Background
Climate is the average weather condition for a particular location over
a long period of time, typically at least thirty years (WMO, 2020).
Climate change represents an alteration in the state of the climate as
documented by changes in the mean and/or the variability of its
properties (IPCC, 2018). Measuring changes in climate is by no means
trivial because several dimensions are involved (e.g., measured through
anomalies, difference from the baseline, velocity of change) and several
variables can be used, in isolation or in combination, to quantify the
changes (e.g., temperature, precipitation, wind speed). Adequately
capturing the wealth of climate change manifestations as well as the
ways in which they interact with living systems requires characterising
multiple dimensions of change across multiple variables (Garcia et al.,
2014).
Conceptually, the metrics of climate change can be quantified at a local
(the pixel-level) or regional (involving multiple cells) scales (Garcia
et al., 2014). The former has a temporal dimension, while the latter
involves both spatial and temporal dimensions. For example, the local
metrics, such as anomalies (e.g., Araújo et al., 2008), standardized
anomalies (e.g., Williams & Jackson, 2007), probability of extreme
events (e.g., Jiménez et al., 2011), and changes in seasonality (e.g.,
Lane et al., 2012), quantify how mean or extreme values in the time
series of climate variables are altered in a given locality (a grid cell
in a raster dataset) over time. In turn, the regional metrics, such as
the emergence of novel climates (Williams et al., 2001), changes in
areas of analogous climate (Ohlemüller et al., 2008), changes in
distance to analogous climate (Nogués-Bravo et al., 2010), and climate
change velocity (Loarie et al., 2009; VanDerWal et al., 2013), first
characterize a climate dimension across a given region and then measure
local changes in the availability of the climate dimension relative to
the regional pattern.
Climate change metrics can be used as proxies for more detailed
assessments of climate change impacts on biodiversity, such as those
used in species distribution modelling (Garcia et al., 2016). They have
also been related to quantities of past extinctions (Nogués-Bravo et
al., 2010) and the occurrence of areas with high concentrations of
species with restricted range sizes (Ohlemüller et al., 2008; Sandel et
al., 2011).
Despite the abundance of climate change metrics and their potential
links with biodiversity (Garcia et al., 2014), empirical studies linking
climate change metrics to biodiversity dynamics are still limited. Most
often, explorations of how climate change metrics relate to biodiversity
patterns are based on a limited number of variables (e.g., anomalies),
although evidence exists that multiple metrics can help capture a wider
range of biodiversity patterns (e.g., González-Trujillo et al., 2023).
One practical reason that has limited more comprehensive explorations of
the relationship of climate metrics with biodiversity patterns is that
no convenient platform exists where all commonly used metrics are
implemented on equal footing. Moreover, the existing tools to quantify
the metrics are scattered, and substantially different in the way they
handle spatiotemporal data, their input and output, and their
user-friendliness. For example, the velocity of climate change is
implemented by the “VoCC” R package (García Molinos et al.,
2019), while the “analogues ” (Hooker et al. 2011) and
“extRemes” (Eric Gilleland, 2021) R packages can measure novel
climates and extreme value analysis, respectively. It follows that each
one of these packages uses a different interface and different input and
output data formats and most of them do not support spatial data. To
facilitate the exploration of climate change metrics and overcome
existing barriers, we provide a unified interface in theclimetrics R package that enables straightforward quantification
and comparison of six different climate change metrics: 1) Standardized
local anomalies; 2) Changes in probabilities of local climate extremes;
3) Changes in areas of analogous climates; 4) Novel climates; 5) Change
in distances to analogous climates; and 6) Climate change velocity.
We now provide three supporting functions (apply.month, kgc and
temporalTrend ) to aggregate time series of climate data for each month
(i.e., generate 12 outcomes corresponding to 12 months), enabling the
classification of Köppen climate zones, and measuring temporal trend at
the pixel-level for a given climate variable (slope of changes over
time), respectively.
We also designed the functions in the climetrics R package, so
that they are user-friendly, providing flexible handling of multiple
data formats (e.g., raster or raster time series) while generating
outputs as raster maps. In addition, the package is linked to the
“rts” R package (Naimi, 2021) for handling raster time series
data. The “rts” package uses the new R package “terra”(Hijmans, 2021b), for manipulating raster data in a very efficient way
(i.e., it is substantially faster than many other R packages as its
functionalities have been implemented using the C++ programming
language). The “terra” package uses a well-known GDAL library
for handling several common raster formats (e.g., GeoTiff, netCDF,
etc.). Therefore, climetrics can quantify climate change metrics
with high computational performance.