What is link2GI?
The link2GI package provides a lightweight linking framework that simplifies access to external GIS and remote-sensing software from R. Its primary purpose is to establish reliable, reproducible connections to command-line interfaces of GRASS GIS, SAGA GIS, Orfeo Toolbox (OTB), and GDAL, without requiring detailed operating-system knowledge.
The package originated from repeated teaching and research scenarios in graduate-level GIS and remote-sensing courses. In restrictive university environments—particularly on Windows systems with limited user privileges—the configuration of external GIS software often becomes the dominant obstacle, long before any actual spatial analysis can begin. link2GI addresses this problem by automating discovery, configuration, and environment setup.
Why link2GI?
R provides a mature and powerful ecosystem for spatial data analysis. Vector data are primarily handled by sf packages, while raster workflows are covered by terra and stars. Specialized formats such as NetCDF are well supported by packages like ncdf4.
Spatial analysis in R is frequently extended via wrapper packages that expose external libraries or algorithms through R-like interfaces, for example geosphere, Distance, igraph, or spatstat. A comprehensive overview of this ecosystem is provided by Geocomputation with R.
Despite this richness, a practical gap remains between R and established open-source GIS and remote-sensing systems. Software such as QGIS, GRASS GIS, SAGA GIS, and OTB provide large collections of mature, optimized, and well-validated algorithms that are not fully replicated within R packages.
Bridging this gap is possible through dedicated wrapper packages. For GRASS GIS (7/8), this role is served by rgrass. For SAGA GIS, RSAGA and Rsagacmd are available. However, for the Orfeo Toolbox, no comprehensive and actively maintained R wrapper exists, despite its importance for large-scale remote-sensing workflows.
GDAL occupies a special position: although its
functionality is widely accessible through R packages such as
sf and terra, these interfaces are bound to
the GDAL version used at build time and do not expose
the full command-line toolchain. As a result, direct access to the
system-installed GDAL binaries—often required for
advanced formats, drivers, or reproducible batch workflows—remains
fragmented and poorly standardized in R.
Consequently, there is no lightweight, version-agnostic R interface that systematically exposes the GDAL command-line utilities while preserving transparency about the actually used binaries. This gap becomes particularly relevant in heterogeneous or multi-user environments, where multiple GDAL installations coexist and reproducibility depends on explicit control over the underlying toolchain.
In practice, users often encounter technical barriers related to operating systems, library dependencies, and version mismatches. GRASS GIS, in particular, requires a strictly defined spatial database and a carefully configured runtime environment. From within R, setting up such an environment—especially on Windows—can be disproportionately complex when the goal is simply to run a single algorithm.
What does “linking” mean?
In the context of link2GI, linking refers to the automated detection, configuration, and validation of all environment variables and paths required to access external GIS software via their command-line interfaces.
The package does not reimplement GIS algorithms and does not replace existing wrappers. Instead, it prepares a consistent and explicit execution environment so that existing tools can be called directly, reproducibly, and transparently from R.
link2GI analyses available installations, identifies compatible versions, and establishes either temporary or persistent environments that comply with the requirements of the respective software. Where possible, it integrates seamlessly with established wrapper packages.
GRASS GIS
GRASS GIS imposes the most stringent requirements. In addition to numerous environment variables, it requires a correctly defined spatial database, including projection, extent, and resolution.
The linkGRASS() function searches for available GRASS
installations, optionally allows interactive version selection, and
initializes a valid GRASS session. Once initialized, users can access
both the rgrass interface and the native GRASS
command-line API within the same R session.
SAGA GIS
SAGA GIS is comparatively straightforward to configure. The
linkSAGA() function locates available installations, sets
the required paths and environment variables, and returns a
configuration object that can be passed to wrapper packages such as
RSAGA. Alternatively, users may directly call the
saga_cmd command-line interface via R’s
system() or system2() functions.
Orfeo Toolbox (OTB)
The Orfeo Toolbox is a powerful framework for remote-sensing tasks such as classification, filtering, and machine-learning-based image analysis. While selected OTB algorithms are accessible through R packages, these interfaces are often limited in scope or not designed for large-scale or production workflows.
link2GI identifies available OTB installations and exposes their command-line interfaces in a structured and reproducible manner. In the absence of a comprehensive R wrapper, the package provides a lightweight introspection mechanism based on OTB’s CLI help system, enabling scriptable and reproducible execution of OTB modules from R.
GDAL
Although GDAL functionality is widely available through R packages,
direct access to GDAL command-line utilities remains important in
several scenarios—particularly following the retirement of
rgdal and related packages.
Differences between GDAL, PROJ, and system libraries can make it difficult to determine which capabilities are actually available on a given machine. link2GI therefore provides explicit discovery and controlled access to system-linked GDAL binaries, allowing users to run GDAL utilities deterministically from R, independent of how GDAL is compiled into other packages.
This approach supports reproducibility, transparent diagnostics, and batch-oriented workflows while deliberately avoiding duplication of existing GDAL-based R functionality.