mapImage() with filledContour producing weird patterns

[mdupilka]

Starting about a months or so ago, using filledContour = TRUE in the mapImage() has been producing weird patterns. Here is an example

Here is the sample code to produce it

require(oce)
data("coastlineWorldMedium", package = "ocedata")
int <- .1
lon <- seq(-179.95, 179.95, int)
lat <- seq(-39.95, 89.95, int) 
m <- matrix(0, nrow = length(lon), ncol = length(lat))
mapPlot(coastlineWorldMedium,
        type = "polygon",
        col = "grey95",
        #projection="+proj=laea +lat0=40 +lat1=60 +lon_0=-110",
        projection = "+proj=lcc +lat_1=30 +lat_2=45 +lon_0=-110",
        #longitudelim = c(minlon, maxlon), latitudelim = c(minlat, maxlat),
        longitudelim = c(-140, -80), latitudelim = c(45, 70),
        polarCircle = 0,
        clip = TRUE, drawBox = TRUE,
        cex.axis = .7  )

mapImage(longitude = lon, latitude = lat, z = m,
         filledContour = TRUE,
         breaks = 10, 
         col = colorRampPalette(c("blue", "red"))(10) )

With filledContour = FALSE this does not happen

[mdupilka]

That worked. Interesting, if I set the fillgap > 0 and leave filledContour = TRUE I get a clean plot.

[dankelley]

Note that the right fillgap value is going to depend on the projection and the view being shown, etc.

PS. I am cleaning up some of the docs. This also involves changing what I was renaming gap in fillGapMatrix() to be fillgap. That's where the confusion came from -- I had decided on a given name for one thing and neglected to copy that same name for another thing that was really the same. I will rerun tests before I push to GH though.

[mdupilka]

I did more testing. Using fillgap=2 and filledContour = TRUE looks to fit my needs for a clean plot

[dankelley]

I made a blog posting comparing the methods. Sometimes I think this kind of thing is useful.

@mdupilka perhaps you can consider closing the issue after you feel it has been addressed. I think I'm done with my flurry of coding, unless there are new problems. In the oce community, we ask issue reporters to close issues. That just seems to be more respectful of reporters. (Related problems that do not quite match a given title should be in new issues, so please don't be shy on that.)

[dankelley]

Sorry, forgot to say that the blog post is at https://dankelley.github.io/dek_blog/2024/03/07/mapimage.html

[dankelley]

Thanks, @mdupilka and @richardsc, for your patience in this. I have merged the "2199" branch into the "develop" branch and then deleted "2199". So, if either of you is now on the "2199" branch, you'll need to hop over to "develop" before trying to "git pull" for any new results.