[英]R: how to create a heat map of averaged values from a grid and plot it with ggplot?
我有一個包含超過 50 000 個值的數據框(見下文),每個值都與一個位置(緯度、經度)相關聯。 我想計算 5° 緯度 x 5° 經度網格的每個單元格的平均值,以創建熱圖。 最終目標是在測深圖上繪制此網格。
我查看了類似這樣的問題 one Average values of a point dataset to a grid dataset 。 但我無法用我自己的數據復制這些例子。 可悲的是,我被困在創建網格的第一步。
我的數據是這樣的:
library(sp)
library(proj4)
coordinates(data) <- c("lon", "lat")
proj4string(data) <- CRS("+init=epsg:4326") #defined CRS to WGS 84
df<- data.frame(data)
> head(df)
lon lat value
1 -48.1673562 57.71791 822.9
2 -48.7430053 57.83568 1302.3
3 -48.5662663 57.82087 1508.0
4 -48.3252052 58.29815 224.0
5 -47.1716772 58.42417 38.0
6 -46.4098311 58.67651 431.2
7 -45.8071218 58.70022 365.6
8 -45.5558936 58.46975 50.0
理想情況下,我想使用 ggplot2 在來自 marmap 包的地圖上繪制網格(見下文):
library(marmap)
library(ggplot2)
atlantic <- getNOAA.bathy(-80, 40, 0, 90, resolution = 25, keep = TRUE)
atl.df <- fortify(atlantic)
map <- ggplot(atl.df, aes(x=x, y=y)) +
geom_raster(aes(fill=z), data=atl.df) +
geom_contour(aes(z=z),
breaks=0, #contour for continent
colour="black", size=1) +
scale_fill_gradientn(values = scales::rescale(c(-5000, 0, 1, 2400)),
colors = c("steelblue4", "#C7E0FF", "gray40", "white"))
聽起來您想將數值變量(緯度和經度)切成均勻的間隔並總結每個間隔內的值。 以下對您有用嗎?
library(dplyr)
df2 <- df %>%
mutate(lon.group = cut(lon, breaks = seq(floor(min(df$lon)), ceiling(max(df$lon)), by = 5),
labels = seq(floor(min(df$lon)) + 2.5, ceiling(max(df$lon)), by = 5)),
lat.group = cut(lat, breaks = seq(floor(min(df$lat)), ceiling(max(df$lat)), by = 5),
labels = seq(floor(min(df$lat)) + 2.5, ceiling(max(df$lat)), by = 5))) %>%
group_by(lon.group, lat.group) %>%
summarise(value = mean(value), .groups = "drop") %>%
mutate(across(where(is.factor), ~as.numeric(as.character(.x))))
樣本數據:
set.seed(444)
n <- 10000
df <- data.frame(lon = runif(n, min = -100, max = -50),
lat = runif(n, min = 30, max = 80),
value = runif(n, min = 0, max = 1000))
> summary(df)
lon lat value
Min. :-99.99 Min. :30.00 Min. : 0.1136
1st Qu.:-87.55 1st Qu.:42.45 1st Qu.: 247.2377
Median :-75.29 Median :55.11 Median : 501.4165
Mean :-75.12 Mean :55.01 Mean : 499.5385
3rd Qu.:-62.69 3rd Qu.:67.63 3rd Qu.: 748.8834
Max. :-50.01 Max. :80.00 Max. : 999.9600
前后數據對比:
gridExtra::grid.arrange(
ggplot(df,
aes(x = lon, y = lat, colour = value)) +
geom_point() +
ggtitle("Original points"),
ggplot(df2,
aes(x = lon.group, y = lat.group, fill = value)) +
geom_raster() +
ggtitle("Summarised grid"),
nrow = 1
)
就像(幾乎!)總是一樣,有一個功能。 我相信marmap::griddify()
是您正在尋找的。 幫助文件指出:
將不規則間隔的 xyz 數據轉換為適合創建具有規則間隔經度和緯度的深海對象的柵格對象。
這是使用您的坐標的腳本:
library(marmap)
library(ggplot2)
# Create fake data
set.seed(42)
n <- 10000
data_irregular <- data.frame(lon = runif(n, min = -80, max = 40),
lat = runif(n, min = 0, max = 90),
value = runif(n, min = 0, max = 1000))
# Fit data into a grid of 30 cells in longitude and 50 cells in latitude
data_grid <- as.bathy(griddify(data_irregular, nlon = 30, nlat = 50))
fortified_grid <- fortify(data_grid)
# Get bathymetric data to plot continent contours
atlantic <- getNOAA.bathy(-80, 40, 0, 90, resolution = 25)
atl_df <- fortify(atlantic)
# Plot with ggplot with gridded data as tiles
map <- ggplot(atl_df, aes(x = x, y = y)) +
geom_raster(data = fortified_grid, aes(fill = z)) +
geom_contour(data = atl_df, aes(z = z),
breaks = 0, # contour for continent
colour = "black", size = 1) +
scale_fill_gradientn(values = scales::rescale(c(-5000, 0, 1, 2400)),
colors = c("steelblue4", "#C7E0FF", "gray40", "white")) +
labs(x = "Longitude", y = "Latitude", fill = "Value")
map +
theme_bw()
結果如下:
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.