snapr

snapr is a proof of concept R package to interact with The ESA Sentinel Applications Platform (SNAP) from R. SNAP provides an impressive set of tools for processing satellite data, many of which are not available in the wider R ecosystem. snapr is a wrapper for SNAP’s Graph Processing Tool (command line interface) and shares similar ambitions to the SNAPISTA python library.

Known Issues:

• The package has only been tested on Linux. It should theoretically work on Windows and MacOS but the automated installation of SNAP is not yet supported on MacOS. On MacOS, snap will have to be installed manually and the path to the SNAP bin folder will have to be set with the SNAPR_BIN environment variable.

• The following snap operators are currently not supported:

BandMaths, BandMerge, BandsDifferenceOp, Binning, DecisionTree, Merge, Mosaic, Multi-size, PixEx, RemoteExecutionOp, SAR-Mosaic, SpectralAngleMapperOp, StatisticsOp, TOPSAR-Merge, Unmix.

To see supported oprators run snapr::get_operators().

• There is little to no validation or checking on the R side for operators and constructing snap graphs. This is something that I believe R will be very well suited to but it is not yet implemented. Therefore we must rely on the somewhat alarming/cryptic messages for the SNAP command line interface.

Installation

You can install the development version of snapr like so:

#install.packages("pak")
pak::pkg_install("Permian-Global-Research/snapr")
library(snapr)
install_snap()

# This is only needed for pacakge development but also if you want access to the 
# snappy python library with {reticulate} for example.
# configure_snappy_python() 

Example

This is a basic example which shows you how to solve a common problem:

library(snapr)
library(terra)
#> terra 1.7.71

mt_st_helens_s1 <- system.file("s1/mt_st_helens_s1.tif", package = "snapr")

out_file <- tempfile(fileext = ".tif")

sg <- snap_graph(
  op_read(
    operator_id = "Reader",
    file = mt_st_helens_s1
  ),
  op_speckle_filter(
    operator_id = "SpeckleFilter",
    sourceProduct = "Reader",
    filter = "Refined Lee"
  ),
  op_write(
    operator_id = "Writer",
    sourceProduct = "SpeckleFilter",
    file = out_file,
    formatName = "GeoTIFF"
  )
)
show_xml(sg)
#> XML process graph: 
#> 
#> <?xml version="1.0" encoding="UTF-8"?>
#> <graph>
#>   <version>1.0</version>
#>   <node id="Reader">
#>     <operator>Read</operator>
#>     <sources/>
#>     <parameters>
#>       <file>/tmp/Rtmpq5pGRz/temp_libpath3c4f3fb1e084/snapr/s1/mt_st_helens_s1.tif</file>
#>       <formatName/>
#>       <pixelRegion/>
#>       <geometryRegion/>
#>       <copyMetadata>true</copyMetadata>
#>       <sourceBands/>
#>       <sourceMasks/>
#>     </parameters>
#>   </node>
#>   <node id="SpeckleFilter">
#>     <operator>Speckle-Filter</operator>
#>     <sources>
#>       <sourceProduct refid="Reader"/>
#>     </sources>
#>     <parameters>
#>       <sourceBands/>
#>       <filter>Refined Lee</filter>
#>       <filterSizeX>3</filterSizeX>
#>       <filterSizeY>3</filterSizeY>
#>       <dampingFactor>2</dampingFactor>
#>       <estimateENL>false</estimateENL>
#>       <enl>1</enl>
#>       <numLooksStr>1</numLooksStr>
#>       <windowSize>7x7</windowSize>
#>       <targetWindowSizeStr>3x3</targetWindowSizeStr>
#>       <sigmaStr>0.9</sigmaStr>
#>       <anSize>50</anSize>
#>     </parameters>
#>   </node>
#>   <node id="Writer">
#>     <operator>Write</operator>
#>     <sources>
#>       <sourceProduct refid="SpeckleFilter"/>
#>     </sources>
#>     <parameters>
#>       <file>/tmp/RtmpfuqYhT/file2423e5474d882.tif</file>
#>       <formatName>GeoTIFF</formatName>
#>       <deleteOutputOnFailure>true</deleteOutputOnFailure>
#>       <writeEntireTileRows>false</writeEntireTileRows>
#>       <clearCacheAfterRowWrite>false</clearCacheAfterRowWrite>
#>     </parameters>
#>   </node>
#> </graph>

suppressMessages(run_graph(sg))

all_bands <- c(rast(mt_st_helens_s1), rast(out_file))
names(all_bands) <- c(
  "Original-VH", "Original-VV",
  "speckle-filter-VH", "speckle-filter-VV"
)

par(mfrow = c(2, 2))
p <- lapply(all_bands, \(x){
  title <- names(x)
  range <- terra::minmax(x, compute = TRUE)
  plot(x, main = title, col = hcl.colors(100, "mako", rev = TRUE), range = range)
})