14
Exploratory Analysis
SLIDE 15
Spatial Means
SLIDE 16
Temporal Means
SLIDE 17
Empirical spatial covariability
SLIDE 18
Empirical spatial covariability
SLIDE 19
- The empirical spatiotemporal covariogram for spatial lag h and
time lag is given by
τ
Spatio-Temporal Covariograms and Semivariograms
SLIDE 20 Spatio-Temporal Covariograms and Semivariograms
- If we assume covariance only depends on displacement in time/
space, we can estimate semivariogram:
SLIDE 21 Spatio-Temporal Covariograms and Semivariograms
SLIDE 22 Empirical Orthogonal Functions
- Empirical orthogonal functions (EOFs) used for dimensionality
reduction and for studying spatial structure
- Recall principal component analysis projects original data onto
new coordinate space where the first coordinate aligns with the axis of largest variation, …
- EOFs are obtained by treating observations at different time
points as “sample” and computing principal components
SLIDE 23
Empirical Orthogonal Functions
SLIDE 24
Lab 2.2: Visualization
SLIDE 25
Lab 2.3: Exploratory Data Analysis