Skip to contents

Snow normals from cd's ERA5-Land annual snow layers

Source: R/wet_snow_normals.R
wet_snow_normals.Rd

Reads cd's annual snow layers (one band per year, ERA5-Land at 0.1 degree, built by cd from the Earth Data Hub hourly store), takes the per-cell mean over years, and resamples it bilinearly onto grid. The mean is cd::cd_baseline()'s definition (the mean of the year values, NA removed), taken per cell instead of over an area. Every year in years must be a band of the file read.

Usage

wet_snow_normals(
  grid,
  vars = c("snowmelt_doy_50", "swe_max"),
  years = 1981:2010,
  dir = "data/snow",
  catalog = NULL,
  cap = c(swe_max = 3000),
  overwrite = FALSE
)

Arguments

grid

Target grid: a SpatRaster or a path to one, in EPSG:4326.

vars

cd variable names, each with an "annual" row in catalog.

years

Years to average.

dir

Cache directory.

catalog

A data.frame with columns variable, period and href, as from cd::cd_catalog(); NULL reads cd's default catalog. A remote href is read through cd::cd_cache_fetch().

cap

Named numeric: an upper limit for the per-cell mean of each named variable, applied before resampling. Variables not named are not capped.

overwrite

Logical. Rebuild the resampled grid even when cached.

Value

Path to a GeoTIFF with one layer per variable, named as in vars. The cache key is the grid, vars, years, cap, the method version and the values of the bands read (not their files, which cd rebuilds with a new run time whenever it appends a year).

Details

The defaults are the two predictors the monthly shares were scored with (#16): snowmelt_doy_50, the calendar day of the year by which half the year's snowmelt has happened, and swe_max, the year's peak snow water equivalent (mm).

ERA5-Land has no cell over the sea, so cells of grid whose own ERA5-Land cell is empty come out NA (GDAL's bilinear reweights around missing neighbours); they are left for the caller to fill.

ERA5-Land's glacier points carry perennial snow: over 1981-2010, 513 BC land cells have a mean swe_max of 5,000 to 10,000 mm, against a seasonal maximum of about 3,000 mm elsewhere. cap limits each cell's mean before resampling, so a glacier cell still reads as deep snow without dominating an upstream mean (decided for #16 before any score was seen).

Examples

if (FALSE) { # interactive()
g <- terra::rast(xmin = -127, xmax = -126, ymin = 54, ymax = 55, res = 1 / 120)
sn <- terra::rast(wet_snow_normals(g, dir = tempdir()))
terra::global(sn, "mean", na.rm = TRUE)
}