| calc {gdalraster} | R Documentation |
Raster calculation
Description
calc() evaluates an R expression for each pixel in a raster layer or
stack of layers. Each layer is defined by a raster filename, band number,
and a variable name to use in the R expression. If not specified, band
defaults to 1 for each input raster.
Variable names default to LETTERS if not specified
(A (layer 1), B (layer 2), ...).
All of the input layers must have the same extent and cell size.
The projection will be read from the first raster in the list
of inputs.
Individual pixel coordinates are also available as variables in the
R expression, as either x/y in the raster projected coordinate system or
inverse projected longitude/latitude.
Multiband output is supported as of gdalraster 1.11.0.
Usage
calc(
expr,
rasterfiles,
bands = NULL,
var.names = NULL,
dstfile = tempfile("rastcalc", fileext = ".tif"),
fmt = NULL,
dtName = "Int16",
out_band = NULL,
options = NULL,
nodata_value = NULL,
setRasterNodataValue = FALSE,
usePixelLonLat = NULL,
write_mode = "safe",
quiet = FALSE
)
Arguments
expr |
An R expression as a character string (e.g., |
rasterfiles |
Character vector of source raster filenames. |
bands |
Integer vector of band numbers to use for each raster layer. |
var.names |
Character vector of variable names to use for each raster layer. |
dstfile |
Character filename of output raster. |
fmt |
Output raster format name (e.g., "GTiff" or "HFA"). Will attempt to guess from the output filename if not specified. |
dtName |
Character name of output data type (e.g., Byte, Int16, UInt16, Int32, UInt32, Float32). |
out_band |
Integer band number(s) in |
options |
Optional list of format-specific creation options in a
vector of "NAME=VALUE" pairs
(e.g., |
nodata_value |
Numeric value to assign if |
setRasterNodataValue |
Logical. |
usePixelLonLat |
This argument is deprecated and will be removed in a
future version. Variable names |
write_mode |
Character. Name of the file write mode for output. One of:
|
quiet |
Logical scalar. If |
Details
The variables in expr are vectors of length raster xsize
(row vectors of the input raster layer(s)).
The expression should return a vector also of length raster xsize
(an output row).
Four special variable names are available in expr:
pixelX and pixelY provide pixel center coordinates in projection units.
pixelLon and pixelLat can also be used, in which case the pixel x/y
coordinates will be inverse projected to longitude/latitude
(in the same geographic coordinate system used by the input projection,
which is read from the first input raster). Note that inverse projection
adds computation time.
To refer to specific bands in a multi-band input file, repeat the filename in
rasterfiles and specify corresponding band numbers in bands, along with
optional variable names in var.names, for example,
rasterfiles = c("multiband.tif", "multiband.tif")
bands = c(4, 5)
var.names = c("B4", "B5")
Output will be written to dstfile. To update a file that already
exists, set write_mode = "update" and set out_band to an existing
band number(s) in dstfile (new bands cannot be created in dstfile).
To write multiband output, expr must return a vector of values
interleaved by band. This is equivalent to, and can also be returned as,
a matrix m with nrow(m) equal to length() of an input vector, and
ncol(m) equal to the number of output bands. In matrix form, each column
contains a vector of output values for a band.
length(m) must be equal to the length() of an input vector multiplied by
length(out_band). The dimensions described above are assumed and not
read from the return value of expr.
Value
Returns the output filename invisibly.
See Also
GDALRaster-class, combine(), rasterToVRT()
Examples
## Using pixel longitude/latitude
# Hopkins bioclimatic index (HI) as described in:
# Bechtold, 2004, West. J. Appl. For. 19(4):245-251.
# Integrates elevation, latitude and longitude into an index of the
# phenological occurrence of springtime. Here it is relativized to
# mean values for an eight-state region in the western US.
# Positive HI means spring is delayed by that number of days relative
# to the reference position, while negative values indicate spring is
# advanced. The original equation had elevation units as feet, so
# converting m to ft in `expr`.
elev_file <- system.file("extdata/storml_elev.tif", package="gdalraster")
# expression to calculate HI
expr <- "round( ((ELEV_M * 3.281 - 5449) / 100) +
((pixelLat - 42.16) * 4) +
((-116.39 - pixelLon) * 1.25) )"
# calc() writes to a tempfile by default
hi_file <- calc(expr = expr,
rasterfiles = elev_file,
var.names = "ELEV_M",
dtName = "Int16",
nodata_value = -32767,
setRasterNodataValue = TRUE)
ds <- new(GDALRaster, hi_file)
# min, max, mean, sd
ds$getStatistics(band=1, approx_ok=FALSE, force=TRUE)
ds$close()
deleteDataset(hi_file)
## Calculate normalized difference vegetation index (NDVI)
# Landast band 4 (red) and band 5 (near infrared):
b4_file <- system.file("extdata/sr_b4_20200829.tif", package="gdalraster")
b5_file <- system.file("extdata/sr_b5_20200829.tif", package="gdalraster")
expr <- "((B5 * 0.0000275 - 0.2) - (B4 * 0.0000275 - 0.2)) /
((B5 * 0.0000275 - 0.2) + (B4 * 0.0000275 - 0.2))"
ndvi_file <- calc(expr = expr,
rasterfiles = c(b4_file, b5_file),
var.names = c("B4", "B5"),
dtName = "Float32",
nodata_value = -32767,
setRasterNodataValue = TRUE)
ds <- new(GDALRaster, ndvi_file)
ds$getStatistics(band=1, approx_ok=FALSE, force=TRUE)
ds$close()
deleteDataset(ndvi_file)
## Reclassify a variable by rule set
# Combine two raster layers and look for specific combinations. Then
# recode to a new value by rule set.
#
# Based on example in:
# Stratton, R.D. 2009. Guidebook on LANDFIRE fuels data acquisition,
# critique, modification, maintenance, and model calibration.
# Gen. Tech. Rep. RMRS-GTR-220. U.S. Department of Agriculture,
# Forest Service, Rocky Mountain Research Station. 54 p.
# Context: Refine national-scale fuels data to improve fire simulation
# results in localized applications.
# Issue: Areas with steep slopes (40+ degrees) were mapped as
# GR1 (101; short, sparse dry climate grass) and
# GR2 (102; low load, dry climate grass) but were not carrying fire.
# Resolution: After viewing these areas in Google Earth,
# NB9 (99; bare ground) was selected as the replacement fuel model.
# look for combinations of slope >= 40 and FBFM 101 or 102
lcp_file <- system.file("extdata/storm_lake.lcp", package="gdalraster")
rasterfiles <- c(lcp_file, lcp_file)
var.names <- c("SLP", "FBFM")
bands <- c(2, 4)
tbl <- combine(rasterfiles, var.names, bands)
nrow(tbl)
tbl_subset <- subset(tbl, SLP >= 40 & FBFM %in% c(101,102))
print(tbl_subset) # twelve combinations meet the criteria
sum(tbl_subset$count) # 85 total pixels
# recode these pixels to 99 (bare ground)
# the LCP driver does not support in-place write so make a copy as GTiff
tif_file <- file.path(tempdir(), "storml_lndscp.tif")
createCopy("GTiff", tif_file, lcp_file)
expr <- "ifelse( SLP >= 40 & FBFM %in% c(101,102), 99, FBFM)"
calc(expr = expr,
rasterfiles = c(lcp_file, lcp_file),
bands = c(2, 4),
var.names = c("SLP", "FBFM"),
dstfile = tif_file,
out_band = 4,
write_mode = "update")
# verify the ouput
rasterfiles <- c(tif_file, tif_file)
tbl <- combine(rasterfiles, var.names, bands)
tbl_subset <- subset(tbl, SLP >= 40 & FBFM %in% c(101,102))
print(tbl_subset)
sum(tbl_subset$count)
# if LCP file format is needed:
# createCopy("LCP", "storml_edited.lcp", tif_file)
deleteDataset(tif_file)