[R-sig-Geo] how to find temporally invariant cluster (TIC) using R.

[CRISTIAN ANDRES VERGARA FERNANDEZ]

Hi everyone,

This is my first post so my apologize if I break any of the rules, or good
practices for posting a question. I read the suggestions, but I may make a
mistake anyway.

I am interested in doing a radiometric normalization using the TIC method
described in:

Chen, X,, L. Vierling, and D. Deering. 2005. A simple and effective
radiometric correction method to improve landscape change detection across
sensors and cross time. Remote Sensing of Environment 98, 63-79 pp.

basically what I need to do is to find the temporally invariant cluster
between two images (at least 2). by the means of a pixel based scatterplot
and a point density map. Then I need to identify the value of the pixels
that are in the center of the clusters. In my case, I am using landsat
images (b1 2000 v/s b1 1986)

I would like to perform this analysis using R, but I have not sucessed.

My first approach was to plot directly using the raster package.

plot(b1_2000, b1_1986) However i can not do the density map later using the
basic plot function.

Then I used the smothScatter function which allows to perform a density map
but I could not do it using a raster object, so I exported the raster as
txt in ARCGIS and then I read it in R

reflex1 <- (read.table("00_reflex.txt", sep = " ", skip = 6))
reflex2 <- (read.table("00_reflex.txt", sep = " ", skip = 6))

Later, I did the scatterplot showing the density of the pixels using the
smoothScatter function.

lab.palette <- colorRampPalette(c("blue", "orange", "red"), space = "Lab")
smoothScatter(reflex1, reflex2, colramp = lab.palette)

However, I am not able to set the scale and the color of the density map
easily, I may use the Transformation  and the colramp options but are not
very handy.

Now, I trying to do it using the ggplot2 package using the Stat_density2d
function but it seems that ggplot2 works only with dataframes. I am also
thinking that the  GeoXp package may be useful for this work.

I would like to ask you, which is the best approach for doing this analysis
using R?

Many thanks in advance

Cristián

	[[alternative HTML version deleted]]