fields (version 5.02)

poly.image: Image plot for cells that are irregular quadrilaterals.

Description

Creates an image using polygon filling based on a grid of irregular quadrilaterals. This function is useful for a regular grid that has been transformed to another nonlinear or rotated coordinate system. This situation comes up in lon-lat grids created under different map projections. Unlike the usual image format this function requires the grid to be specified as two matrices x and y that given the grid x and y coordinates explicitly for every grid point.

Usage

poly.image(x, y, z, col = tim.colors(64), transparent.color = "white", 
 midpoint = FALSE, zlim = range(z, na.rm = TRUE), 
 xlim = range(x), ylim = range(y), add = FALSE, border=NA,...)
poly.image.regrid(x)

Details

This function is straightforward except in the case when the dimensions of x,y, and z are equal. In this case the relationship of the values to the grid cells is ambigious and the switch midpoint gives two possible solutions. The z values at 4 neighboring grid cells can be averaged to estimate a new value interpreted to be at the center of the grid. This is done when midpoint is TRUE. Alternatively the full set of z values can be retained by redefining the grid. This is accomplisehd by finding the midpoints of x and y grid points and adding two outside rows and cols to complete the grid. The new result is a new grid that is is (M+1)X (N+1) if z is MXN. These new grid points define cells that contain each of the original grid points as their midpoints. Of course the advantage of this alternative is that the values of z are preserved in the image plot; a feature that may be important for some uses.

The function image.plot uses this function internally when image information is passed in this format and can add a legend. In most cases just use image.plot.

The function poly.image.regrid does a simple averaging and extrapolation of the grid locations to shift from midpoints to corners. In the interior grid corners are found by the average of the 4 closest midpoints. For the edges the corners are just extrapolated based on the separation of nieghboring grid cells.

See Also

image.plot

Examples

data(RCMexample)
set.panel( 1,2)
par(pty="s")
# plot with grid modified
poly.image( RCMexample$x, RCMexample$y, RCMexample$z[,,1])

# use midpoints of z
poly.image( RCMexample$x, RCMexample$y, RCMexample$z[,,1],midpoint=TRUE)

# images are very similar. 
set.panel()
# Regridding of x and y
l1<- poly.image.regrid( RCMexample$x)
l2<- poly.image.regrid( RCMexample$y)

# test that this works
i<- 1:10
plot( l1[i,i], l2[i,i])
points( RCMexample$x[i,i], RCMexample$y[i,i],col="red")