将自然年转换为水文年时间序列 Rasterbrick R

Question

给定一个带有自然年份的rasterbrick (Jan-Dec)，如何将其转换为具有水文年份（从1st October and ending on the following 30th September 开始）？与here 的问题类似（但不准确）。

样本数据：

     nl <- 768
     s <- brick(nrows = 510,  ncols = 1068,
                   xmn = -180, xmx = 180, ymn = -90, ymx = 90,
                   crs = "+proj=longlat +datum=WGS84",
                   nl = nl,values=TRUE)
        dates <- seq(as.Date("1950-01-01"), as.Date("2013-12-31"), by = "month")
        s <- setZ(s, dates)
vals <- 1:ncell(s)
s[]=vals

Answer 1

我刚才wrote a function这个。

to_wateryear <- function(thedate){
  thedate = as.Date(thedate)
  year = as.integer(substr(thedate, 1, 4))
  cutoff =  cutoff = as.Date(paste(year, '-10-01', sep=''))
  return(ifelse(thedate < cutoff, 
           year*1000 + thedate - as.Date(paste(year - 1, '-09-30', sep='')),
           (year + 1)*1000 + thedate - as.Date(paste(year, '-09-30', sep=''))))
}

例子：

to_wateryear(as.Date(Sys.time()))
to_wateryear('2013-10-01')
# leap year behavior
to_wateryear('2012-09-30')
to_wateryear('2013-09-30')
to_wateryear('2012-02-29')
to_wateryear('2013-03-31')

编辑从您的 cmets 看来，您真正想要的是按水年拆分您的栅格砖，而不是根据水文年来表示日期。

library(raster)
nl <- 768
s <- brick(nrows = 510,  ncols = 1068, 
  xmn = -180, xmx = 180, ymn = -90, ymx = 90,
  crs = "+proj=longlat +datum=WGS84",
  nl = nl,values=TRUE)

dates <- seq(as.Date("1950-01-01"), as.Date("2013-12-31"), by = "month")
# use water year
s <- setZ(s, to_wateryear(dates))
vals <- 1:ncell(s)
s[]=vals

# split by water year
hyears = unique(getZ(s) %/% 1000)
res = vector("list", length(hyears))
for(i in 1:length(hyears))
  res[[i]] = s[[which(getZ(s) %/% 1000 == hyears[i])]]
names(res) = paste0("HY", hyears)