MalditoBarbudo/sfn_data_fixes.R

## sfn_data_fixes.R
######################################################################
# CODE TO FIX KNOWN BUGS IN THE 0.1.3 VERSION OF THE SAPFLUXNET
# DATABASE
#
# INSTRUCTIONS:
# 1. COPY THIS FILE TO AN EMPTY R SCRIPT IN YOUR LOCAL MACHINE
# 2. INSTALL tidyverse AND sapfluxnetr IF NEEDED
# 3. CHANGE LAST LINE AS DESIRED
# 4. EXECUTE ALL THE SCRIPT
#######################################################################

# install.packages(tidyverse)
# install.packages(sapfluxnetr)
library(tidyverse)
library(sapfluxnetr)

## Site functions ####
ESP_YUN_T3_THI_fixes <- function(folder, csv = FALSE) {

  if (!file.exists(file.path(folder, 'ESP_YUN_T3_THI.RData'))) {
    return()
  }

  bad_site <- read_sfn_data('ESP_YUN_T3_THI', folder)

  # check sapflow
  if ('ESP_YUN_T3_THI_Api_Js_1' %in% names(get_sapf_data(bad_site))) {
    good_sapf_data <- bad_site %>%
      get_sapf_data()
  } else {
    good_sapf_data <- bad_site %>%
      get_sapf_data() %>%
      dplyr::select(
        ESP_YUN_T3_THI_Api_Js_1 = ESP_YUN_T3_Api_Js_1,
        ESP_YUN_T3_THI_Api_Js_2 = ESP_YUN_T3_Api_Js_2,
        ESP_YUN_T3_THI_Api_Js_3 = ESP_YUN_T3_Api_Js_3,
        ESP_YUN_T3_THI_Api_Js_4 = ESP_YUN_T3_Api_Js_4,
        ESP_YUN_T3_THI_Api_Js_5 = ESP_YUN_T3_Api_Js_5,
        ESP_YUN_T3_THI_Api_Js_6 = ESP_YUN_T3_Api_Js_6
      )
  }
  # check flags
  if ('ESP_YUN_T3_THI_Api_Js_1' %in% names(get_sapf_flags(bad_site))) {
    good_sapf_flags <- bad_site %>%
      get_sapf_flags()
  } else {
    good_sapf_flags <- bad_site %>%
      get_sapf_flags() %>%
      dplyr::select(
        ESP_YUN_T3_THI_Api_Js_1 = ESP_YUN_T3_Api_Js_1,
        ESP_YUN_T3_THI_Api_Js_2 = ESP_YUN_T3_Api_Js_2,
        ESP_YUN_T3_THI_Api_Js_3 = ESP_YUN_T3_Api_Js_3,
        ESP_YUN_T3_THI_Api_Js_4 = ESP_YUN_T3_Api_Js_4,
        ESP_YUN_T3_THI_Api_Js_5 = ESP_YUN_T3_Api_Js_5,
        ESP_YUN_T3_THI_Api_Js_6 = ESP_YUN_T3_Api_Js_6
      )
  }
  # check plant names in plant metadata
  is_plant_md_ok <- bad_site %>%
    get_plant_md() %>%
    pull(pl_code) %>%
    str_detect(pattern = '_THI_') %>%
    all()

  if (is_plant_md_ok) {
    good_plant_md <- bad_site %>%
      get_plant_md()
  } else {
    good_plant_md <- bad_site %>%
      get_plant_md() %>%
      dplyr::mutate(
        pl_code = stringr::str_replace(pl_code, 'ESP_YUN_T3_', 'ESP_YUN_T3_THI_')
      )
  }

  # finaly build the fixed site
  ESP_YUN_T3_THI <- bad_site
  get_sapf_data(ESP_YUN_T3_THI) <- good_sapf_data
  get_sapf_flags(ESP_YUN_T3_THI) <- good_sapf_flags
  get_plant_md(ESP_YUN_T3_THI) <- good_plant_md
  # and save it
  save(ESP_YUN_T3_THI, file = file.path(folder, 'ESP_YUN_T3_THI.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(ESP_YUN_T3_THI) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_YUN_T3_THI)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_YUN_T3_THI_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(ESP_YUN_T3_THI) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_YUN_T3_THI)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_YUN_T3_THI_sapf_flags.csv'
      ))

    # plant_md
    get_plant_md(ESP_YUN_T3_THI) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_YUN_T3_THI_plant_md.csv'
      ))
  }
}

COL_MAC_SAF_RAD_fixes <- function(folder, csv = FALSE) {

  if (!file.exists(file.path(folder, 'COL_MAC_SAF_RAD.RData'))) {
    return()
  }

  bad_site <- read_sfn_data('COL_MAC_SAF_RAD', folder)
  # check sapflow
  if ('COL_MAC_SAF_RAD_Tca_Js_1' %in% names(get_sapf_data(bad_site))) {
    good_sapf_data <- bad_site %>%
      get_sapf_data()
  } else {
    good_sapf_data <- bad_site %>%
      get_sapf_data() %>%
      dplyr::select(
        COL_MAC_SAF_RAD_Tca_Js_1 = COL_MAC_SAF_Tca_Js_1,
        COL_MAC_SAF_RAD_Tca_Js_2 = COL_MAC_SAF_Tca_Js_2,
        COL_MAC_SAF_RAD_Tca_Js_3 = COL_MAC_SAF_Tca_Js_3
      )
  }
  # check flags
  if ('COL_MAC_SAF_RAD_Tca_Js_1' %in% names(get_sapf_flags(bad_site))) {
    good_sapf_flags <- bad_site %>%
      get_sapf_flags()
  } else {
    good_sapf_flags <- bad_site %>%
      get_sapf_flags() %>%
      dplyr::select(
        COL_MAC_SAF_RAD_Tca_Js_1 = COL_MAC_SAF_Tca_Js_1,
        COL_MAC_SAF_RAD_Tca_Js_2 = COL_MAC_SAF_Tca_Js_2,
        COL_MAC_SAF_RAD_Tca_Js_3 = COL_MAC_SAF_Tca_Js_3
      )
  }
  # check plant names in plant metadata
  is_plant_md_ok <- bad_site %>%
    get_plant_md() %>%
    pull(pl_code) %>%
    str_detect(pattern = '_RAD_') %>%
    all()

  if (is_plant_md_ok) {
    good_plant_md <- bad_site %>%
      get_plant_md()
  } else {
    good_plant_md <- bad_site %>%
      get_plant_md() %>%
      dplyr::mutate(
        pl_code = stringr::str_replace(pl_code, 'COL_MAC_SAF_', 'COL_MAC_SAF_RAD_')
      )
  }

  # finaly build the fixed site
  COL_MAC_SAF_RAD <- bad_site
  get_sapf_data(COL_MAC_SAF_RAD) <- good_sapf_data
  get_sapf_flags(COL_MAC_SAF_RAD) <- good_sapf_flags
  get_plant_md(COL_MAC_SAF_RAD) <- good_plant_md
  # and save it
  save(COL_MAC_SAF_RAD, file = file.path(folder, 'COL_MAC_SAF_RAD.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(COL_MAC_SAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(COL_MAC_SAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(COL_MAC_SAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(COL_MAC_SAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_sapf_flags.csv'
      ))

    # plant_md
    get_plant_md(COL_MAC_SAF_RAD) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_plant_md.csv'
      ))
  }
}

AUS_MAR_HSW_HIG_fixes <- function(folder, csv = FALSE) {

  if (!file.exists(file.path(folder, 'AUS_MAR_HSW_HIG.RData'))) {
    return()
  }

  bad_site <- read_sfn_data('AUS_MAR_HSW_HIG', folder)
  # check sapflow
  if ('AUS_MAR_HSW_HIG_Eob_Js_1' %in% names(get_sapf_data(bad_site))) {
    good_sapf_data <- bad_site %>%
      get_sapf_data()
  } else {
    good_sapf_data <- bad_site %>%
      get_sapf_data() %>%
      dplyr::select(
        AUS_MAR_HSW_HIG_Eob_Js_1 = AUS_MAR_HSW_Eob_Js_1,
        AUS_MAR_HSW_HIG_Eob_Js_2 = AUS_MAR_HSW_Eob_Js_2,
        AUS_MAR_HSW_HIG_Eob_Js_3 = AUS_MAR_HSW_Eob_Js_3,
        AUS_MAR_HSW_HIG_Ecy_Js_4 = AUS_MAR_HSW_Ecy_Js_4
      )
  }
  # check flags
  if ('AUS_MAR_HSW_HIG_Eob_Js_1' %in% names(get_sapf_flags(bad_site))) {
    good_sapf_flags <- bad_site %>%
      get_sapf_flags()
  } else {
    good_sapf_flags <- bad_site %>%
      get_sapf_flags() %>%
      dplyr::select(
        AUS_MAR_HSW_HIG_Eob_Js_1 = AUS_MAR_HSW_Eob_Js_1,
        AUS_MAR_HSW_HIG_Eob_Js_2 = AUS_MAR_HSW_Eob_Js_2,
        AUS_MAR_HSW_HIG_Eob_Js_3 = AUS_MAR_HSW_Eob_Js_3,
        AUS_MAR_HSW_HIG_Ecy_Js_4 = AUS_MAR_HSW_Ecy_Js_4
      )
  }
  # check plant names in plant metadata
  is_plant_md_ok <- bad_site %>%
    get_plant_md() %>%
    pull(pl_code) %>%
    str_detect(pattern = '_HIG_') %>%
    all()

  if (is_plant_md_ok) {
    good_plant_md <- bad_site %>%
      get_plant_md()
  } else {
    good_plant_md <- bad_site %>%
      get_plant_md() %>%
      dplyr::mutate(
        pl_code = stringr::str_replace(pl_code, 'AUS_MAR_HSW_', 'AUS_MAR_HSW_HIG_')
      )
  }

  # finaly build the fixed site
  AUS_MAR_HSW_HIG <- bad_site
  get_sapf_data(AUS_MAR_HSW_HIG) <- good_sapf_data
  get_sapf_flags(AUS_MAR_HSW_HIG) <- good_sapf_flags
  get_plant_md(AUS_MAR_HSW_HIG) <- good_plant_md
  # and save it
  save(AUS_MAR_HSW_HIG, file = file.path(folder, 'AUS_MAR_HSW_HIG.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(AUS_MAR_HSW_HIG) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(AUS_MAR_HSW_HIG)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'AUS_MAR_HSW_HIG_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(AUS_MAR_HSW_HIG) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(AUS_MAR_HSW_HIG)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'AUS_MAR_HSW_HIG_sapf_flags.csv'
      ))

    # plant_md
    get_plant_md(AUS_MAR_HSW_HIG) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'AUS_MAR_HSW_HIG_plant_md.csv'
      ))
  }
}


USA_SMI_SER_fixes <- function(folder, csv = FALSE) {

  if (!file.exists(file.path(folder, 'USA_SMI_SER.RData'))) {
    return()
  }

  bad_site <- read_sfn_data('USA_SMI_SER', folder)

  # change the units in sapf data only if we are in plant level, if not, we only need
  # to change the metadata, as the sapwood values were the original ones.
  if (stringr::str_detect(folder, 'plant')) {
    USA_SMI_SER <- bad_site %>%
      sfn_mutate_at(
        .vars = vars(one_of(names(get_sapf_data(bad_site)[,-1]))),
        .funs = list(~ .x * 1e4)
      )
  } else {
    USA_SMI_SER <- bad_site
  }

  # check plant md
  good_plant_md <- USA_SMI_SER %>%
    get_plant_md() %>%
    mutate(pl_sapw_area = pl_sapw_area * 1e4)

  # finaly build the fixed site
  get_plant_md(USA_SMI_SER) <- good_plant_md
  # and save it
  save(USA_SMI_SER, file = file.path(folder, 'USA_SMI_SER.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_SMI_SER) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_SMI_SER)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_SMI_SER_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_SMI_SER) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_SMI_SER)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_SMI_SER_sapf_flags.csv'
      ))

    # plant_md
    get_plant_md(USA_SMI_SER) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_SMI_SER_plant_md.csv'
      ))
  }

}

USA_SMI_SCB_fixes <- function(folder, csv = FALSE) {

  if (!file.exists(file.path(folder, 'USA_SMI_SCB.RData'))) {
    return()
  }

  bad_site <- read_sfn_data('USA_SMI_SCB', folder)

  # change the units in sapf data only if we are in plant level, if not, we only need
  # to change the metadata, as the sapwood values were the original ones.
  if (stringr::str_detect(folder, 'plant')) {
    USA_SMI_SCB <- bad_site %>%
      sfn_mutate_at(
        .vars = vars(one_of(names(get_sapf_data(bad_site)[,-1]))),
        .funs = list(~ .x * 1e4)
      )
  } else {
    USA_SMI_SCB <- bad_site
  }

  # check plant md
  good_plant_md <- USA_SMI_SCB %>%
    get_plant_md() %>%
    mutate(pl_sapw_area = pl_sapw_area * 1e4)

  # finaly build the fixed site
  get_plant_md(USA_SMI_SCB) <- good_plant_md
  # and save it
  save(USA_SMI_SCB, file = file.path(folder, 'USA_SMI_SCB.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_SMI_SCB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_SMI_SCB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_SMI_SCB_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_SMI_SCB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_SMI_SCB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_SMI_SCB_sapf_flags.csv'
      ))

    # plant_md
    get_plant_md(USA_SMI_SCB) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_SMI_SCB_plant_md.csv'
      ))
  }

}

USA_CHE_ASP_fixes <- function(folder, csv = FALSE) {

  if (!file.exists(file.path(folder, 'USA_CHE_ASP.RData'))) {
    return()
  }

  USA_CHE_ASP <- read_sfn_data('USA_CHE_ASP', folder)

  # In this case, there is only to fix the metadata. Sapwood depth is not used to
  # calculate sapwood area, as this value is already provided.
  good_plant_md <- USA_CHE_ASP %>%
    get_plant_md() %>%
    mutate(pl_sapw_depth = pl_sapw_depth / 10)
  # finaly build the fixed site
  get_plant_md(USA_CHE_ASP) <- good_plant_md
  # and save it
  save(USA_CHE_ASP, file = file.path(folder, 'USA_CHE_ASP.RData'))

  # csv files
  if (isTRUE(csv)) {
    # plant_md
    get_plant_md(USA_CHE_ASP) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_CHE_ASP_plant_md.csv'
      ))
  }
}

RUS_POG_VAR_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, 'RUS_POG_VAR.RData'))) {
    return()
  }

  # Here we need to set stand basal area percentage to NA
  RUS_POG_VAR <- read_sfn_data('RUS_POG_VAR', folder)

  good_species_md <- RUS_POG_VAR %>%
    get_species_md() %>%
    mutate(sp_basal_area_perc = NA_integer_)

  # finaly build the fixed site
  get_species_md(RUS_POG_VAR) <- good_species_md
  # and save it
  save(RUS_POG_VAR, file = file.path(folder, 'RUS_POG_VAR.RData'))

  # csv files
  if (isTRUE(csv)) {
    # species_md
    get_species_md(RUS_POG_VAR) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'RUS_POG_VAR_species_md.csv'
      ))
  }

}

## TZ issues fixes ####

# sites <- c(
#   'ZAF_SOU_SOU', 'ZAF_RAD', 'ZAF_FRA_FRA', 'RUS_FYO', 'IDN_JAM_RUB',
#   "IDN_JAM_OIL", 'ESP_PRA', "ESP_LAS", 'USA_WVF', 'USA_TNB',
#   'USA_TNO', 'USA_TNP', 'USA_TNY', 'USA_PAR_FER', 'USA_INM',
#   'COL_MAC_SAF_RAD'
# )
#
# good_tz <- c(
#   "Etc/GMT-2", "Etc/GMT-2", "Etc/GMT-2", "Etc/GMT-3", "Etc/GMT-7",
#   "Etc/GMT-7", "Etc/GMT-1", "Etc/GMT+0",  "Etc/GMT+5", "Etc/GMT+5",
#   "Etc/GMT+5", "Etc/GMT+5", "Etc/GMT+5", "Etc/GMT+5", "Etc/GMT+5",
#   "Etc/GMT+5"
# )

solar_timestamp_creator <- function(timestamp, lat, long) {

  mst <- solaR::local2Solar(timestamp, long)

  # 2.2.1 warning if solartimestamp has repeated values (due to rounding)
  if (length(mst) != length(unique(mst))) {
    warning('solar mean time generates repeated timestamps. ',
            'Please revise the original timestamp for repeated values.')
  }

  # STEP 3
  # Calculating Apparent Solar Time (Mean Solar Time + Equation of Time)
  # 2.1 Equation of time
  solD <- solaR::fSolD(lat, mst)
  EoT <- solaR::r2sec(solD$EoT)

  ast <- lapply(as.Date(strptime(zoo::index(EoT), format = '%Y-%m-%d')),
                function(id, vect)
                  (vect[as.Date(vect) == id] +
                     zoo::coredata(EoT)[which(as.Date(strptime(zoo::index(EoT),
                                                               format = '%Y-%m-%d')) == id)]),
                vect = mst)

  ast <- do.call("c", ast)

  return(ast)
}

ZAF_SOU_SOU_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('ZAF_SOU_SOU','.RData')))) {
    return()
  }

  ZAF_SOU_SOU <- read_sfn_data('ZAF_SOU_SOU', folder)
  site_metadata <- get_site_md(ZAF_SOU_SOU)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- ZAF_SOU_SOU %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-2")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- ZAF_SOU_SOU %>%
    get_env_md() %>%
    mutate(env_time_zone = "18UTC+02:00, B")

  get_timestamp(ZAF_SOU_SOU) <- good_timestamp
  get_solar_timestamp(ZAF_SOU_SOU) <- good_solar_timestamp
  get_env_md(ZAF_SOU_SOU) <- good_env_md

  save(ZAF_SOU_SOU, file = file.path(folder, 'ZAF_SOU_SOU.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(ZAF_SOU_SOU) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_SOU_SOU)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_SOU_SOU_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(ZAF_SOU_SOU) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_SOU_SOU)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_SOU_SOU_sapf_flags.csv'
      ))

    # env_data
    get_env_data(ZAF_SOU_SOU) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_SOU_SOU)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_SOU_SOU_env_data.csv'
      ))

    # env_flags
    get_env_flags(ZAF_SOU_SOU) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_SOU_SOU)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_SOU_SOU_env_flags.csv'
      ))

    # env_md
    get_env_md(ZAF_SOU_SOU) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_SOU_SOU_env_md.csv'
      ))
  }
}

ZAF_RAD_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('ZAF_RAD','.RData')))) {
    return()
  }

  ZAF_RAD <- read_sfn_data('ZAF_RAD', folder)
  site_metadata <- get_site_md(ZAF_RAD)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- ZAF_RAD %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-2")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- ZAF_RAD %>%
    get_env_md() %>%
    mutate(env_time_zone = "18UTC+02:00, B")

  get_timestamp(ZAF_RAD) <- good_timestamp
  get_solar_timestamp(ZAF_RAD) <- good_solar_timestamp
  get_env_md(ZAF_RAD) <- good_env_md

  save(ZAF_RAD, file = file.path(folder, 'ZAF_RAD.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(ZAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_RAD_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(ZAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_RAD_sapf_flags.csv'
      ))

    # env_data
    get_env_data(ZAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_RAD_env_data.csv'
      ))

    # env_flags
    get_env_flags(ZAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_RAD_env_flags.csv'
      ))

    # env_md
    get_env_md(ZAF_RAD) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_RAD_env_md.csv'
      ))
  }
}

ZAF_FRA_FRA_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('ZAF_FRA_FRA','.RData')))) {
    return()
  }

  ZAF_FRA_FRA <- read_sfn_data('ZAF_FRA_FRA', folder)
  site_metadata <- get_site_md(ZAF_FRA_FRA)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- ZAF_FRA_FRA %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-2")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- ZAF_FRA_FRA %>%
    get_env_md() %>%
    mutate(env_time_zone = "18UTC+02:00, B")

  get_timestamp(ZAF_FRA_FRA) <- good_timestamp
  get_solar_timestamp(ZAF_FRA_FRA) <- good_solar_timestamp
  get_env_md(ZAF_FRA_FRA) <- good_env_md

  save(ZAF_FRA_FRA, file = file.path(folder, 'ZAF_FRA_FRA.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(ZAF_FRA_FRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_FRA_FRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_FRA_FRA_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(ZAF_FRA_FRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_FRA_FRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_FRA_FRA_sapf_flags.csv'
      ))

    # env_data
    get_env_data(ZAF_FRA_FRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_FRA_FRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_FRA_FRA_env_data.csv'
      ))

    # env_flags
    get_env_flags(ZAF_FRA_FRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ZAF_FRA_FRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_FRA_FRA_env_flags.csv'
      ))

    # env_md
    get_env_md(ZAF_FRA_FRA) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ZAF_FRA_FRA_env_md.csv'
      ))
  }
}

RUS_FYO_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('RUS_FYO','.RData')))) {
    return()
  }

  RUS_FYO <- read_sfn_data('RUS_FYO', folder)
  site_metadata <- get_site_md(RUS_FYO)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- RUS_FYO %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-3")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- RUS_FYO %>%
    get_env_md() %>%
    mutate(env_time_zone = "19UTC+03:00, C")

  get_timestamp(RUS_FYO) <- good_timestamp
  get_solar_timestamp(RUS_FYO) <- good_solar_timestamp
  get_env_md(RUS_FYO) <- good_env_md

  save(RUS_FYO, file = file.path(folder, 'RUS_FYO.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(RUS_FYO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(RUS_FYO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'RUS_FYO_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(RUS_FYO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(RUS_FYO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'RUS_FYO_sapf_flags.csv'
      ))

    # env_data
    get_env_data(RUS_FYO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(RUS_FYO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'RUS_FYO_env_data.csv'
      ))

    # env_flags
    get_env_flags(RUS_FYO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(RUS_FYO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'RUS_FYO_env_flags.csv'
      ))

    # env_md
    get_env_md(RUS_FYO) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'RUS_FYO_env_md.csv'
      ))
  }
}

IDN_JAM_RUB_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('IDN_JAM_RUB','.RData')))) {
    return()
  }

  IDN_JAM_RUB <- read_sfn_data('IDN_JAM_RUB', folder)
  site_metadata <- get_site_md(IDN_JAM_RUB)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- IDN_JAM_RUB %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-7")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- IDN_JAM_RUB %>%
    get_env_md() %>%
    mutate(env_time_zone = "28UTC+07:00, G")

  get_timestamp(IDN_JAM_RUB) <- good_timestamp
  get_solar_timestamp(IDN_JAM_RUB) <- good_solar_timestamp
  get_env_md(IDN_JAM_RUB) <- good_env_md

  save(IDN_JAM_RUB, file = file.path(folder, 'IDN_JAM_RUB.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(IDN_JAM_RUB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_RUB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_RUB_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(IDN_JAM_RUB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_RUB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_RUB_sapf_flags.csv'
      ))

    # env_data
    get_env_data(IDN_JAM_RUB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_RUB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_RUB_env_data.csv'
      ))

    # env_flags
    get_env_flags(IDN_JAM_RUB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_RUB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_RUB_env_flags.csv'
      ))

    # env_md
    get_env_md(IDN_JAM_RUB) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_RUB_env_md.csv'
      ))
  }
}

IDN_JAM_OIL_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('IDN_JAM_OIL','.RData')))) {
    return()
  }

  IDN_JAM_OIL <- read_sfn_data('IDN_JAM_OIL', folder)
  site_metadata <- get_site_md(IDN_JAM_OIL)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- IDN_JAM_OIL %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-7")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- IDN_JAM_OIL %>%
    get_env_md() %>%
    mutate(env_time_zone = "28UTC+07:00, G")

  get_timestamp(IDN_JAM_OIL) <- good_timestamp
  get_solar_timestamp(IDN_JAM_OIL) <- good_solar_timestamp
  get_env_md(IDN_JAM_OIL) <- good_env_md

  save(IDN_JAM_OIL, file = file.path(folder, 'IDN_JAM_OIL.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(IDN_JAM_OIL) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_OIL)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_OIL_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(IDN_JAM_OIL) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_OIL)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_OIL_sapf_flags.csv'
      ))

    # env_data
    get_env_data(IDN_JAM_OIL) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_OIL)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_OIL_env_data.csv'
      ))

    # env_flags
    get_env_flags(IDN_JAM_OIL) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(IDN_JAM_OIL)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_OIL_env_flags.csv'
      ))

    # env_md
    get_env_md(IDN_JAM_OIL) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'IDN_JAM_OIL_env_md.csv'
      ))
  }
}

ESP_PRA_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('ESP_PRA','.RData')))) {
    return()
  }

  ESP_PRA <- read_sfn_data('ESP_PRA', folder)
  site_metadata <- get_site_md(ESP_PRA)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- ESP_PRA %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT-1")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- ESP_PRA %>%
    get_env_md() %>%
    mutate(env_time_zone = "17UTC+01:00, A")

  get_timestamp(ESP_PRA) <- good_timestamp
  get_solar_timestamp(ESP_PRA) <- good_solar_timestamp
  get_env_md(ESP_PRA) <- good_env_md

  save(ESP_PRA, file = file.path(folder, 'ESP_PRA.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(ESP_PRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_PRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_PRA_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(ESP_PRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_PRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_PRA_sapf_flags.csv'
      ))

    # env_data
    get_env_data(ESP_PRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_PRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_PRA_env_data.csv'
      ))

    # env_flags
    get_env_flags(ESP_PRA) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_PRA)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_PRA_env_flags.csv'
      ))

    # env_md
    get_env_md(ESP_PRA) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_PRA_env_md.csv'
      ))
  }
}

ESP_LAS_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('ESP_LAS','.RData')))) {
    return()
  }

  ESP_LAS <- read_sfn_data('ESP_LAS', folder)
  site_metadata <- get_site_md(ESP_LAS)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- ESP_LAS %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+0")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- ESP_LAS %>%
    get_env_md() %>%
    mutate(env_time_zone = "16UTC±00:00, Z")

  get_timestamp(ESP_LAS) <- good_timestamp
  get_solar_timestamp(ESP_LAS) <- good_solar_timestamp
  get_env_md(ESP_LAS) <- good_env_md

  save(ESP_LAS, file = file.path(folder, 'ESP_LAS.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(ESP_LAS) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_LAS)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_LAS_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(ESP_LAS) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_LAS)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_LAS_sapf_flags.csv'
      ))

    # env_data
    get_env_data(ESP_LAS) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_LAS)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_LAS_env_data.csv'
      ))

    # env_flags
    get_env_flags(ESP_LAS) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(ESP_LAS)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_LAS_env_flags.csv'
      ))

    # env_md
    get_env_md(ESP_LAS) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'ESP_LAS_env_md.csv'
      ))
  }
}

USA_WVF_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_WVF','.RData')))) {
    return()
  }

  USA_WVF <- read_sfn_data('USA_WVF', folder)
  site_metadata <- get_site_md(USA_WVF)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_WVF %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_WVF %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_WVF) <- good_timestamp
  get_solar_timestamp(USA_WVF) <- good_solar_timestamp
  get_env_md(USA_WVF) <- good_env_md

  save(USA_WVF, file = file.path(folder, 'USA_WVF.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_WVF) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_WVF)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_WVF_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_WVF) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_WVF)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_WVF_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_WVF) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_WVF)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_WVF_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_WVF) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_WVF)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_WVF_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_WVF) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_WVF_env_md.csv'
      ))
  }
}

USA_TNB_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_TNB','.RData')))) {
    return()
  }

  USA_TNB <- read_sfn_data('USA_TNB', folder)
  site_metadata <- get_site_md(USA_TNB)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_TNB %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_TNB %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_TNB) <- good_timestamp
  get_solar_timestamp(USA_TNB) <- good_solar_timestamp
  get_env_md(USA_TNB) <- good_env_md

  save(USA_TNB, file = file.path(folder, 'USA_TNB.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_TNB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNB_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_TNB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNB_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_TNB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNB_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_TNB) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNB)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNB_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_TNB) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNB_env_md.csv'
      ))
  }
}

USA_TNO_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_TNO','.RData')))) {
    return()
  }

  USA_TNO <- read_sfn_data('USA_TNO', folder)
  site_metadata <- get_site_md(USA_TNO)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_TNO %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_TNO %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_TNO) <- good_timestamp
  get_solar_timestamp(USA_TNO) <- good_solar_timestamp
  get_env_md(USA_TNO) <- good_env_md

  save(USA_TNO, file = file.path(folder, 'USA_TNO.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_TNO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNO_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_TNO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNO_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_TNO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNO_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_TNO) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNO)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNO_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_TNO) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNO_env_md.csv'
      ))
  }
}

USA_TNP_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_TNP','.RData')))) {
    return()
  }

  USA_TNP <- read_sfn_data('USA_TNP', folder)
  site_metadata <- get_site_md(USA_TNP)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_TNP %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_TNP %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_TNP) <- good_timestamp
  get_solar_timestamp(USA_TNP) <- good_solar_timestamp
  get_env_md(USA_TNP) <- good_env_md

  save(USA_TNP, file = file.path(folder, 'USA_TNP.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_TNP) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNP)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNP_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_TNP) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNP)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNP_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_TNP) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNP)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNP_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_TNP) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNP)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNP_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_TNP) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNP_env_md.csv'
      ))
  }
}

USA_TNY_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_TNY','.RData')))) {
    return()
  }

  USA_TNY <- read_sfn_data('USA_TNY', folder)
  site_metadata <- get_site_md(USA_TNY)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_TNY %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_TNY %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_TNY) <- good_timestamp
  get_solar_timestamp(USA_TNY) <- good_solar_timestamp
  get_env_md(USA_TNY) <- good_env_md

  save(USA_TNY, file = file.path(folder, 'USA_TNY.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_TNY) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNY)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNY_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_TNY) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNY)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNY_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_TNY) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNY)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNY_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_TNY) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_TNY)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNY_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_TNY) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_TNY_env_md.csv'
      ))
  }
}

USA_PAR_FER_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_PAR_FER','.RData')))) {
    return()
  }

  USA_PAR_FER <- read_sfn_data('USA_PAR_FER', folder)
  site_metadata <- get_site_md(USA_PAR_FER)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_PAR_FER %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_PAR_FER %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_PAR_FER) <- good_timestamp
  get_solar_timestamp(USA_PAR_FER) <- good_solar_timestamp
  get_env_md(USA_PAR_FER) <- good_env_md

  save(USA_PAR_FER, file = file.path(folder, 'USA_PAR_FER.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_PAR_FER) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_PAR_FER)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_PAR_FER_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_PAR_FER) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_PAR_FER)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_PAR_FER_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_PAR_FER) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_PAR_FER)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_PAR_FER_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_PAR_FER) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_PAR_FER)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_PAR_FER_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_PAR_FER) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_PAR_FER_env_md.csv'
      ))
  }
}

USA_INM_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('USA_INM','.RData')))) {
    return()
  }

  USA_INM <- read_sfn_data('USA_INM', folder)
  site_metadata <- get_site_md(USA_INM)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- USA_INM %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- USA_INM %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(USA_INM) <- good_timestamp
  get_solar_timestamp(USA_INM) <- good_solar_timestamp
  get_env_md(USA_INM) <- good_env_md

  save(USA_INM, file = file.path(folder, 'USA_INM.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(USA_INM) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_INM)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_INM_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(USA_INM) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_INM)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_INM_sapf_flags.csv'
      ))

    # env_data
    get_env_data(USA_INM) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_INM)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_INM_env_data.csv'
      ))

    # env_flags
    get_env_flags(USA_INM) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(USA_INM)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_INM_env_flags.csv'
      ))

    # env_md
    get_env_md(USA_INM) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'USA_INM_env_md.csv'
      ))
  }
}

COL_MAC_SAF_RAD_fixes <- function(folder, csv) {

  if (!file.exists(file.path(folder, paste0('COL_MAC_SAF_RAD','.RData')))) {
    return()
  }

  COL_MAC_SAF_RAD <- read_sfn_data('COL_MAC_SAF_RAD', folder)
  site_metadata <- get_site_md(COL_MAC_SAF_RAD)


  # we need to update both timestamps, and the environmental metadata
  good_timestamp <- COL_MAC_SAF_RAD %>%
    get_timestamp() %>%
    lubridate::force_tz("Etc/GMT+5")

  good_solar_timestamp <- good_timestamp %>%
    solar_timestamp_creator(lat = site_metadata$si_lat, long = site_metadata$si_long)

  good_env_md <- COL_MAC_SAF_RAD %>%
    get_env_md() %>%
    mutate(env_time_zone = "9UTC-05:00, R")

  get_timestamp(COL_MAC_SAF_RAD) <- good_timestamp
  get_solar_timestamp(COL_MAC_SAF_RAD) <- good_solar_timestamp
  get_env_md(COL_MAC_SAF_RAD) <- good_env_md

  save(COL_MAC_SAF_RAD, file = file.path(folder, 'COL_MAC_SAF_RAD.RData'))

  # csv files
  if (isTRUE(csv)) {
    # sapf_data
    get_sapf_data(COL_MAC_SAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(COL_MAC_SAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_sapf_data.csv'
      ))

    # sapf_flags
    get_sapf_flags(COL_MAC_SAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(COL_MAC_SAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_sapf_flags.csv'
      ))

    # env_data
    get_env_data(COL_MAC_SAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(COL_MAC_SAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_env_data.csv'
      ))

    # env_flags
    get_env_flags(COL_MAC_SAF_RAD) %>%
      dplyr::mutate(TIMESTAMP_solar = get_solar_timestamp(COL_MAC_SAF_RAD)) %>%
      dplyr::select(dplyr::starts_with('TIMESTAMP'), everything()) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_env_flags.csv'
      ))


    # env_md
    get_env_md(COL_MAC_SAF_RAD) %>%
      readr::write_csv(path = file.path(
        stringr::str_replace(folder, 'RData', 'csv'),
        'COL_MAC_SAF_RAD_env_md.csv'
      ))
  }
}

## MAIN FUNCTION ####
sfn_data_fixes <- function(db_folder, csv = FALSE) {

  data_levels_folders <- file.path(db_folder, 'RData', c('plant', 'sapwood', 'leaf'))
  # sites <- c('AUS_MAR_HSW_HIG', 'COL_MAC_SAF_RAD', 'ESP_YUN_T3_THI')

  for (folder in data_levels_folders) {
    AUS_MAR_HSW_HIG_fixes(folder, csv)
    COL_MAC_SAF_RAD_fixes(folder, csv)
    ESP_YUN_T3_THI_fixes(folder, csv)
    USA_SMI_SER_fixes(folder, csv)
    USA_SMI_SCB_fixes(folder, csv)
    USA_CHE_ASP_fixes(folder, csv)
    RUS_POG_VAR_fixes(folder, csv)
    ZAF_SOU_SOU_fixes(folder, csv)
    ZAF_RAD_fixes(folder, csv)
    ZAF_FRA_FRA_fixes(folder, csv)
    RUS_FYO_fixes(folder, csv)
    IDN_JAM_RUB_fixes(folder, csv)
    IDN_JAM_OIL_fixes(folder, csv)
    ESP_PRA_fixes(folder, csv)
    ESP_LAS_fixes(folder, csv)
    USA_WVF_fixes(folder, csv)
    USA_TNB_fixes(folder, csv)
    USA_TNO_fixes(folder, csv)
    USA_TNP_fixes(folder, csv)
    USA_TNY_fixes(folder, csv)
    USA_PAR_FER_fixes(folder, csv)
    USA_INM_fixes(folder, csv)
    COL_MAC_SAF_RAD_fixes(folder, csv)
    sapfluxnetr:::.write_metadata_cache(file.path(folder))
  }
}

## Execute fixes ####
# CHANGE folder VALUE WITH THE PATH TO THE DATABASE VERSION (0.1.3)
# CHANGE csv VALUE TO FALSE IF YOU DONT WANT TO FIX THE CSV FILES
sfn_data_fixes(db_folder = '.', csv = TRUE)