如何聚合季节数据？请求绘制1920-2019年冬春季流量数据

Hey Kelsey, let's tackle this with a much cleaner, maintainable approach using dplyr and lubridate—no more manually typing out every start/end date! Here's how to calculate winter-spring discharge means for 1920-2019 and visualize the results:

Step 1: Load Packages & Retrieve Data #

First, let's get your daily flow data (we'll keep things tidy from the start):

library(dataRetrieval)
library(dplyr)
library(lubridate)
library(ggplot2)
library(ggpmisc)

# Retrieve daily discharge data
siteNo <- "02202500"
pCode <- "00060"
daily <- readNWISdv(siteNo, pCode, "1920-10-01", "2019-09-30") %>%
  renameNWISColumns() %>%
  select(Date, Flow) %>% # Keep only columns we need
  mutate(Flow_m3s = Flow * 0.0283168) # Convert to m³/s upfront

Step 2: Define Winter-Spring Seasons #

Your goal is to group December of year N to May of year N+1 as a single winter-spring season. We'll create a season_year column to label each date with the "target year" of its season (e.g., Dec 1920 to May 1921 gets labeled as 1921):

daily_seasonal <- daily %>%
  mutate(
    month = month(Date),
    # Assign season year: Dec belongs to next year's season; Jan-May belong to current year
    season_year = ifelse(month == 12, year(Date) + 1, year(Date))
  ) %>%
  # Filter only winter-spring months (Dec, Jan, Feb, Mar, Apr, May)
  filter(month %in% c(12, 1, 2, 3, 4, 5)) %>%
  # Keep seasons within your 1920-2019 range (covers Dec 1920 to May 2019)
  filter(season_year >= 1921 & season_year <= 2019)

Step 3: Calculate Seasonal Mean Discharge #

Now use dplyr to group by season_year and compute the mean flow:

seasonal_means <- daily_seasonal %>%
  group_by(season_year) %>%
  summarize(
    mean_flow_m3s = mean(Flow_m3s, na.rm = TRUE) # Handle missing values safely
  ) %>%
  ungroup() %>%
  # Add a midpoint date for smooth plotting (e.g., March of the season year)
  mutate(season_midpoint = ymd(paste0(season_year, "-03-15")))

Step 4: Plot the Seasonal Means #

Now we can create your desired plot with the seasonal means, including the linear trend and stats:

my.formula <- mean_flow_m3s ~ season_midpoint

p1 <- ggplot(data = seasonal_means, aes(x = season_midpoint, y = mean_flow_m3s)) +
  geom_line(color = "steelblue", linewidth = 1) +
  geom_smooth(method = "lm", se = FALSE, color = "darkred") +
  stat_poly_eq(
    formula = my.formula,
    eq.with.lhs = "italic(hat(y))~`=`~",
    aes(label = paste(..eq.label.., ..rr.label.., sep = "~~~")),
    parse = TRUE,
    color = "darkred"
  ) +
  theme_classic() +
  labs(
    x = "",
    y = expression(Discharge~(m^{3}~s^{-1}))
  ) +
  scale_x_date(
    breaks = "5 years",
    labels = scales::date_format("%Y"),
    expand = c(0, 0)
  ) +
  scale_y_continuous(limits = c(0, 600)) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

p1

Key Fixes & Improvements Over Your Initial Code #

• No manual date lists: We use lubridate to dynamically assign seasons, which is far less error-prone and works for any date range.
• Tidy data workflow: dplyr pipes keep the code readable and avoid messy data frame reassignments.
• Proper grouping: We explicitly group by season_year to calculate means for each winter-spring period, instead of using sapply which wasn't targeting the right groups.
• Missing value handling: Added na.rm=TRUE to the mean calculation to avoid errors if there are gaps in the discharge data.

内容的提问来源于stack exchange，提问作者Kelsey