CSTools (version 5.2.0)

RainFARM: RainFARM stochastic precipitation downscaling (reduced version)

Description

This function implements the RainFARM stochastic precipitation downscaling method and accepts in input an array with named dims ("lon", "lat") and one or more dimension (such as "ftime", "sdate" or "time") over which to average automatically determined spectral slopes. Adapted for climate downscaling and including orographic correction. References: Terzago, S. et al. (2018). NHESS 18(11), 2825-2840. tools:::Rd_expr_doi("10.5194/nhess-18-2825-2018"), D'Onofrio et al. (2014), J of Hydrometeorology 15, 830-843; Rebora et. al. (2006), JHM 7, 724.

Usage

RainFARM(
  data,
  lon,
  lat,
  nf,
  weights = 1,
  nens = 1,
  slope = 0,
  kmin = 1,
  fglob = FALSE,
  fsmooth = TRUE,
  nprocs = 1,
  time_dim = NULL,
  lon_dim = "lon",
  lat_dim = "lat",
  drop_realization_dim = FALSE,
  verbose = FALSE
)

Value

RainFARM() Returns a list containing the fine-scale longitudes, latitudes and the sequence of nens downscaled fields. If

nens > 1 an additional dimension named "realization" is added to the output array after the "member" dimension (if it exists and unless

drop_realization_dim = TRUE is specified). The ordering of the remaining dimensions in the exp element of the input object is maintained.

Arguments

data

Precipitation array to downscale. The input array is expected to have at least two dimensions named "lon" and "lat" by default (these default names can be changed with the lon_dim and lat_dim parameters) and one or more dimensions over which to average these slopes, which can be specified by parameter time_dim. The number of longitudes and latitudes in the input data is expected to be even and the same. If not the function will perform a subsetting to ensure this condition.

lon

Vector or array of longitudes.

lat

Vector or array of latitudes.

nf

Refinement factor for downscaling (the output resolution is increased by this factor).

weights

Multi-dimensional array with climatological weights which can be obtained using the CST_RFWeights function. If weights = 1. (default) no weights are used. The names of these dimensions must be at least the same longitudinal and latitudinal dimension names as data.

nens

Number of ensemble members to produce (default: nens = 1).

slope

Prescribed spectral slope. The default is slope = 0. meaning that the slope is determined automatically over the dimensions specified by time_dim. A 1D array with named dimension can be provided (see details and examples).

kmin

First wavenumber for spectral slope (default: kmin = 1).

fglob

Logical to conseve global precipitation over the domain (default: FALSE).

fsmooth

Logical to conserve precipitation with a smoothing kernel (default: TRUE).

nprocs

The number of parallel processes to spawn for the use for parallel computation in multiple cores. (default: 1)

time_dim

String or character array with name(s) of time dimension(s) (e.g. "ftime", "sdate", "time" ...) over which to compute spectral slopes. If a character array of dimension names is provided, the spectral slopes will be computed over all elements belonging to those dimensions. If omitted one of c("ftime", "sdate", "time") is searched and the first one with more than one element is chosen.

lon_dim

Name of lon dimension ("lon" by default).

lat_dim

Name of lat dimension ("lat" by default).

drop_realization_dim

Logical to remove the "realization" stochastic ensemble dimension (default: FALSE) with the following behaviour if set to TRUE:

1. if nens == 1: the dimension is dropped;

2. if nens > 1 and a "member" dimension exists: the "realization" and "member" dimensions are compacted (multiplied) and the resulting dimension is named "member";

3. if nens > 1 and a "member" dimension does not exist: the "realization" dimension is renamed to "member".

verbose

logical for verbose output (default: FALSE).

Author

Jost von Hardenberg - ISAC-CNR,

Details

Wether parameter 'slope' and 'weights' presents seasonality dependency, a dimension name should match between these parameters and the input data in parameter 'data'. See example 2 below where weights and slope vary with 'sdate' dimension.

Examples

# Example for the 'reduced' RainFARM function 
nf <- 8   # Choose a downscaling by factor 8
exp <- 1 : (2 * 3 * 4 * 8 * 8)
dim(exp) <- c(dataset = 1, member = 2, sdate = 3, ftime = 4, lat = 8, lon = 8)
lon <- seq(10, 13.5, 0.5)
lat <- seq(40, 43.5, 0.5)
# Create a test array of weights
ww <- array(1., dim = c(lon = 8 * nf, lat = 8 * nf))
res <- RainFARM(data = exp, lon = lon, lat = lat, nf = nf, 
               weights = ww, nens = 3, time_dim = 'ftime')