subsetting

Subsetting

stars objects are lists of identically shaped arrays, with one array per variable. This is analogous to the data.frame which is a list of identically shaped vectors, with one vector per variable. Note: there is no requirement that each variable be of the same data type as all of the others (numeric, character, factor, etc.) Thus is is possible to perform manipulations on stars that are analogous to manipulations on data.frame objects.

Tidy manipulation methods

There is a wonderful vignette that describes tidyverse methods for stars - there is no need to reinvent the wheel here.

Base [ and [[ subsetting.

This is actually well described in this vignette but I thought it worth embellishment here.

First we load a toy dataset. Here we make it a 2 variable dataset.

toy <- c(toy_multi(), toy_multi() |> dplyr::mutate(b1.b2.b3.b4.b5 = b1.b2.b3.b4.b5 /500)) |>
  rlang::set_names(c("A", "B"))
toy

## stars object with 3 dimensions and 2 attributes
## attribute(s):
##     Min. 1st Qu.  Median    Mean 3rd Qu. Max. NA's
## A  1.000 115.500 234.000 237.000 356.500  500  225
## B  0.002   0.231   0.468   0.474   0.713    1  225
## dimension(s):
##      from to offset delta    values x/y
## x       1 10      0     1      NULL [x]
## y       1 10     10    -1      NULL [y]
## band    1  5     NA    NA b1,...,b5

Below is a table outline different methods for manipulation. Pay close attention to slicing and the order of arguments to [ which are usually [attribute/variable, x/lon, y/lat, band/layer]. Note that stars is customizable so the order of dimensions, other than attribute/variable is really use defined.

Action	Subset	Tidy	Outcome
select attributes(s)	toy[‘A’]	select(toy, A)	stars object
extract entire array	toy[[‘A’]]	pull(toy, A)	n-d array
slice	toy[2,1:3,6:10,2:3]	select(toy, B) |\> slice(x, 1:3) |\> slice(y, 6:10) |\> slice(band, 2:3)	stars object

select attributes

Selection of one or more attributes.

toy['A']

## stars object with 3 dimensions and 1 attribute
## attribute(s):
##    Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## A     1   115.5    234  237   356.5  500  225
## dimension(s):
##      from to offset delta    values x/y
## x       1 10      0     1      NULL [x]
## y       1 10     10    -1      NULL [y]
## band    1  5     NA    NA b1,...,b5

select(toy, A)

## stars object with 3 dimensions and 1 attribute
## attribute(s):
##    Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## A     1   115.5    234  237   356.5  500  225
## dimension(s):
##      from to offset delta    values x/y
## x       1 10      0     1      NULL [x]
## y       1 10     10    -1      NULL [y]
## band    1  5     NA    NA b1,...,b5

extract entire array

Extraction of the array of data for one attribute.

toy[['A']] |> str()

##  num [1:10, 1:10, 1:5] 1 2 3 4 5 6 7 8 9 10 ...
##  - attr(*, "names")= chr [1:500] "b11" "b12" "b13" "b14" ...

pull(toy, A) |> str()

##  num [1:10, 1:10, 1:5] 1 2 3 4 5 6 7 8 9 10 ...

slice

Extraction of slices of the array.

toy[2,1:3, 6:10, 2:3]

## stars object with 3 dimensions and 1 attribute
## attribute(s):
##    Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## B    NA      NA     NA  NaN      NA   NA   30
## dimension(s):
##      from to offset delta values x/y
## x       1  3      0     1   NULL [x]
## y       6 10     10    -1   NULL [y]
## band    2  3     NA    NA b2, b3

select(toy, B) |> slice(x, 1:3) |> slice(y, 6:10) |> slice(band, 2:3)

## stars object with 3 dimensions and 1 attribute
## attribute(s):
##    Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## B    NA      NA     NA  NaN      NA   NA   30
## dimension(s):
##      from to offset delta values x/y
## x       1  3      0     1   NULL [x]
## y       6 10     10    -1   NULL [y]
## band    2  3     NA    NA b2, b3