NAME
r.mess - Computes multivariate environmental similarity surface (MES)
KEYWORDS
similarity,
raster,
modelling
SYNOPSIS
r.mess
r.mess --help
r.mess [-mnkci] env=names[,names,...] [env_proj=names[,names,...]] [ref_rast=name] [ref_vect=name] output=name [digits=string] [--overwrite] [--help] [--verbose] [--quiet] [--ui]
Flags:
- -m
- Calculate Most dissimilar variable (MoD)
- -n
- Area with negative MESS
- -k
- sum(IES), where IES < 0
- -c
- Number of IES layers with values < 0
- -i
- Remove individual environmental similarity layers (IES)
- --overwrite
- Allow output files to overwrite existing files
- --help
- Print usage summary
- --verbose
- Verbose module output
- --quiet
- Quiet module output
- --ui
- Force launching GUI dialog
Parameters:
- env=names[,names,...] [required]
- Reference conditions
- env_proj=names[,names,...]
- Projected conditions
- ref_rast=name
- Reference area (raster)
- Reference areas (1 = presence, 0 or null = absence)
- ref_vect=name
- Reference points (vector)
- Point vector layer with presence locations
- output=name [required]
- Root name of the output MESS data layers
- digits=string
- Precision of your input layers values
- Default: 3
The Multivariate Environmental Similarity (MES) surfaces was
proposed by Elith et al (2010) [1] and originally implemented in the
Maxent software.
The MES provides a measure of the portional distance
of any points (in the projection data) with respect to the range of
individual covariates from reference data. More precisely, the MES
represents how similar a point is to a reference set of points, with
respect to a set of predictor variables (V1, V2, ...). The values in
the MESS are influenced by the full distribution of the reference
points, so that sites within the environmental range of the
reference points but in relatively unusual environments will have a
smaller value than those in very common environments". See the
supplementary materials of Elith et al. (2010) [1] for more details.
r.mess computes the MES and the individual similarity
layers (IES - the user can select to delete these layers) and
optionally a number of other layers that help to further interpret
the MES values
- the area where for at least one of the variables has a value
that falls outside the range of values found in the reference set
- the most dissimilar variable (MoD)
- the sum of the IES layers where IES < 0. This is similar to
the NT1 measure as proposed by Mesgaran et al. 2014 [2]
- the number of layers with negative values
The user can compare a set of reference / baseline conditions
(ref) and projected / test conditions (proj). For the reference
conditions the whole region can be used (no reference areas or
points are given). Alternatively, one can define a set of
reference/sample points (presvect) or reference / sample areas
(presrast) against which other areas are to be compared. The
projected conditions can be future conditions in the same area
(similarity across time), or conditions in another area (similarity
between two different areas). See the examples for more details.
Note that a mask is taken into account when computing the frequency
distribution of the reference data layers, but is removed when
computing the output layers. This means that instead of using a
raster layer to delimit an reference / sample area (
ref_rast,
see example 2), one can use the mask to delimit a reference area,
and compute how similar the areas area outside the mask.
The examples below use the bioclimatic variables bio1 (mean annual
temperature), bio12 (annual precipitation) and bio15 (precipitation
seasonality) in Kenya and Uganda. All climate layers (current and
future) are from http://www.worldclim.org. The protected areas layer
includes all nationally designated protected areas with a IUCN
category of II or higher from
protectedplanet.net.
The simplest case is when only a set of reference data layers (
env
) is provided. The multi-variate similarity values of the
resulting map are a measure of how similar conditions in a location
are to the mediam conditions in the whole region.
g.region raster=bio1
r.mess env=bio1,bio12,bio15 output=Ex_01
Thus, in the maps above, the value in each pixel represents how
similar conditions are in that pixel to the median conditions in the
whole region, in terms of mean annual temperature (bio1), mean
annual precipitation (bio12), precipitation seasonality (bio15) and
the three combined (MES).
In the second example, conditions in the whole region are compared
to those in the region's protected areas (ppa), which thus serves as
the reference/sample area. See
van Breugel et al.(2015)
[3] for an example how this can be useful.
g.region raster=bio1
r.mess -m -n -i env=bio1,bio12,bio15 ref_rast=ppa output=Ex_02
In the figure below the map with the protected areas, the MES, the
most dissimilar variables and the areas with novel conditions are
given. They show that the protected areas cover most of the region's
annual precipitation, mean annual temperature and precipitation
seasonality gradients. Areas with novel conditions can be found in
northern Kenya.
Similarity between long-term average conditions based on the period
1950-2000 (
env) and projections for climate conditions in
2070 under RCP85 based on the IPSL General Circulation Models (
env_proj). No reference points or areas are defined in this
example, so the whole region is used as reference. Note that this is
equivalent to what the Maxent program does when computing the MESS
layers.
g.region raster=bio1
r.mess env=bio1,bio12,bio15 env_proj=IPSL_bio1,IPSL_bio12,IPSL_bio15
output=Ex_03
Results (below) shows that there is a fairly large area with novel
conditions. Note that in the MES map, the values are based on
the highest negative value across the input variables (here bio1,
bio12, bio15). In the SumNeg map, values of all input
variables are summed when negative. The Count map shows for
each raster cell how many variables have a negative similarity
scores. Thus, the values in the MES and SumNeg maps
only differ were the MES of more than one variable is negative (dark
grey areas in the Count map).
[1] Elith, J., Kearney, M., & Phillips, S.
2010. The art of modelling range-shifting species. Methods in
Ecology and Evolution 1:330-342.
[2] Mesgaran, M.B., Cousens, R.D. & Webber, B.L. (2014) Here be
dragons: a tool for quantifying novelty due to covariate range and
correlation change when projecting species distribution models.
Diversity & Distributions, 20: 1147-1159, DOI: 10.1111/ddi.12209.
[3] van Breugel, P., Kindt, R., Lillesø, J.-P.B., & van Breugel,
M. 2015. Environmental Gap Analysis to Prioritize Conservation
Efforts in Eastern Africa. PLoS ONE 10: e0121444.
Paulo van Breugel, paulo at ecodiv.org
Last changed: $Date: 2017-06-05 20:02:55 +0200 (Mon, 05 Jun 2017) $
SOURCE CODE
Available at: r.mess source code (history)
Main index |
Raster index |
Topics index |
Keywords index |
Graphical index |
Full index
© 2003-2018
GRASS Development Team,
GRASS GIS 7.4.1svn Reference Manual