A ternary density plot combines a three-component ternary diagram with kernel density estimation to visualize where compositional data concentrates. Instead of showing individual points, this visualization uses a heatmap overlay to reveal the underlying probability distribution of compositions, making it ideal for identifying clusters, modes, and patterns in large compositional datasets.
#' anyplot.ai
#' ternary-density: Ternary Density Plot
#' Library: ggplot2 3.5.1 | R 4.4.1
#' Quality: 87/100 | Created: 2026-05-19
library(ggplot2)
library(MASS)
library(ragg)
set.seed(42)
# Theme tokens
THEME <- Sys.getenv("ANYPLOT_THEME", "light")
PAGE_BG <- if (THEME == "light") "#FAF8F1" else "#1A1A17"
ELEVATED_BG <- if (THEME == "light") "#FFFDF6" else "#242420"
INK <- if (THEME == "light") "#1A1A17" else "#F0EFE8"
INK_SOFT <- if (THEME == "light") "#4A4A44" else "#B8B7B0"
# Equilateral triangle height (side = 1)
TRI_H <- sqrt(3) / 2
# Data: sediment composition (sand / silt / clay) – three geological facies
n <- 600
# Sandy facies: concentrated near sand (top) vertex
sand1 <- rbeta(n, 8, 2)
split1 <- runif(n, 0.25, 0.75)
silt1 <- (1 - sand1) * split1
clay1 <- (1 - sand1) * (1 - split1)
# Silty facies: concentrated near silt (bottom-left) vertex
silt2 <- rbeta(n, 8, 2)
split2 <- runif(n, 0.25, 0.75)
sand2 <- (1 - silt2) * split2
clay2 <- (1 - silt2) * (1 - split2)
# Clayey facies: concentrated near clay (bottom-right) vertex
clay3 <- rbeta(n, 8, 2)
split3 <- runif(n, 0.25, 0.75)
sand3 <- (1 - clay3) * split3
silt3 <- (1 - clay3) * (1 - split3)
sand <- c(sand1, sand2, sand3)
silt <- c(silt1, silt2, silt3)
clay <- c(clay1, clay2, clay3)
# Ternary -> Cartesian: Sand=top (0.5, TRI_H), Silt=bottom-left (0,0), Clay=bottom-right (1,0)
cart_x <- 0.5 * sand + clay
cart_y <- TRI_H * sand
# KDE on Cartesian coordinates (triangle lies in [0,1] x [0, TRI_H])
n_grid <- 300
kde <- kde2d(cart_x, cart_y, n = n_grid, lims = c(0, 1, 0, TRI_H))
density_df <- expand.grid(x = kde$x, y = kde$y)
density_df$z <- as.vector(kde$z)
# Mask density values outside the equilateral triangle
inside <- with(density_df,
y >= 0 &
x >= y / sqrt(3) &
x <= 1 - y / sqrt(3)
)
density_df$z[!inside] <- NA
# Ternary reference grid at 20 / 40 / 60 / 80 % levels (three families of parallels)
grid_segs <- do.call(rbind, lapply(c(0.2, 0.4, 0.6, 0.8), function(f) {
# Parallel to bottom (constant sand = f)
yv <- f * TRI_H
# Parallel to left edge (constant clay = f)
# Parallel to right edge (constant silt = f)
rbind(
data.frame(x = yv / sqrt(3), xend = 1 - yv / sqrt(3), y = yv, yend = yv),
data.frame(x = f, xend = 0.5*(1-f)+f, y = 0, yend = (1-f)*TRI_H),
data.frame(x = 1-f, xend = 0.5*(1-f), y = 0, yend = (1-f)*TRI_H)
)
}))
# Triangle border
tri_border <- data.frame(
x = c(0.5, 0, 1, 0.5),
y = c(TRI_H, 0, 0, TRI_H)
)
# Plot
p <- ggplot() +
geom_raster(
data = density_df,
aes(x = x, y = y, fill = z),
interpolate = TRUE
) +
geom_contour(
data = density_df,
aes(x = x, y = y, z = z),
color = "white",
alpha = 0.55,
linewidth = 0.7,
na.rm = TRUE
) +
geom_segment(
data = grid_segs,
aes(x = x, xend = xend, y = y, yend = yend),
color = INK_SOFT,
alpha = 0.30,
linewidth = 0.4
) +
geom_path(
data = tri_border,
aes(x = x, y = y),
color = INK,
linewidth = 1.4
) +
# Vertex labels
annotate("text", x = 0.5, y = TRI_H + 0.09, label = "Sand",
color = INK, size = 9, fontface = "bold") +
annotate("text", x = -0.09, y = -0.07, label = "Silt",
color = INK, size = 9, fontface = "bold") +
annotate("text", x = 1.09, y = -0.07, label = "Clay",
color = INK, size = 9, fontface = "bold") +
scale_fill_viridis_c(
option = "viridis",
name = "Density",
na.value = "transparent",
guide = guide_colorbar(barwidth = 1.8, barheight = 14,
title.position = "top", title.hjust = 0.5)
) +
coord_fixed(
xlim = c(-0.16, 1.16),
ylim = c(-0.16, TRI_H + 0.18)
) +
labs(
title = "ternary-density · r · ggplot2 · anyplot.ai",
subtitle = "Sediment composition: kernel density of sand / silt / clay proportions across three geological facies"
) +
theme_void(base_size = 14) +
theme(
plot.background = element_rect(fill = PAGE_BG, color = PAGE_BG),
panel.background = element_rect(fill = PAGE_BG, color = NA),
plot.title = element_text(color = INK, size = 24, hjust = 0.5,
margin = margin(t = 14, b = 6)),
plot.subtitle = element_text(color = INK_SOFT, size = 15, hjust = 0.5,
margin = margin(b = 10)),
legend.title = element_text(color = INK, size = 18),
legend.text = element_text(color = INK_SOFT, size = 14),
legend.background = element_rect(fill = ELEVATED_BG, color = NA),
legend.position = "right",
plot.margin = margin(10, 80, 20, 80)
)
# Save
ggsave(
filename = sprintf("plot-%s.png", THEME),
plot = p,
device = ragg::agg_png,
width = 12,
height = 12,
units = "in",
dpi = 300
)
Part of Ternary Density Plot on anyplot.ai.