Estimate sub-daily shortwave radiation from daily means

Builds a clear-sky irradiance curve from solar geometry and rescales it so that each day's mean reproduces the supplied daily mean MET_radswd. This preserves the observed daily energy while giving a physically shaped diurnal cycle with genuine night-time zeros.

Usage

estimate_hourly_swr(
  met,
  lat,
  lon,
  tz = NULL,
  timestep = c("hour", "3hour"),
  interval = c("ending", "beginning", "instant"),
  cloud = 0
)

Arguments

met

data frame with a Date column (class Date, or POSIXct at daily resolution) and MET_radswd in W m-2.

lat, lon

position in decimal degrees.

tz

timezone the daily dates refer to, and in which the sub-daily output is labelled. Default NULL, which uses the tz attribute of met when present, otherwise "UTC". Pass a fixed-offset zone such as "Etc/GMT-12" (NZST) to place the peak at local solar noon.

timestep

"hour" (default) or "3hour".

interval

what a timestamp denotes: "ending" (default) means the value is the mean over the interval ending at that label, which is the convention of ERA5 accumulated fluxes and hence of extract_era5_hourly_met(); "beginning" is the mean over the following interval; "instant" is the instantaneous value at the label. Averaging over an interval is done on a 10-minute sub-grid, so sunrise and sunset steps are handled correctly.

cloud

optional cloud fraction, scalar or one value per row of met, used to damp the clear-sky curve before rescaling (the rescale makes the daily total insensitive to this, but it changes the shape slightly).

Value

data frame with Date (POSIXct in tz) and MET_radswd (W m-2). Days whose clear-sky mean is zero (polar night) return zeros.

Details

Unlike the equivalent routine in AEME, timestamps are handled explicitly: daily dates are expanded into sub-daily steps in tz, the solar geometry is evaluated at the corresponding UTC instants, and the result is returned labelled in tz. The shortwave peak therefore lands at local solar noon rather than being offset by the UTC offset.

Examples

met <- data.frame(Date = seq(as.Date("2024-01-01"), by = "day",
                             length.out = 5),
                  MET_radswd = c(300, 120, 280, 90, 310))
hr <- estimate_hourly_swr(met, lat = -38.08, lon = 176.27)
head(hr)
#>                  Date MET_radswd
#> 1 2024-01-01 00:00:00   831.2122
#> 2 2024-01-01 01:00:00   843.2225
#> 3 2024-01-01 02:00:00   805.7792
#> 4 2024-01-01 03:00:00   721.3019
#> 5 2024-01-01 04:00:00   596.4732
#> 6 2024-01-01 05:00:00   441.8697