Skip to contents

Extract daily CMIP6 / CCAM projection meteorology for a point

Source: R/extract_climate_point.R
extract_climate_point.Rd

Reads local daily climate-projection netCDF files - one variable per file, as produced by the NIWA CCAM downscaling of CMIP6 - samples them at a point, decodes the model calendar to real dates, and returns the series in AEME MET_* names and units. Built for the delta-change step of scenario_workflow: extract the model historical and future series with this function, form monthly change factors, and apply them to a bias_correct_daily_baseline().

Usage

extract_climate_point(
  path,
  lon,
  lat,
  vars = c("MET_tmpair", "MET_pprain", "MET_wndspd", "MET_radswd", "MET_humrel",
    "MET_radlwd", "MET_prsttn"),
  experiments = NULL,
  method = c("bilinear", "nearest"),
  calendar = c("auto", "365_day", "360_day", "366_day", "standard"),
  tz = "UTC",
  strict = FALSE,
  verbose = TRUE
)

Arguments

path

directory holding the .nc files, or a character vector of file paths.

lon, lat

point of interest, decimal degrees (WGS84).

vars

variables to return, as AEME MET_* names or CMIP short names (do not mix the two). Default is the seven mapped variables; only those present on disk are returned.

experiments

optional character vector to keep (e.g. c("historical", "ssp245")); NULL (default) returns every experiment found.

method

"bilinear" (default) interpolation between the four surrounding grid nodes, or "nearest" grid node.

calendar

"auto" (default, read from the time variable's calendar attribute), or force one of "365_day", "360_day", "366_day", "standard". A 365_day / 360_day series is mapped onto real dates, so it has no 29 February (and, for 360_day, no 31st).

tz

time zone recorded on the result; the series is daily and carries no time of day. Default "UTC".

strict

if TRUE, any NA introduced by merging variables with mismatched date coverage raises an error instead of a warning/message.

verbose

print each file as it is read.

Value

a data frame with Date (class Date), experiment (character) and one column per requested variable - stacked over experiments, wide over variables. Attributes lon, lat, method, tz.

Details

Files are matched by the convention <variable>_<experiment>_*.nc (for example tas_ssp245_NorESM2-MM_CCAM_daily_NZ5km_bc.nc -> variable tas, experiment ssp245). Recognised variables and their AEME targets: tas -> MET_tmpair, pr -> MET_pprain, sfcWind -> MET_wndspd, rsds -> MET_radswd, hurs -> MET_humrel, rlds -> MET_radlwd, ps -> MET_prsttn.

Units are converted from each file's own units attribute: Kelvin -> degC, kg m-2 s-1 -> mm day-1, metres -> mm. Everything else is passed through (the CCAM pr files are already mm day-1). _FillValue is applied by ncdf4.

See also

scenario_workflow, bias_correct_daily_baseline(), disaggregate_met_to_hourly()

Examples

if (FALSE) { # \dontrun{
cmip <- extract_climate_point("inst/extdata/rotorua_cmip6",
                            lon = 176.2717, lat = -38.0790)
split(cmip, cmip$experiment) |> lapply(head, 2)
} # }