Basic Usage of NLMR
NLMR is an R package
for simulating neutral landscape models (NLMs): synthetic landscapes
used to explore landscape-scale ecological patterns and processes. The
package follows the same general idea as the Python package
NLMpy (see Etherington et al.
2014), but returns results as SpatRaster objects so
they can be used directly with terra and other raster
workflows. Most of the common NLMs described in the literature are
available. The package also includes tools for classifying continuous
outputs into categories.
For more background on neutral landscape models and the package design, see the accompanying publication Sciani, Fritsch, Scherer and Simpkins (2018).
Quickstart
NLMR exposes 15 exported landscape-generation functions
covering 16 algorithm variants. The variants span a spectrum of spatial
structure, from spatially random landscapes (nlm_random)
through intermediate patterned forms to deterministic planar gradients
(nlm_planargradient) (see Palmer
1992). The curdling generator, nlm_curds(), can
produce either random curdling or wheyed curdling, which is why the
package has one more algorithm variant than exported function. For a
full browseable catalog of generators, use
catalog_nlm().
The plot below shows a basic random landscape with no intentional spatial structure.
A simple random landscape can be generated like this:
Combining landscapes
NLM rasters can be combined to create new landscape patterns. A base raster can be added to one or more secondary rasters, with optional scaling factors controlling how strongly each secondary raster contributes. This example combines a gradient with random variation to create a more complex surface.
base <- NLMR::nlm_edgegradient(ncol = 100, nrow = 100)
secondary_1 <- NLMR::nlm_distancegradient(
ncol = 100,
nrow = 100,
origin = c(10, 10, 10, 10)
)
secondary_2 <- NLMR::nlm_random(ncol = 100, nrow = 100)
combined <- base + (secondary_1 + secondary_2 * 0.2)
terra::plot(combined)
Classifying categories
Landscape rasters generated by NLMR are continuous by
default and often fall between 0 and 1. Use
landscape_classify() to convert a continuous raster into
categorical classes when you want a discrete landscape representation.
This example turns a continuous gradient into four discrete classes.
continuous <- NLMR::nlm_edgegradient(ncol = 100, nrow = 100)
classified <- landscape_classify(continuous, n = 4)
terra::plot(classified)