使用R语言circular包过滤跨0地理方向数据失败求助

问题根源 #

你用between()筛选跨0°的环形方向区间（270°-0°-90°）失效，核心原因是：

• circular类型数据内部以弧度存储，270°对应的弧度（≈4.712）远大于90°对应的弧度（≈1.571）。
• between()是线性区间判断，要求第一个参数小于第二个参数，否则无法匹配任何数据——没有数值能同时满足「大于大值且小于小值」。

解决方案 #

针对环形方向的跨0区间，需要用两个线性区间的逻辑或来筛选，或者使用circular包专门的环形区间判断函数is.inside.circular()。

修改后的可运行代码 #

library(circular)
library(viridis)
library(dplyr)
library(ggplot2)

# 生成模拟数据
df <- data.frame(flight_dir = runif(n = 1000, min = 0, max = 359),
                 hourOfNight = sample(1:8, 1000, replace = TRUE),
                 n_birds = sample(100:3000, 1000, replace = TRUE),
                 mean_headwind = sample(-10:10, 1000, replace = TRUE))

# 转换为circular类型
df <- df %>% mutate(flight_dir = as.circular(flight_dir,
                                             type = "directions",
                                             units = "degrees",
                                             template = "geographics",
                                             modulo = "2pi",
                                             zero = pi/2,
                                             rotation = "clock"))

# 方案1：手动拆分区间（逻辑或）
df_filtered <- df %>% 
  filter(
    # 匹配270°到360° 或 0°到90°的方向
    (flight_dir >= circular(270, units = "degrees", template = "geographics", zero = pi/2, rotation = "clock") | 
       flight_dir <= circular(90, units = "degrees", template = "geographics", zero = pi/2, rotation = "clock")),
    n_birds >= 1000
  )

# 方案2：用is.inside.circular（更推荐，专门处理环形区间）
# interval <- circular(c(270, 90), type = "directions", units = "degrees", 
#                      template = "geographics", zero = pi/2, rotation = "clock", modulo = "2pi")
# df_filtered <- df %>% filter(is.inside.circular(flight_dir, interval), n_birds >= 1000)

# 绘图
ggplot(df_filtered, aes(x = hourOfNight, y = mean_headwind)) +
  geom_hline(yintercept = 0, linetype = "dashed", size = .5) +
  geom_smooth(aes(weight = n_birds), method = "loess") +
  geom_point(aes(size = n_birds, fill = n_birds), shape = 21, alpha = .5) +
  scale_fill_viridis_c(name = "MTR", option = "turbo", guide = "legend") +
  scale_size(name = "MTR", guide = "legend") +
  theme_bw()

关键提示 #

• 避免用between()处理环形数据的跨0区间，线性判断逻辑不适用环形空间。
• is.inside.circular()会自动识别环形区间的跨0特性，代码更简洁且不易出错，是处理这类问题的最优选择。

内容的提问来源于stack exchange，提问作者Simon Hirschhofer