R statistics/rgrass

From GRASS-Wiki
Revision as of 14:39, 22 October 2022 by Neteler (talk | contribs) (Selected updates; more needed!)
Jump to navigation Jump to search

Terminology

Using R in conjunction with GRASS GIS can have two meanings:

  • Using R within GRASS GIS session, i.e. you start R (or RStudio) from the GRASS GIS command line. You may like this variant if you are primarily a GRASS GIS user.
  • Using GRASS GIS within a R session, i.e. you connect to a GRASS GIS location/mapset from within R (or RStudio). You may like this variant if you are primarily a R user.


References: see "Overview" in R_statistics

Installation

See R_statistics/Installation

R within GRASS

Using R within GRASS GIS session, i.e. you start R (or RStudio) from the GRASS GIS command line.

Startup

  • First start a GRASS GIS session. Then, at the GRASS command line start R (for a 'rstudio' session, see below)
In this example we will use the North Carolina sample dataset.

Reset the region settings to the defaults

GRASS (nc_spm_08_grass7):~ > g.region -d

Launch R from the GRASS prompt

GRASS (nc_spm_08_grass7):~ > R

Load the rgrass library:

library(rgrass)

If you plan to follow the example using the North Carolina sample dataset, load the rgdal library:

library(rgdal)

Get the GRASS environment (mapset, region, map projection, etc.); you can display the metadata for your location by printing G:

G <- gmeta()
G
gisdbase    /home/mneteler/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]]],
        PRIMEM["Greenwich",0,
            ANGLEUNIT["degree",0.0174532925199433]],
        ID["EPSG",4152]],
    CONVERSION["SPCS83 North Carolina zone (meters)",
        METHOD["Lambert Conic Conformal (2SP)",
            ID["EPSG",9802]],
        PARAMETER["Latitude of false origin",33.75,
            ANGLEUNIT["degree",0.0174532925199433],
            ID["EPSG",8821]],
        PARAMETER["Longitude of false origin",-79,
            ANGLEUNIT["degree",0.0174532925199433],
            ID["EPSG",8822]],
        PARAMETER["Latitude of 1st standard parallel",36.1666666666667,
            ANGLEUNIT["degree",0.0174532925199433],
            ID["EPSG",8823]],
        PARAMETER["Latitude of 2nd standard parallel",34.3333333333333,
            ANGLEUNIT["degree",0.0174532925199433],
            ID["EPSG",8824]],
        PARAMETER["Easting at false origin",609601.22,
            LENGTHUNIT["metre",1],
            ID["EPSG",8826]],
        PARAMETER["Northing at false origin",0,
            LENGTHUNIT["metre",1],
            ID["EPSG",8827]]],
    CS[Cartesian,2],
        AXIS["easting (X)",east,
            ORDER[1],
            LENGTHUNIT["metre",1]],
        AXIS["northing (Y)",north,
            ORDER[2],
            LENGTHUNIT["metre",1]],
    USAGE[
        SCOPE["Engineering survey, topographic mapping."],
        AREA["United States (USA) - North Carolina - counties of Alamance; Alexander; Alleghany; Anson; Ashe; Avery; Beaufort; Bertie; Bladen; Brunswick; Buncombe; Burke; Cabarrus; Caldwell; Camden; Carteret; Caswell; Catawba; Chatham; Cherokee; Chowan; Clay; Cleveland; Columbus; Craven; Cumberland; Currituck; Dare; Davidson; Davie; Duplin; Durham; Edgecombe; Forsyth; Franklin; Gaston; Gates; Graham; Granville; Greene; Guilford; Halifax; Harnett; Haywood; Henderson; Hertford; Hoke; Hyde; Iredell; Jackson; Johnston; Jones; Lee; Lenoir; Lincoln; Macon; Madison; Martin; McDowell; Mecklenburg; Mitchell; Montgomery; Moore; Nash; New Hanover; Northampton; Onslow; Orange; Pamlico; Pasquotank; Pender; Perquimans; Person; Pitt; Polk; Randolph; Richmond; Robeson; Rockingham; Rowan; Rutherford; Sampson; Scotland; Stanly; Stokes; Surry; Swain; Transylvania; Tyrrell; Union; Vance; Wake; Warren; Washington; Watauga; Wayne; Wilkes; Wilson; Yadkin; Yancey."],
        BBOX[33.83,-84.33,36.59,-75.38]],
    ID["EPSG",3358]]

Listing of existing maps

List available vector maps:

execGRASS("g.list", parameters = list(type = "vector"))

List selected vector maps (wildcard):

execGRASS("g.list", parameters = list(type = "vector", pattern = "precip*"))
 precip
 precip_30ynormals
 precip_30ynormals_3d

Save selected vector maps into R vector:

my_vmaps <- execGRASS("g.list", parameters = list(type = "vector", pattern = "precip*"))
attributes(my_vmaps)
attributes(my_vmaps)$resOut
 [1] "precip"               "precip_30ynormals"    "precip_30ynormals_3d"

List available raster maps:

execGRASS("g.list", parameters = list(type = "raster"))

List selected raster maps (wildcard):

execGRASS("g.list", parameters = list(type = "raster", pattern = "lsat7_2002*"))
 lsat7_2002_10
 lsat7_2002_20
 lsat7_2002_30
 lsat7_2002_40
 lsat7_2002_50
 lsat7_2002_61
 lsat7_2002_62
 lsat7_2002_70
 lsat7_2002_80

Reading in data from GRASS

Example 1: raster maps

Manual page: read_RAST()

Read in two GRASS raster maps (the maps "geology_30m" and "elevation" from the North Carolina sample data location) into the R current session as a new object "ncdata" (containing then two SpatialGridDataFrames as well as metadata):

# Avoid this error:
#   Error in .read_rast_non_plugin(vname = vname, NODATA = NODATA, driverFileExt = driverFileExt,  : 
#    no stars import yet
use_sp()

# the cat parameter indicates which raster values to be returned as factors
# - geology_30m is a categorical map (i.e., it comes with classes)
# - elevation is a continuous surface
ncdata <- read_RAST(c("geology_30m", "elevation"), cat=c(TRUE, FALSE))

(A warning may appear since in the "geology_30m" map two labels are not unique - as found in the original data.)

We can verify the new R object:

str(ncdata)
Formal class 'SpatialGridDataFrame' [package "sp"] with 4 slots
 ..@ data       :'data.frame':	16616 obs. of  2 variables:

and also check the data structure in more detail:

str(ncdata@data)
'data.frame':	16616 obs. of  2 variables:
 $ geology_30m: Factor w/ 12 levels "CZfg_217","CZlg_262",..: NA NA NA NA NA NA NA NA NA NA ...
 $ elevation  : num  NA NA NA NA NA NA NA NA NA NA ...

The metadata are now accessed and available, but are not (yet) used to structure the sp class objects, in this case a SpatialGridDataFrame object filled with data from two North Carolina layers. Here is a plot of the elevation data:

image(ncdata, "elevation", col = terrain.colors(20))

Add a title to the plot:

title("North Carolina elevation")

In addition, we can show what is going on inside the objects read into R:

str(G)
List of 26
 $ DEBUG        : chr "0"
 $ LOCATION_NAME: chr "nc_spm_08_grass7"
 $ GISDBASE     : chr "/home/veroandreo/grassdata"
 $ MAPSET       : chr "PERMANENT"
 $ GUI          : chr "wxpython"
 $ projection   : chr "99"
 $ zone         : chr "0"
 $ n            : num 228500
 $ s            : num 215000
 $ w            : num 630000
 $ e            : num 645000
 $ t            : num 1
 $ b            : num 0
 $ nsres        : num 27.5
 $ nsres3       : num 10
 $ ewres        : num 37.5
 $ ewres3       : num 10
 $ tbres        : num 1
 $ rows         : int 491
 $ rows3        : int 1350
 $ cols         : int 400
 $ cols3        : int 1500
 $ depths       : int 1
 $ cells        : chr "196400"
 $ cells3       : chr "2025000"
 $ proj4        : chr "+proj=lcc +lat_1=36.16666666666666 +lat_2=34.33333333333334 +lat_0=33.75 +lon_0=-79 +x_0=609601.22 +y_0=0 +no_defs +a=6378137 +"| __truncated__
 - attr(*, "class")= chr "gmeta"

summary(ncdata)
Object of class SpatialGridDataFrame
Coordinates:
        min    max
[1,] 630000 645000
[2,] 215000 228500
Is projected: TRUE 
proj4string :
[+proj=lcc +lat_1=36.16666666666666 +lat_2=34.33333333333334
+lat_0=33.75 +lon_0=-79 +x_0=609601.22 +y_0=0 +no_defs +a=6378137
+rf=298.257222101 +towgs84=0.000,0.000,0.000 +to_meter=1]
Grid attributes:
  cellcentre.offset cellsize cells.dim
1          630018.8 37.50000       400
2          215013.7 27.49491       491
Data attributes:
   geology_30m      elevation     
 CZfg_217:70381   Min.   : 55.92  
 CZig_270:66861   1st Qu.: 94.78  
 CZbg_405:24561   Median :108.88  
 CZlg_262:19287   Mean   :110.38  
 CZam_862: 6017   3rd Qu.:126.78  
 CZbg_910: 4351   Max.   :156.25  
 (Other) : 4942

Example 2: raster maps

Here an example for a single raster map transfer from GRASS GIS to R:

library(rgrass)
execGRASS("g.region", raster = "elevation", flags = "p")
ncdata <- read_RAST("elevation", cat=FALSE)
summary(ncdata)
spplot(ncdata, col = terrain.colors(20))

Example 3: vector maps

Here an example for a single vector map transfer from GRASS GIS to R:

# needed prerequisite
library(rgrass)
use_sf()
library(sf)

execGRASS("v.info", map="schools", layer="1")
vInfo("schools")
myschools <- read_VECT("schools")
print(summary(myschools))

Summarizing data

We can create a table of cell counts:

table(ncdata$geology_30m)
CZfg_217 CZlg_262 CZig_270 CZbg_405 CZve_583 CZam_720 CZg_766 CZam_862 CZbg_910 Km_921 CZbg_945 CZam_946 CZam_948
70381 19287 66861 24561 2089 467 691 6017 4351 1211 1 398 85

And compare with the equivalent GRASS module:

execGRASS("r.stats", flags=c("c", "l"), parameters=list(input="geology_30m"), ignore.stderr=TRUE)
217 CZfg 70381
262 CZlg 19287
270 CZig 66861
405 CZbg 24561
583 CZve 2089
720 CZam 467
766 CZg 691
862 CZam 6017
910 CZbg 4351
921 Km 1211
945 CZbg 1
946 CZam 398
948 CZam 85

Create a box plot of geologic types at different elevations:

boxplot(ncdata$elevation ~ ncdata$geology_30m, medlwd = 1)

Querying maps

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 first 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)

ncdata$sqdem <- sqrt(ncdata$elevation)

Export data from R back into a GRASS raster map:

writeRAST(ncdata, "sqdemNC", zcol="sqdem", ignore.stderr=TRUE)

Check that it imported into GRASS ok:

execGRASS("r.info", parameters=list(map="sqdemNC"))
 +----------------------------------------------------------------------------+
 | Map:      sqdemNC                        Date: Sun Jul 19 13:06:34 2015    |
 | Mapset:   PERMANENT                      Login of Creator: veroandreo      |
 | Location: nc_spm_08_grass7                                                 |
 | DataBase: /home/veroandreo/grassdata                                       |
 | Title:     ( sqdemNC )                                                     |
 | Timestamp: none                                                            |
 |----------------------------------------------------------------------------|
 |                                                                            |
 |   Type of Map:  raster               Number of Categories: 0               |
 |   Data Type:    DCELL                                                      |
 |   Rows:         491                                                        |
 |   Columns:      400                                                        |
 |   Total Cells:  196400                                                     |
 |        Projection: Lambert Conformal Conic                                 |
 |            N:     228500    S: 215000.0002   Res: 27.49490794              |
 |            E:     645000    W:     630000   Res:  37.5                     |
 |   Range of data:    min = 7.47818253045085  max = 12.5000787351036         |
 |                                                                            |
 |   Data Description:                                                        |
 |    generated by r.in.bin                                                   |
 |                                                                            |
 |   Comments:                                                                |
 |                                                                            |
 +----------------------------------------------------------------------------+

Using RStudio in a GRASS GIS session

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

Using GRASS GIS within a R session, i.e. you connect to a GRASS GIS location/mapset from within R (or RStudio).

Startup

In the first place, find out the path to the GRASS GIS library. This can be easily done with the following command (still outside of R; or through a system() call inside of R):

# OSGeo4W, Linux, Mac OSX users:
grass --config path

It may report something like:

# OSGeo4W users:
C:\OSGeo4W64\apps\grass\grass

# Linux, Mac OSX users:
/usr/local/grass

To call GRASS GIS functionality from R, start R and use the initGRASS() function to define the GRASS settings. If GRASS GIS was installed through OSGeo4W, 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", 
         SG="elevation")

Linux and Mac OSX users just need to start R or RStudio the usual way, load the rgrass package and initialize GRASS GIS with initGRASS() as shown below:

# load rgrass package
library(rgrass)
# initialisation of GRASS in the North Carolina sample dataset
initGRASS(gisBase = "/usr/local/grass", 
          home = tempdir(), 
          gisDbase = "/home/veroandreo/grassdata/",
          location = "nc_spm_08_grass7", 
          mapset = "user1", 
          SG="elevation")

Note: the optional SG parameter is a 'SpatialGrid' object to define the ‘DEFAULT_WIND’ of the temporary location.

# set computational region to default (optional)
system("g.region -dp")
# verify metadata
gmeta()

# get two raster maps into R space
ncdata <- read_RAST(c("geology_30m", "elevation"), cat=c(TRUE, FALSE))

# calculate object summaries
summary(ncdata$geology_30m)
 CZfg_217 CZlg_262 CZig_270 CZbg_405 CZve_583 CZam_720  CZg_766 CZam_862 
     292       78      277      102        8        1        2       25 
 CZbg_910   Km_921 CZam_946     NA's 
      18        5        2  1009790

R in GRASS in batch mode

Run the following script with

R CMD BATCH batch.R
library(rgrass)
# initialisation of GRASS in the North Carolina sample dataset (example with a compiled version)
initGRASS(gisBase = "/home/veroandreo/software/grass/dist.x86_64-pc-linux-gnu", 
          home = tempdir(), 
          gisDbase = "/home/veroandreo/grassdata/",
          location = "nc_spm_08_grass7", mapset = "user1", SG="elevation",
          override = TRUE)
# set region to default
system("g.region -dp")
# verify
gmeta()
# read data into R
ncdata <- read_RAST(c("geology_30m", "elevation"), cat=c(TRUE, FALSE))
# summary of geology map
summary(ncdata$geology_30m)
proc.time()

The result is (shorted here):

cat batch.Rout
    R version 4.0.2 (2020-06-22) -- "Taking Off Again"
    Copyright (C) 2020 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)
    > # initialisation and the use of north carolina dataset
    > initGRASS(gisBase = "/home/veroandreo/software/grass/dist.x86_64-pc-linux-gnu", home = tempdir(), 
    +           gisDbase = "/home/veroandreo/grassdata/",
    +           location = "nc_spm_08_grass7", mapset = "user1", SG="elevation",
    +           override = TRUE)
    gisdbase    /home/veroandreo/grassdata/ 
    location    nc_spm_08_grass7 
    mapset      user1 
    rows        13500 
    columns     15000 
    north       228500 
    south       215000 
    west        630000 
    east        645000 
    nsres       1 
    ewres       1 
    projection  +proj=lcc +lat_1=36.16666666666666 +lat_2=34.33333333333334
    +lat_0=33.75 +lon_0=-79 +x_0=609601.22 +y_0=0 +no_defs +a=6378137
    +rf=298.257222101 +towgs84=0.000,0.000,0.000 +type=crs +to_meter=1  
    
    > 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 
    ...
    > ncdata <- read_RAST(c("geology_30m", "elevation"), cat=c(TRUE, FALSE))
    ...
    > summary(ncdata$geology_30m)
    CZfg_217 CZlg_262 CZig_270 CZbg_405 CZve_583 CZam_720  CZg_766 CZam_862 
         292       78      277      102        8        1        2       25 
    CZbg_910   Km_921 CZam_946     NA's 
          18        5        2  1009790 
    > proc.time()
       user  system elapsed 
      8.061   2.016  10.048

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 $HOME/tmp
export TMP=$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.