tidyst_kquiver {eks}

Tidy and geospatial kernel density quiver estimate

Description

Tidy and geospatial versions of a kernel density quiver estimate for 2-dimensional data.

Usage

tidy_kquiver(data, thin=5, transf=1/4, neg.grad=FALSE)
st_kquiver(x, thin=5, transf=1/4, neg.grad=FALSE, scale=1)

Arguments

data

tidy kernel density gradient estimate (output from tidy_kdde(deriv_order=1))

x

geospatial kernel density gradient estimate (output from st_kdde(deriv_order=1))

thin

number to thin out estimation grid. Default is 5.

transf

power index in transformation. Default is 1/4.

neg.grad

flag to compute arrows in negative gradient direction. Default is FALSE.

scale

scale factor to normalise length of arrows. Default is 1.

Details

A kernel quiver estimate is a modification of the standard kernel density gradient estimate in *_kdde where the density derivatives are not given in the separate groups as indexed in deriv_group, but as extra columns u (for deriv_group=(1,0)) and v (for deriv_group=(0,1)).

The bandwidth matrix of smoothing parameters is computed as in ks::kdde(deriv_order=1).

Value

The output from tidy_kquiver has the same structure as the input kernel density gradient estimate, with the added columns u,v for the density gradient value in the x-, y-axis. This structure is compatible with the ggquiver::geom_quiver layer function for quiver plots.

Since ggquiver::geom_quiver is not compatible with geom_sf layers, the output from st_kquiver has added columns lon, lat, lon_end, lat_end, len which are required in geom_segment.

Examples

## tidy kernel quiver estimate
data(crabs, package="MASS")
crabs2 <- dplyr::select(crabs, FL, CW)
t1 <- tidy_kde(crabs2)
t2 <- tidy_kdde(crabs2, deriv_order=1)
t3 <- tidy_kquiver(t2, thin=5) 
gt <- ggplot2::ggplot(t1, ggplot2::aes(x=FL, y=CW)) 
gt + geom_contour_filled_ks(colour=1, cont=seq(10,90,by=10)) +
    colorspace::scale_fill_discrete_sequential(alpha=0.5) + 
    ggquiver::geom_quiver(data=t3, ggplot2::aes(u=u, v=v), colour=6)

## geospatial kernel `quiver' estimate
data(wa)
data(grevilleasf)
hakeoides <- dplyr::filter(grevilleasf, species=="hakeoides")
hakeoides_coord <- st_add_coordinates(hakeoides)
s1 <- st_kde(hakeoides)
s2 <- st_kdde(hakeoides, deriv_order=1)
s3 <- st_kquiver(s2, thin=9) 

## base R plot
xlim <- c(1.2e5, 1.1e6); ylim <- c(6.1e6, 7.2e6)
plot(wa, xlim=xlim, ylim=ylim)
plot(s1, add=TRUE, alpha=0.5)
plot(s3$tidy_ks$ks[[1]], add=TRUE, display="quiver")

## geom_sf plot - ggquiver::geom_quiver not compatible with ggplot2::geom_sf layers
## use instead geom_segment 
gs <- ggplot2::ggplot(s1) + ggplot2::geom_sf(data=wa, fill=NA)  +
    ggthemes::theme_map()
gs + ggplot2::geom_sf(data=st_get_contour(s1),
    ggplot2::aes(fill=label_percent(contlabel)), alpha=0.5) + 
    ggplot2::coord_sf(xlim=xlim, ylim=ylim) + 
    ggplot2::geom_segment(data=s3$sf, 
    ggplot2::aes(x=lon, xend=lon_end, y=lat, yend=lat_end), 
    arrow=grid::arrow(length=0.05*s3$sf$len)) + 
    colorspace::scale_fill_discrete_sequential("Heat2")