I am plotting a large lat-lon NetCDF raster
over an R leaflet
map using shinydashboard
. When I click on the map, a popup comes out and shows row, column, lat-lon position and value of the clicked raster point. (See reproducible code below)
The problem is that I am experiencing a shift in the raster if the raster is large enough. For example, here I clicked on a point which should have a value, but the result is that the identified point is the one above.
I believe this has to do with the fact that the raster used by leaflet
is projected, while the raw data I use to identify the points is Lat-Lon, since the clicked point is returned as Lat-Lon by leaflet
. I cannot use the projected file (depth
) since its units are in meters, not degrees!
Even if I tried to reproject those meters to degrees, I got a shift.
Here is a basic runnable example of the code:
#Libraries
library(leaflet)
library(raster)
library(shinydashboard)
library(shiny)
#Input data
download.file("https://www.dropbox.com/s/y9ekjod2pt09rvv/test.nc?dl=0", destfile="test.nc")
inputFile = "test.nc"
inputVarName = "Depth"
lldepth <- raster(inputFile, varname=inputVarName)
lldepth[Which(lldepth<=0, cells=T)] <- NA #Set all cells <=0 to NA
ext <- extent(lldepth)
resol <- res(lldepth)
projection(lldepth) <- "+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0"
#Project for leaflet
depth <- projectRasterForLeaflet(lldepth)
#Prepare UI
sbwidth=200
sidebar <- dashboardSidebar(width=sbwidth)
body <- dashboardBody(
box( #https://stackoverflow.com/questions/31278938/how-can-i-make-my-shiny-leafletoutput-have-height-100-while-inside-a-navbarpa
div(class="outer",width = NULL, solidHeader = TRUE, tags$style(type = "text/css", paste0(".outer {position: fixed; top: 50px; left: ", sbwidth, "px; right: 0; bottom: 0px; overflow: hidden; padding: 0}")),
leafletOutput("map", width = "100%", height = "100%")
)
)
)
ui <- dashboardPage(
dashboardHeader(title = "A title"),
sidebar,
body
)
#
#Server instance
server <- function(input, output, session) {
output$map <- renderLeaflet({#Set extent
leaflet() %>%
fitBounds(ext[1], ext[3], ext[2], ext[4])
})
observe({#Observer to show Popups on click
click <- input$map_click
if (!is.null(click)) {
showpos(x=click$lng, y=click$lat)
}
})
showpos <- function(x=NULL, y=NULL) {#Show popup on clicks
#Translate Lat-Lon to cell number using the unprojected raster
#This is because the projected raster is not in degrees, we cannot use it!
cell <- cellFromXY(lldepth, c(x, y))
if (!is.na(cell)) {#If the click is inside the raster...
xy <- xyFromCell(lldepth, cell) #Get the center of the cell
x <- xy[1]
y <- xy[2]
#Get row and column, to print later
rc <- rowColFromCell(lldepth, cell)
#Get value of the given cell
val = depth[cell]
content <- paste0("X=",rc[2],
"; Y=",rc[1],
"; Lon=", round(x, 5),
"; Lat=", round(y, 5),
"; Depth=", round(val, 1), " m")
proxy <- leafletProxy("map")
#add Popup
proxy %>% clearPopups() %>% addPopups(x, y, popup = content)
#add rectangles for testing
proxy %>% clearShapes() %>% addRectangles(x-resol[1]/2, y-resol[2]/2, x+resol[1]/2, y+resol[2]/2)
}
}
#Plot the raster
leafletProxy("map") %>%
addRasterImage(depth, opacity=0.8, project=FALSE, group="Example", layerId="Example", colors=colorNumeric(terrain.colors(10), values(depth), na.color = "black"))
}
print(shinyApp(ui, server))
How can I correctly identify the points, if the raster is large?
EDIT: I also wanted to provide some additional links to (possibly) related documentation or questions:
I have found that I can reproject back the X-Y (lon-lat) position given by input$map_click
.
In this case I assumed the input projection to be Lon-Lat, but I think it doesn't necessarily have to be. It just needs to have Lat-Lon units.
#Set projections
inputProj <- "+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0"
leafletProj <- "+proj=merc +lon_0=0 +k=1 +x_0=0 +y_0=0 +a=6378137 +b=6378137 +towgs84=0,0,0,0,0,0,0 +units=m +nadgrids=@null +wktext +no_defs"
#Note that for some reason "+nadgrids=@null +wktext" is very important
#as hinted to by other questions and answers linked in my question.
xy <- SpatialPoints(data.frame(x,y))
proj4string(xy) <- inputProj
xy <- as.data.frame(spTransform(xy, leafletProj))
#Get the cell number from the newly transformed metric X and Y.
cell <- cellFromXY(depth, c(xy$x, xy$y))
#At this point, you can also retrace back the center of the cell in
#leaflet coordinates, starting from the cell number!
xy <- SpatialPoints(xyFromCell(depth, cell))
proj4string(xy) <- leafletProj
xy <- as.data.frame(spTransform(xy, inputProj))
#Here XY will again be in lat-lon, if you projection says so,
#indicating the center of the clicked cell
来源:https://stackoverflow.com/questions/37523323/identify-position-of-a-click-on-a-raster-in-leaflet-in-r