Monthly Distribution Fitting on Gridded Data — fitlm_monthly

Gridded counterpart of fitlm_monthly that fits a list of candidate distributions to every calendar month of a NetCDF raster or an sxts object via the L-moments method. The input is first normalised to an sxts object (accepting sxts, Raster*, or NetCDF filename and variable name), and each calendar month present in the record is extracted as a sub-grid. Per-month fitting is dispatched to fitlm_nc, which delegates to fitlm_nxts for per-cell L-moment fitting and returns parameter rasters, theoretical and sample L-moment rasters, GoF rasters, and a density GoF comparison plot. Parallelism is controlled by parallel_by: "cells" parallelises the cell-level fits within each month, whereas "months" parallelises across months but is serial across cells. The shared_memory flag determines whether the grid is shared with workers via a file-backed big.matrix or serialised in column chunks (suitable for multi-node plans). This is reccomended for single machine usage and especially windows for efficiency and reduced RAM consumption.The returned list is named by month and contains the full fitlm_nc output for each.

Usage

fitlm_monthly_nc(
  data = NULL,
  filename = NA,
  varname = NA,
  candidates = "norm",
  ignore_zeros = FALSE,
  zero_threshold = 0.01,
  parallel = FALSE,
  ncores = 2,
  shared_memory = TRUE,
  parallel_by = c("cells", "months"),
  order = NULL,
  ...
)

Arguments

data: An sxts object, or a Raster* object. Leave NULL when supplying filename and varname.
filename: A NetCDF file name to import if data is not provided.
varname: The name of the variable to extract from filename.
candidates: A character vector of distribution names to fit.
ignore_zeros: A logical value, if TRUE zeros will be ignored. Default is FALSE.
zero_threshold: The threshold below which values are considered zero. Default is 0.01.
parallel: Logical, whether to use parallel processing.
ncores: Number of cores to use for parallel computations. Default is 2.
shared_memory: Logical, when parallel on the cell axis, share the grid with workers via a file-backed big.matrix (mmap, single machine) instead of serialising column chunks. Set FALSE for multi-node plan(cluster) setups. Default TRUE.
parallel_by: Character, the axis to parallelise over when parallel = TRUE: "cells" (default) parallelises the grid-cell fits within each month via fitlm_nxts; "months" parallelises the per-month fits, each run serially across cells.
order: Optional named list mapping a candidate name to the vector of L-moment orders matched by its optimiser, e.g. list(gengamma = 1:5, expweibull = 1:3). Only the numerically-fitted distributions accept it; passed through to fitlm_nc. Default NULL.
...: Additional arguments passed to nc2xts when filename and varname are provided.

Value

A named list with one element per calendar month present in the data (named after month.name). Each element is the standard fitlm_nc output: a list with fit_results (per-candidate rasters) and gof_plots.

Examples

if (FALSE) { # \dontrun{
# Simulated 3-cell grid over two years
n <- 730
dates <- seq.Date(as.Date("2020-01-01"), by = "day", length.out = n)
vals <- cbind(cell1 = rgamma(n, shape = 0.8, scale = 3),
              cell2 = rgamma(n, shape = 1.2, scale = 2),
              cell3 = rgamma(n, shape = 0.6, scale = 4))
coords <- data.frame(lon = c(10, 11, 12), lat = c(45, 46, 47))
grid <- sxts(vals, order.by = dates, coords = coords,
             projection = "+proj=longlat +datum=WGS84")

monthly_fits <- fitlm_monthly_nc(grid, candidates = c("exp", "gamma3"),
                                  ignore_zeros = TRUE)
names(monthly_fits)
} # }