I have a points shapefile and I want to create a rectangular buffer (22 x 16 km) around each point. My goal is to create a 16x22 km grid where each of the points is the centroid of the grid. Seed the attached image for an example of what I want to do.

enter image description here

  • Have you tried gBuffer from the rgeos package?
    – user32309
    Commented Sep 16, 2014 at 9:17
  • As far as I understand, gBuffer only has a "width" argument. I would need to specify length and width separately. Commented Sep 16, 2014 at 9:30
  • 1
    You can convert your points to a raster and then convert that raster back to a vector.
    – johanvdw
    Commented Sep 16, 2014 at 9:39
  • @johanvdw, which tool would you recommend? Commented Sep 16, 2014 at 9:46
  • 1
    In GRASS, check v.mkgrid
    – markusN
    Commented Sep 18, 2014 at 5:13

4 Answers 4


Here is a suggestion using GridTopology from R.


# coordinates of some points
x <- c(44, 66, 88, 22, 44, 66, 44, 66)
y <- c(64, 64, 64, 48, 48, 48, 32, 32)
sp <- SpatialPoints(cbind (x,y))

# dimensions of grid
topleftCorner <- bbox(sp)[,1]
columns <- length(unique(x))
rows <- length(unique(y))
cellWidth <- 22
cellHeight <- 16

# make the grid
grd <- GridTopology(topleftCorner, c(cellWidth,cellHeight), c(columns,rows))
grdPolys <- as.SpatialPolygons.GridTopology(grd) #rectangles

# remove the polys which don't have a point
grdPolys <- grdPolys[over(sp, grdPolys),]

plot(sp, col="red", add=T, pch=1)

enter image description here


Here is a approach made with GRASS on al Linux machine.

Import your vector layer with the point features:

v.in.ogr -o dsn=/your_shapefile_directory layer=your_vector_layer output=points_moved

Then move (in meter) your points with half of the size of the grid you want to have (first X, second Y direction):

v.edit map=points_moved type=point tool=move move=-11000,-8000 where="(id >= 1)"

(The where clause here is optional)

After this create the grid:

for i in `v.out.ascii input=points_moved format=point fs=,`;
do k=$((k+1)); v.mkgrid map="pp$k" grid=1,1 position=coor coor=${i%,*} box=22000,16000; done

In the v.mkgrid function there you can define coordinates on which the grid will be snapped. To get the coordinates from the points use v.out.ascii.Defining coor=${i%,*} in v.mkgrid sets the coordinates for the grids (grid=1,1 > number of rows and columns in grid), avoiding a z-value as a third value from the ASCII output.

The pp* layer you can merge with v.patch.

for i in `g.mlist type=vect pattern="pp*"`; do names="$i,$names"; done
v.patch -e input=${names%,} output=merged_lines

Compare the points, points_moved and the merged layer to see the result.

!!! I have used spatial data with a metric coordinate reference system (DHDN / Gauss-Kruger zone 4)

When you have coordinate reference system like WGS84 (EPSG:4326) you have to use values in degrees.


If you change the "End cap style" to "Square" it will make squares around the point. Your points would have to be perfectly spaced for the grids to connect perfectly though and even then you may have some gaps or overlap.


Would it be just as easy to:

Create the grid layer with the dimensions that you want using the MMQGIS plugin and the Create | Create Grid option.

Create the grid as a polygon.

Select the grids that don't match your points and delete them.

Use the Vector | Geometry Tools | Polygon Centroids tool to 'regain' the centroids

I realise that my method doesn't use your existing centroid points but I wonder if your existing points are already in a set pattern that would easily fit into a grid anyway.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.