r - 为什么 stat_density2d 并不总是正确地描述密度？

Question

我目前正在做一个项目，到目前为止我已经提取了数据。我正在对其中的一些进行可视化。这就是问题开始的地方。我对 ggmap 密度绘图有疑问，尤其是在点数较少的区域。我可以举例说明。

以下代码用于重现该问题：

library(ggmap)
cont <- c(-45,-35)
bins <- 12
higher <- rgb(1,0,0,1)
low <- rgb(0,1,0,0)
gmap <- ggmap(get_googlemap(center = c(cont[1],cont[2]), 
                          zoom = 3, 
                          color = "bw", maptype = "terrain", 
                          style = 
              "styles=feature:all|element:labels|visibility:off"))

csvf <- read.csv(paste("Filename.csv", sep = ""))
csvf_1st <- select(filter(csvf, bcdate == eventdate & 
                   bctime < "0 days 09:00:00" &bctime > "0 days 03:00:00" &
                   lat != 0 & lon != 0), c(lat,lon))
c1 <- filter(csvf, as.Date(bcdate) == as.Date(eventdate)+1 &
             bctime < "0 days 09:30:00"&bctime > "0 days 03:00:00" & 
             lat != 0 & lon != 0)
c2 <- filter(csvf, as.Date(bcdate) == as.Date(eventdate)+2 & 
             bctime < "0 days 09:00:00" &bctime > "0 days 03:00:00"& 
             lat != 0 & lon != 0)
csvf_2nd <- select(c1, c(lat,lon))
csvf_last <-select(c2, c(lat,lon))
day1 <- gmap + stat_density2d(data = csvf_1st, 
                              mapping = aes(x = lon, y = lat, 
                              fill = (..level..), 
                              alpha = (..level..),
                              size = 0.0001),
                              geom = "polygon", bins = bins, na.rm 
                             =TRUE) +
                    scale_fill_gradient(low = low, high = higher) + 
                    geom_point(data = csvf_1st, mapping = aes(x = lon, y = lat))
day2 <- gmap + stat_density2d(data = csvf_2nd, 
                          mapping = aes(x = lon, y = lat, 
                                        fill = (..level..), 
                                        alpha = (..level..),
                                        size = 0.0001),
                          geom = "polygon", bins = bins, 
                          na.rm = TRUE) +
                    scale_fill_gradient(low = low, high = higher) + 
                    geom_point(data = csvf_2nd ,mapping = aes(x = lon, y = lat))
day3 <- gmap + stat_density2d(data = csvf_last, 
                          mapping = aes(x = lon, y = lat, 
                                        fill = (..level..), 
                                        alpha = (..level..),
                                        size = 0.0001),
                          geom = "polygon", bins = bins, na.rm = TRUE) +
                    scale_fill_gradient(low = low, high = higher) + 
                    geom_point(data = csvf_last ,mapping = aes(x = lon, y = lat))
print(day1)
print(day2)
print(day3)

此处的代码使用全球第一天、第二天和第三天的 csv 数据（下一行中的超链接），并仅绘制南美洲的点，并在其顶部绘制密度映射。此链接适用于包含数据点的 csv 文件，应添加到“Filename.csv”中，以添加数据框。

现在您已经有了数据和图表，它们应该是这样的：

我的问题是，为什么第 2 天和第 3 天没有描绘正确的密度？