## Installing sen2r

Instruction are provided separately for each OS:

Alternatively, sen2r can be used in a Docker container.

### On Linux systems

#### Installing dependencies

Before installing sen2r, it is recommended to check R (v. 3.4.4 or higher) and the following libraries to be present:

Some external dependencies are required to run specific sen2r features:

• Sen2Cor is used to perform atmospheric correction of Sentinel-2 Level-1C products: it is required by the package, unless you choose not to correct products locally (using only Level-1C – TOA products or downloading directly Level-2A products);
• GDAL is required to smooth/bufferise a cloud mask, and can be optionally used to compute spectral indices, RGB images and thumbnails;
• aria2 is an alternative downloader which can be used to faster the download of SAFE archives; it can be optionally installed and used.

To install both package dependencies and runtime binaries (excepting Sen2Cor, which is installed by sen2r):

• On Ubuntu:

• Ubuntu 20.04 LTS Focal and newer:

sudo apt install -y r-base gdal-bin aria2 libpython2-dev libudunits2-dev libgdal-dev libjq-dev libprotobuf-dev protobuf-compiler libv8-dev libssl-dev libcairo2-dev
• Older distros:

sudo apt-get install -y r-base gdal-bin python-gdal aria2 libpython-dev libudunits2-dev libgdal-dev libjq-dev libprotobuf-dev protobuf-compiler libv8-dev libssl-dev libcairo2-dev

On Ubuntu 16.04 LTS Xenial, the repositories ubuntugis and opencpu must be added before installing the packages:

sudo add-apt-repository -y ppa:opencpu/jq
sudo apt-get update -q

Installation was tested over Ubuntu 20.04 LTS focal and 18.04 LTS Bionic; non-LTS versions > 18.04 and Ubuntu 16.04 LTS Xenial should work. Other releases (Ubuntu non-LTS < 18.04; Ubuntu LTS ≤ 14.04) are not supported.

• On Debian:

sudo apt-get install -y r-base gdal-bin python-gdal aria2 libpython-dev libudunits2-dev libgdal-dev libjq-dev libprotobuf-dev protobuf-compiler libnode-dev libssl-dev libcairo2-dev proj-epsg

Installation was tested over Debian 9 Stretch; older releases are not supported.

• On Fedora:

sudo yum install R gdal gdal-python aria2 python-devels gdal-devel proj-devel geos-devel udunits2-devel jq-devel protobuf-devel v8-devel openssl-devel cairo-devel libcurl-devel netcdf-devel libxml2-devel

Installation was tested over Fedora 29.

• On ArchLinux:

sudo pacman -S --noconfirm r base-devel gcc-fortran python-gdal wget aria2 jq protobuf git
cd /tmp; git clone https://aur.archlinux.org/gyp-git.git
cd gyp-git; makepkg -si --noconfirm
cd /tmp; git clone https://aur.archlinux.org/v8-3.14.git
cd v8-3.14; makepkg -si --noconfirm
cd /tmp; git clone https://aur.archlinux.org/udunits.git
cd udunits; makepkg -si --noconfirm

Installation was tested on date 2018-11-21.

#### Installing sen2r

Before proceeding, it is suggested to update the existing R packages:

update.packages(ask = FALSE)

Doing that, the following warning could appear:

Warning in install.packages(update[instlib == l, "Package"], l, contriburl = contriburl,  :
'lib = "/usr/lib/R/library"' is not writable
Would you like to use a personal library instead?  (y/n)

If so, press y and continue. Moreover, if no personal libraries already exist, it is asked to create a new one: ask y and continue.

After that, the package can be installed from CRAN:

install.packages("sen2r", dependencies = TRUE)

The parameter dependencies = TRUE is used to install also suggested dependencies, which includes Shiny packages required to run the Graphical User Interface. Anyway, it could be omitted as Shiny packages can be installed later, before running the GUI.

Alternatively, the development version can be installed from GitHub with the R package remotes. To do it:

1. install the package remotes, if missing:

install.packages("remotes")
2. load it and install sen2r:

library(remotes)
install_github("ranghetti/sen2r", dependencies = TRUE)

This will install the R package along with its package dependencies, containing all the functions necessary to preprocess data.

Finally, Sen2Cor can be installed from R:

library(sen2r)
install_sen2cor()

### On Windows

R must be installed before proceeding.

In the case it was already installed, it is suggested to update the existing R packages:

update.packages(ask = FALSE)

sen2r can be installed from CRAN:

install.packages("sen2r")

Alternatively, the development version can be installed from GitHub with the R package remotes. To do it:

1. install the package remotes, if missing:

install.packages("remotes")
2. load it and install sen2r:

library(remotes)
install_github("ranghetti/sen2r")

This will install the R package along with its package dependencies, containing all the functions necessary to preprocess data.

### On MacOS

R must be installed before proceeding.

In the case it was already installed, it is suggested to update the existing R packages:

update.packages(ask = FALSE)

Packages can be installed using the compiled binaries, except from sp, sf, rgdal and rgeos, which are recommended to be compiled. To do it, use the command

install.packages(c("sp", "sf", "rgdal", "rgeos"), type = "source")

after having installed GDAL as described above.

sen2r can be installed from CRAN:

install.packages("sen2r")

Alternatively, the development version can be installed from GitHub with the R package remotes. To do it:

1. install the package remotes, if missing:

install.packages("remotes")
2. load it and install sen2r:

library(remotes)
install_github("ranghetti/sen2r")

This will install the R package along with its package dependencies, containing all the functions necessary to preprocess data.

### Installing Sen2Cor

Sen2Cor is used to perform atmospheric correction of Sentinel-2 Level-1C products, so its installation can be skipped in case this function is not required by the user (using only Level-1C – TOA products or downloading directly Level-2A products).

It can be graphically checked / installed launching the function

library(sen2r)
check_sen2r_deps()

This function opens a GUI which help to check if this and other dependencies are satisfied; from this GUI a new Sen2Cor installation can be performed, or an existing environment can be linked to Sen2Cor.

Alternatively, it can be installed from the command line with the following function:

install_sen2cor()

### Installing other dependencies (generally unrequired and discouraged)

Other dependencies can be configure to run specific sen2r features; their installation is generally unrequired (they do not add relevant improvements to the package) and discouraged (errors could occur).

• GDAL is no more required, being used only to smooth/bufferise a cloud mask (and optionally to compute spectral indices, RGB images and thumbnails). On Windows, to avoid incompatibilities, it is strictly mandatory to use a GDAL installation provided with the OSGeo4W installer: if this requirement is not satisfied, it is possible to install GDAL with OSGeo4W directly from R (see below). Note: configuring a runtime GDAL environment could fail for different reasons (e.g. problems related to GDAL environment like missing libraries, or custom environment variables which could conflict). sen2r maintainers are not responsible in case of GDAL-related issues (no assistance is provided in this case).
• aria2 is an alternative downloader which can be used to faster the download of SAFE archives; it can be optionally installed and used. Since the number of concurrent downloads from ESA SciHub is limited to 2, the use of aria2 is generally not faster than the default downloader; for this reason, its use is recommended only in case of problems with the default downloader.

These dependences can be graphically checked launching the function

library(sen2r)
check_sen2r_deps()

Alternatively, they can be installed individually from the command line with the following functions (see their respective references for further details):

check_gdal()
install_aria2()

#### Note on GDAL installation on MacOS

In order to avoid incompatibilities with GDAL, it is strictly mandatory to use a GDAL installation provided with the OSGeo Homebrew repository. To do it:

1. if Homebrew is missing on your system, install it opening a terminal and typing:

/usr/bin/ruby -e "\$(curl -fsSL https://github.com/Homebrew/install/raw/master/install)"

(see the Homebrew webpage for details);

2. install GDAL by typing:

brew tap osgeo/osgeo4mac
brew install osgeo-gdal-python

(see the OSGeo Homebrew webpage for details). At this step, an error could be returned in case another GDAL installation was already present; in this case, follow the instruction returned in the terminal to solve it (probably some packages must be unlinked with the instructions brew unlink <package>).