Create a surface from “pre-gridded” points

I have a large data.frame which has 3 variables Longitude, Latitudeand Temp.

The data is arranged so that it is regularly spaced on a "grid" of 1/4 degree - so that dput(head(dat)) gives:

structure(list(Longitude = c(0.125, 0.375, 0.625, 0.875, 1.125, 
1.375), Latitude = c(0.125, 0.125, 0.125, 0.125, 0.125, 0.125
), Temp = c(25.2163, 25.1917, 25.1593, 25.125, 25.0908, 25.0612
)), .Names = c("Longitude", "Latitude", "Temp"), row.names = c(NA, 
6L), class = "data.frame").

I am having problems re-arranging it to the required format.

I would like to create a regular surface object (typically a list), where x and y are the grid values and z is a corresponding matrix of the surface. This is the usual format used by persp, contour, imageetc.

Using this surface object I will could then be able to easily interpolate to a matrix of locations using interp.surffrom the fieldspackage.

Any suggestions would be great.

Suppose your data is like

set.seed(123)
d <- data.frame(lon=rep(seq(0,1,0.25), times=5),
           lat=rep(seq(0,1,0.25), each=5),
           temp=sample(1:25, 25, replace=TRUE))
head(d, 8)
#    lon  lat temp
# 1 0.00 0.00    8
# 2 0.25 0.00   20
# 3 0.50 0.00   11
# 4 0.75 0.00   23
# 5 1.00 0.00   24
# 6 0.00 0.25    2
# 7 0.25 0.25   14
# 8 0.50 0.25   23

We create a z matrix that represent the values for each point in the grid. We then put the locations of the grid lines (x and y) into a list, together with z.

library(reshape2)
z <- acast(d, lat~lon, value.var="temp")
X <- list(x=sort(unique(d$lon)), 
          y=sort(unique(d$lat)), 
          z=z)

image(X, col=gray.colors(25))
with(d, text(lon, lat, labels=temp))