# tree ring data
library(mccrr)
library(lubridate)
library(tidyverse)
library(jsonlite)
library(tibble)
library(ggplot2)
library(dplyr)
library(tidyr)

pdsi <- readr::read_csv("C:\\Users\\Andrew\\Desktop\\Statistics and Data Analysis\\climate_data\\NADAv2a.csv")
metadata <- jsonlite::fromJSON(txt = "C:\\Users\\Andrew\\Desktop\\Statistics and Data Analysis\\climate_data\\metadata\\noaa-recon-6319.json", simplifyVector = F)

pdsi_L <- pdsi %>% 
  pivot_longer(-year, 
               names_to = "station", 
               values_to = "pdsi", 
               values_drop_na = TRUE)


meta_raw <- enframe(unlist(metadata))

meta_clean <- meta_raw %>% 
  # takes unlisted data, removes the several thousand variables we don't need, 
  # creates a variable to be widened later
  # then recodes them by row number (dangerous! but they *are* in order)
  # to join them to the original pdsi data set.
  filter(stringr::str_detect(name, 'coordinates')) %>% 
  mutate(latlong = case_when(
    grepl("1", name, fixed=TRUE) ~ "longitude", 
    grepl("2", name, fixed=TRUE) ~ "latitude"
  )) %>% 
  group_by(name) %>% 
  mutate(station = row_number()) %>% 
  ungroup() %>% 
  select(-name) %>% 
  pivot_wider(names_from = latlong, values_from = value)



trees <- pdsi_L %>% 
  mutate(station = as.numeric(station)) %>% 
  left_join(meta_clean, by = "station")