In stineb/fvar:

library(tidyverse)
library(lubridate)
load("data/ddf_CH-Dav.RData")
ddf <- ddf %>% as_tibble()
source("~/mct/R/mct2.R")  # get this from https://github.com/stineb/mct/blob/master/R/mct2.R

## look at data frame
ddf

## plot precip
ddf %>% 
  ggplot(aes(x = date, y = P_F)) +
  geom_line()

## plot ET
ddf %>% 
  ggplot(aes(x = date, y = ET)) +
  geom_line()

## plot ET vs LE => almost perfectly linear
ddf %>% 
  ggplot(aes(x = LE_F_MDS, y = ET)) +
  geom_point()

## average annual ET and P => looks reasonable: P > ET
ddf %>% 
  mutate(year = lubridate::year(date)) %>% 
  group_by(year) %>% 
  summarise(precip = sum(P_F), et = sum(ET)) %>% 
  summarise_all(mean) %>% 
  dplyr::select(-year)

## any missing values? No -> ok
any(is.na(ddf$P_F))
any(is.na(ddf$ET))

## any missing dates? => only last date missing -> shouldn't be a problem
date_range <- seq(min(ddf$date), max(ddf$date), by = 1) 
date_range[!date_range %in% ddf$date] 

Water deficits are defined as negative numbers (positive numbers are a water surplus).

ddf <- ddf %>% 
  mutate(bal = P_F - ET)

Apply mct function.

out <- mct(ddf, "bal", "date", thresh_terminate = 0.0, thresh_drop = 0.9)

Look at CWD time series. => this doesn't look exactly the same as yours.

ggplot() +
  geom_rect(
    data=out$inst, 
    aes(xmin=date_start, xmax=date_end, ymin=-99, ymax=99999), 
    fill=rgb(0,0,0,0.3), 
    color=NA) +
  geom_line(data = out$df, aes(date, P_F), size = 0.3, color="royalblue") +
  geom_line(data = out$df, aes(date, deficit), color="tomato") +
  geom_line(data = out$df, aes(date, ET)) +
  coord_cartesian(ylim=c(0, 100), xlim = c(lubridate::ymd("2007-01-01"), lubridate::ymd("2011-01-01"))) +
  theme_classic() +
  labs(subtitle = "red: CWD, blue: precip, black: ET", x = "Date", y = "(mm)")

Note that the grey bars represent instances.

stineb/fvar documentation built on Oct. 15, 2022, 12:06 p.m.