Read two GRASS raster maps, "geology_30m" and "elevation", from the North Carolina sample data location into the R current session as a new object "ncdata". This object is a SpatRaster layer from terra package:
Read two GRASS raster maps, "geology_30m" and "elevation", from the North Carolina sample data location into the R current session as a new object "ncdata". This object is a SpatRaster layer from terra package:
Line 163:
Line 161:
</source>
</source>
[[Image:ncdata.png|center]]
[[Image:ncdata_new.png|center]]
In addition, we can show what is going on inside the objects read into R:
In addition, we can show what is going on inside the objects read into R:
Line 363:
Line 361:
[[File:Grass7_rstudio.png|thumb|500px|center|RStudio used in GRASS GIS session]]
[[File:Grass7_rstudio.png|thumb|500px|center|RStudio used in GRASS GIS session]]
== GRASS within R ==
== GRASS within R ==
== GRASS within R ==
Line 371:
Line 367:
The first argument in initGRASS() is gisBase=. This argument refers to the path where GRASS is installed. In the latest rgrass
The first argument in initGRASS() is gisBase=. This argument refers to the path where GRASS is installed. In the latest rgrass
version, if the gisBase argument is missing, initGRASS() will do a minimal effort of guessing it. Firstly, if a GRASS_INSTALLATION environment variable is availabe, then its value will automatically be used for gisBase. If not, then the system command <tt>grass --config path</> will be tried to get the value of GISBASE that corresponds to the GRASS installation in the system PATH (if any).
version, if the gisBase argument is missing, initGRASS() will do a minimal effort of guessing it. Firstly, if a GRASS_INSTALLATION environment variable is availabe, then its value will automatically be used for gisBase. If not, then the system command <tt>grass --config path</tt> will be tried to get the value of GISBASE that corresponds to the GRASS installation in the system PATH (if any).
If users, in any case want to pass the path to GRASS installation themselves, they can run:
If users, still want/need to pass the path to GRASS installation themselves, they can run:
<source lang="bash">
<source lang="bash">
Line 430:
Line 426:
override = TRUE)
override = TRUE)
</source>
</source>
=== Temporary location ===
=== Temporary location ===
Line 438:
Line 433:
<source lang="rsplus">
<source lang="rsplus">
library(terra)
x <- rast()
x <- rast()
x
x
Line 447:
Line 443:
</source>
</source>
Then, we use that object to provide an EPSG to our temporary location via the SG= parameter and start GRASS:
Then, we use that object to provide an EPSG to our temporary location via the SG= parameter and we start GRASS:
<source lang="rsplus">
<source lang="rsplus">
Line 455:
Line 451:
</source>
</source>
After that, users can write raster and vector maps into GRASS with write_RAST() and write_VECT(), use GRASS GIS modules over those data with execGRASS() and export the results back to R with read_RAST() and read_VECT() as exemplified above.
After that, users can write their raster and vector maps into GRASS with write_RAST() and write_VECT(), use GRASS GIS modules over those data with execGRASS() and export the results back to R with read_RAST() and read_VECT() as exemplified above.
=== Existing location ===
=== Existing location ===
In the case of existing locations, users will provide the path to them as shown before and read data from GRASS database into R
In the case of existing locations and mapsets, users will provide the path to them via the location= and mapset= parameters, and then read data from GRASS database into R through read_RAST() and read_VECT(), do their analyses or modeling in R and potentially write results back into GRASS with write_RAST() and write_VECT(). See examples above.
through read_RAST() and read_VECT(), do their analyses or modeling in R and potentially write results back into GRASS with write_RAST() and write_VECT().
== R in GRASS in batch mode ==
== R in GRASS in batch mode ==
Line 472:
Line 467:
<source lang="rsplus">
<source lang="rsplus">
library(rgrass)
library(rgrass)
# initialisation of GRASS in the North Carolina sample dataset (example with a compiled version)
Read two GRASS raster maps, "geology_30m" and "elevation", from the North Carolina sample data location into the R current session as a new object "ncdata". This object is a SpatRaster layer from terra package:
In addition, we can show what is going on inside the objects read into R:
summary(ncdata)
geology_30m elevation
Min. :217.0 Min. : 56.01
1st Qu.:217.0 1st Qu.: 94.77
Median :270.0 Median :108.86
Mean :311.2 Mean :110.36
3rd Qu.:270.0 3rd Qu.:126.80
Max. :948.0 Max. :156.07
Example 2: single raster map
Here an example to transfer a single raster map from GRASS GIS to R:
Here an example to transfer a single vector map from GRASS GIS to R. In this case, vector maps from GRASS will be converted into SpatVector objects from the terra package in R:
Sometimes you may want to query GRASS GIS maps from your R session. As an example, here the transfer of raster pixel values at the position of vector points:
# set the computational region to the raster map:
execGRASS("g.region", raster = "elev_state_500m", flags = "p")
# query raster map at vector points, transfer result into R
goutput <- execGRASS("r.what", map="elev_state_500m", points="geodetic_pts", separator=",", intern=TRUE)
str(goutput)
chr [1:29939] "571530.81289275,265739.968425953,,187.8082200648" ...
# parse it
con <- textConnection(goutput)
go1 <- read.csv(con, header=FALSE)
str(go1)
'data.frame': 29939 obs. of 4 variables:
$ V1: num 571531 571359 571976 572391 573011 ...
$ V2: num 265740 265987 267049 267513 269615 ...
$ V3: logi NA NA NA NA NA NA ...
$ V4: Factor w/ 22738 levels "-0.0048115728",..: 6859 6642 6749 6411 6356 6904 7506 7224 6908 7167 ...
summary(go1)
V1 V2 V3 V4
Min. :121862 Min. : 7991 Mode:logical 0 : 723
1st Qu.:462706 1st Qu.:162508 NA's:29939 * : 293
Median :610328 Median :204502 0.3048 : 47
Mean :588514 Mean :200038 0.6096 : 44
3rd Qu.:717610 3rd Qu.:247633 1.524 : 42
Max. :946743 Max. :327186 0.9144 : 23
(Other):28767
Exporting data back to GRASS
Finally, a SpatialGridDataFrame object is written back to a GRASS raster map:
First prepare some data: (square root of elevation)
If you are most used to run R through RStudio, but still want to have all GRASS data available for performing
any analyses without loosing the possibility of still using GRASS command line, you can run:
GRASS>rstudio&
or, if you already are working on a certain RStudio project:
GRASS>rstudio/path/to/project/folder/&
Then, you load rgrass library in your RStudio project
library(rgrass)
and you are ready to go.
RStudio used in GRASS GIS session
GRASS within R
In this case, we start GRASS GIS within an R session, i.e., we use initGRASS() to start GRASS in a temporary location or in an existing GRASS GIS location.
The first argument in initGRASS() is gisBase=. This argument refers to the path where GRASS is installed. In the latest rgrass
version, if the gisBase argument is missing, initGRASS() will do a minimal effort of guessing it. Firstly, if a GRASS_INSTALLATION environment variable is availabe, then its value will automatically be used for gisBase. If not, then the system command grass --config path will be tried to get the value of GISBASE that corresponds to the GRASS installation in the system PATH (if any).
If users, still want/need to pass the path to GRASS installation themselves, they can run:
# OSGeo4W shell, Linux, Mac OSX users:
grass--configpath
It may report something like:
# OSGeo4W users:
C:\OSGeo4W64\apps\grass\grass
# Linux, Mac OSX users:
/usr/local/grass
If GRASS GIS was installed through OSGeo4W, users should first start the OSGeo4W Shell. Note that it may start in C: or a directory
where the user does not have writing permission. Before calling R or RStudio, first change to a directory where you are
sure you have writing permission.
## MS-Windows users (example for an OSGeo4W 64bit installation)
# change to a directory with writing permission
C:\>d:
D:\>cd temp
D:\temp>cd testR
# start R within this directory:
D:\temp\testR>R
# to start RStudio from OSGeo4W Shell, run: "C:/Program Files/RStudio/bin/rstudio.exe"
# load rgrass package
library(rgrass)
# initialisation of GRASS in the North Carolina sample dataset
initGRASS(gisBase = "C:/OSGeo4W64/apps/grass/grass",
gisDbase = "C:/Users/marissa/grassdata/",
location = "nc_spm_08_grass7",
mapset = "user1",
home=tempdir(),
override = TRUE)
Linux and Mac OSX users just need to start R or RStudio the usual way, load the rgrass package and initialize GRASS as shown below:
library(rgrass)
# initialisation of GRASS in the North Carolina sample dataset
initGRASS(gisBase = "/usr/lib64/grass82",
gisDbase = "/home/veroandreo/grassdata/",
location = "nc_basic_spm_grass7",
mapset = "user1",
home=tempdir(),
override = TRUE)
Temporary location
If we don’t pass an existing location to initGRASS(), a temporary GRASS location will be created and set. By default this will be
an XY location, i.e., no CRS defined. To define a CRS for the temporary location, we create a terra SpatRaster object that by default will be EPSG:4326.
After that, users can write their raster and vector maps into GRASS with write_RAST() and write_VECT(), use GRASS GIS modules over those data with execGRASS() and export the results back to R with read_RAST() and read_VECT() as exemplified above.
Existing location
In the case of existing locations and mapsets, users will provide the path to them via the location= and mapset= parameters, and then read data from GRASS database into R through read_RAST() and read_VECT(), do their analyses or modeling in R and potentially write results back into GRASS with write_RAST() and write_VECT(). See examples above.
R in GRASS in batch mode
Run the following script with
RCMDBATCHbatch.R
library(rgrass)
library(terra)
# initialisation of GRASS in the North Carolina sample dataset
initGRASS(gisBase = "/usr/lib64/grass82",
home = tempdir(),
gisDbase = "/home/veroandreo/grassdata/",
location = "nc_spm_08_grass7",
mapset = "user1",
override = TRUE)
# set region to default
system("g.region -dp")
# verify
gmeta()
# read data into R
ncdata <- read_RAST("geology_30m", cat=TRUE)
# summary of geology map
summary(ncdata$geology_30m)
proc.time()
The result is (shorted here):
catbatch.Rout
R version 4.1.3 (2022-03-10) -- "One Push-Up"
Copyright (C) 2022 The R Foundation for Statistical Computing
Platform: x86_64-redhat-linux-gnu (64-bit)
> library(rgrass)
Loading required package: XML
GRASS GIS interface loaded with GRASS version: (GRASS not running)
> library(terra)
terra 1.7.23
> # initialisation of GRASS in the North Carolina sample dataset
> initGRASS(gisBase = "/usr/lib64/grass82",
+ home = tempdir(),
+ gisDbase = "/home/veroandreo/grassdata/",
+ location = "nc_spm_08_grass7",
+ mapset = "user1",
+ override = TRUE)
gisdbase /home/veroandreo/grassdata/
location nc_spm_08_grass7
mapset user1
rows 620
columns 1630
north 320000
south 10000
west 120000
east 935000
nsres 500
ewres 500
projection:
PROJCRS["NAD83(HARN) / North Carolina",
BASEGEOGCRS["NAD83(HARN)",
DATUM["NAD83 (High Accuracy Reference Network)",
ELLIPSOID["GRS 1980",6378137,298.257222101,
> # set region to default
> system("g.region -dp")
projection: 99 (Lambert Conformal Conic)
zone: 0
datum: nad83
ellipsoid: a=6378137 es=0.006694380022900787
north: 320000
south: 10000
west: 120000
east: 935000
nsres: 500
ewres: 500
rows: 620
cols: 1630
cells: 1010600
> gmeta()
gisdbase /home/veroandreo/grassdata/
location nc_spm_08_grass7
mapset user1
rows 620
columns 1630
north 320000
south 10000
west 120000
east 935000
nsres 500
ewres 500
projection:
PROJCRS["NAD83(HARN) / North Carolina",
BASEGEOGCRS["NAD83(HARN)",
DATUM["NAD83 (High Accuracy Reference Network)",
ELLIPSOID["GRS 1980",6378137,298.257222101,
LENGTHUNIT["metre",1]]],
...
...
BBOX[33.83,-84.33,36.59,-75.38]],
ID["EPSG",3358]]
...
> ncdata <- read_RAST("elev_state_500m")
...
r.out.gdal complete. File </home/veroandreo/tmp/RtmpoA1IKV/file2844e28321574.grd> created.
> summary(ncdata)
file2844e28321574
Min. : -6.47
1st Qu.: 14.36
Median : 86.55
Mean : 214.55
3rd Qu.: 252.66
Max. :1878.39
NA's :46188
> proc.time()
user system elapsed
15.574 1.275 16.998
Troubleshooting
Running out of disk space
Linux: A common issue is that /tmp/ is used for temporary files albeit being often a small partition. To change that to a larger directory, you may add to your $HOME/.bashrc the entry:
# change TMP directory of R (note: of course also another directory than suggested here is fine)
mkdir-p$HOME/tmp
exportTMP=$HOME/tmp
The drawback of this approach is that on modern Linux systems the /tmp/ is a fast RAM disk (based on tempfs) while HOME directories are often on slower spinning disks (unless you have a SSD drive). At least you no longer run out of disk space easily.
