R/plotKML.SpatialPredictions.R

In Envirometrix/plotKML: Visualization of Spatial and Spatio-Temporal Objects in Google Earth

setMethod("plotKML", "SpatialPredictions", function(
  obj,
  folder.name = normalizeFilename(deparse(substitute(obj, env=parent.frame()))),
  file.name = paste(folder.name, ".kml", sep=""),
  colour,
  grid2poly = FALSE,
  obj.summary = FALSE,
  plot.svar = FALSE,
  pngwidth = 210, 
  pngheight = 580,
  pngpointsize = 14,
  metadata = NULL,
  kmz = get("kmz", envir = plotKML.opts),
  open.kml = TRUE,
  ...
){

  ## Guess aesthetics if missing:
  varname <- paste(obj@variable)
  if(missing(colour)){ 
    obj@predicted@data[,"colour"] <- obj@predicted@data[,varname]
  } else {
    if(is.name(colour)|is.call(colour)){
      obj@predicted@data[,"colour"] <- eval(colour, obj@predicted@data)
    } else {
      obj@predicted@data[,"colour"] <- obj@predicted@data[,as.character(colour)]      
    }
  }
  pred <- obj@predicted["colour"]
  
  ## sampling locations:
  locs <- obj@observed
  labs <- paste(signif(locs@data[,varname], 3))
    
  ## summary properties of the RK model:
  if(obj.summary==TRUE){
    xx <- summary(obj)
    xd <- unlist(xx[!names(xx) %in% c("bonds", "breaks")])
    md <- data.frame(Names=attr(xd, "names"), Values=xd,  stringsAsFactors = FALSE)
    html <- kml_description(md, asText = TRUE, cwidth = 120, twidth = 240)
  } 
  
  if(grid2poly == TRUE){
    pol <- grid2poly(pred)
  }

  kml_open(folder.name = folder.name, file.name = file.name)
  
  if(obj.summary==TRUE){
    ## add a description for the whole folder:
    kml.out <- get("kml.out", envir=plotKML.fileIO)
    description_txt <- sprintf('<description><![CDATA[%s]]></description>', html)
    parseXMLAndAdd(description_txt, parent=kml.out[["Document"]])  
    assign('kml.out', kml.out, envir=plotKML.fileIO)
  }
  
  if(grid2poly == TRUE){ 
    kml_layer(pol, colour = colour, ...)
  } else {
    kml_layer(pred, colour = colour, raster_name = paste(varname, "_predicted.png", sep=""), metadata = metadata, ...)
  }
  
  if(plot.svar==TRUE){
  ## plot the prediction variance?
    svarname <- paste(obj@variable, ".", "svar", sep="")
    svar <- obj@predicted[svarname]
    names(svar) <- "colour"
    kml_layer(svar, colour = colour, colour_scale = get("colour_scale_svar", envir = plotKML.opts), raster_name = paste(svarname, "_svar.png", sep=""), plot.legend = FALSE)
  } 
  
  kml_layer(obj = locs, points_names = labs)

  ## plot the correlation graph and variogram:
  if(any(names(obj@validation) %in% "var1.pred")){
    if(all(!is.na(obj@validation$var1.pred)) & all(!is.na(obj@validation$observed))){
      png(filename=paste(varname, "_gstatplots.png", sep=""), width=pngwidth, height=pngheight, bg="white", pointsize=pngpointsize)
      plot.SpatialPredictions(obj, plot.predictions=FALSE, vertical=TRUE)
      dev.off()
      
      ## add the SpatialPredictions plot:
      kml_screen(image.file = paste(varname, "_gstatplots.png", sep=""), position = "LL", sname = "gstatModel summary plot")
    }
  }
  
  ## close the file:
  kml_close(file.name = file.name)
  if (kmz == TRUE){
      kml_compress(file.name = file.name)
  }
  ## open KML file in the default browser:
  if(open.kml==TRUE){
    kml_View(file.name)
  } else {
    message(paste("Object written to:", file.name))
  }
})