In rpkgs/rfluxnet: Process fluxnet data

5个站点没有LAI数据，通过lai&GPP R2最大，将其位置调整到相邻网格

2024-03-22, Dongdong Kong https://code.earthengine.google.com/d2c78bba5034f692f114ee64b0fb05c7

library(sf)
library(sf2)
library(tidyr)
library(Ipaper)
library(dplyr)
library(lubridate)
library(hydroTools)

st2 = read_sf("./data-raw/st212_adjust_5points_loc/shp/flux_st5_adjusted.shp") |> 
  st_cast("POINT") |> 
  as.data.table()
st2[, group := as.character(1:.N), .(site)]

# 来源于下面的评估结果
info = tribble(
  ~site, ~group, 
  "DE-Akm", "1",
  "DE-RuS", "3",
  "IT-Ro1", "2", 
  "US-ORv", "1",
  "US-WPT", "2"
)

# st2用于修正站点位置信息
st2 = merge(info, st2) |> cbind("source"="fluxnet2015")

sites_bad <- c("US-WPT", "US-ORv", "DE-Akm", "DE-RuS", "IT-Ro1")
df_d8 = fread("Z:/Researches/PMLV2/rfluxnet.R/OUTPUT/fluxsites212_FULLSET_D8_v20240322 (80%).csv")
df <- df_d8[site %in% sites_bad, .(site, date, NEE, GPP_NT, GPP_DT)]

indir = "data-raw/st212_adjust_5points_loc/LAI_csv_temp"
f = glue("{indir}/st5_fix_LAI_2000-2023_MODIS_061_MOD15A2H.csv")
df_lai <- fread(f) |>
  separate(site2, into = c("site", "group"), sep = "_") |>
  data.table() |>
  mutate(date = as_date(date)) %>%
  # add_dn() |>
  # mutate(date2 = date_ydn(year, d8)) %>%
  .[, .(site, group, date, FparExtra_QC, LAI = Lai_500m)]

d = merge(df_lai, df)
info = d[, GOF(LAI, GPP_NT), .(site, group)]
info[, .SD[which.max(R2), .(group, R, R2, pvalue)], .(site)]

info <- d[, GOF(LAI, NEE), .(site, group)]
info[, .SD[which.max(R2), .(group, R, R2, pvalue)], .(site)]

湿地LAI&NEE为何是负相关？ LAI&GPP是正相关，而且相关系数极高

# 最终挑选的站点, NEE
merge(df_lai, info) |> select(-group) |> unique()

#   site   group      R    R2   pvalue
#   <chr>  <chr>  <dbl> <dbl>    <dbl>
# 1 DE-Akm 1     -0.505 0.255 2.65e-19
# 2 DE-RuS 3     -0.494 0.244 9.86e-13
# 3 IT-Ro1 2     -0.629 0.396 4.77e-46
# 4 US-ORv 1     -0.783 0.613 1.27e-10
# 5 US-WPT 2     -0.681 0.463 4.08e-20

重新制作一个st_flux212 st_plumber170: 可补充49个站点

st = rbind(
  st_plumber170[site %!in% st_flux212$site, .(site, lon, lat, source = "plumber2")],
  st_flux212[, .(site, lon, lat, source = "fluxnet2015")]
)
# IT-Ro1: both
st_flux261 = rbind(
    st[site %!in% sites_bad, ], 
    select(st2, -group)) |> 
  arrange(site)

st[site %in% sites_bad, ]
st_flux261[site %in% sites_bad, ]

usethis::use_data(st_flux261, overwrite=TRUE)
# 5个站点的位置需要调整

已经对站点位置进行了修正, 但修正代码丢失

# 补充冠层高度信息
sites = st_flux261$site

st = site_metadata[SiteCode %in% sites, .(site = SiteCode, 
  z_obs = MeasurementHeight, 
  z_canopy = CanopyHeight,
  z_tower = TowerHeight,
  IGBP_vegetation_short, IGBP_vegetation_long, 
  CABLE_PFT, 
  Exclude, 
  Exclude_reason
)]

st[!is.na(z_obs)]
st2[!is.na(z_obs)] # 217站点有
st2[!is.na(z_canopy)] # 217站点有

sites_good = st_plumber170$site
st2 = rbind(
  st[site %!in% sites_good, .(site, z_obs, z_canopy)],
  st_plumber170[, .(site, z_obs, z_canopy)]
)

meta = fread("data-raw/Site_metadata.csv")
# st = site_metadata[SiteCode %in% sites, .(site = SiteCode, 
#   z_obs = MeasurementHeight, 
#   z_canopy = CanopyHeight,
#   z_tower = TowerHeight,
#   IGBP_vegetation_short, IGBP_vegetation_long, 
#   CABLE_PFT, 
#   Exclude, 
#   Exclude_reason
# )]
# st[!is.na(z_obs)]