Make a multi-site simulation using lists of climate, soil, and parameter input objects.

Wrapper function for run_LWFB90 to make multiple parallel simulations of one or several parameter sets, for a series of sites with individual climate and soil, or individual parameter sets for each climate/soil combinations.

run_multisite_LWFB90(
  options_b90,
  param_b90,
  soil = NULL,
  climate,
  climate_args = NULL,
  all_combinations = FALSE,
  cores = 2,
  show_progress = TRUE,
  ...
)

Arguments

options_b90: Named list of model control options to be used in all simulations
param_b90: Named list of parameters to be used in all simulations, or a list of multiple parameter sets.
soil: Data.frame with soil properties to be used in all simulations, or a list of data.frames with different soil profiles.
climate: Data.frame with climate data, or a list of climate data.frames. Alternatively, a function can be supplied that returns a data.frame. Arguments to the function can be passed via climate_args.
climate_args: List of named lists of arguments passed to climate, if this is a function.
all_combinations: Logical: Set up and run all possible combinations of individual param_b90, climate and soil objects? Default is FALSE, running one object or the list of param_b90 objects for a series of climate/soil combinations.
cores: Number of cores to use for parallel processing.
show_progress: Logical: Show progress bar? Default is TRUE. See also section Progress bar below.
...: Further arguments passed to run_LWFB90.

Value

A named list with the results of the single runs as returned by run_LWFB90. Simulation or processing errors are passed on. The names of the returned list entries are concatenated from the names of the input list entries in the following form: <climate> <soil> <param_b90>. If climate is a function, the names for <climate> are taken from the names of climate_args.

Data management

The returned list of single run results can become very large, if many simulations are performed and the selected output contains daily resolution data sets, especially daily layer-wise soil moisture data. To not overload memory, it is advised to reduce the returned simulation results to a minimum, by carefully selecting the output, and make use of the option to pass a list of functions to run_LWFB90 via argument output_fun. These functions perform directly on the output of a single run simulation, and can be used for aggregating model output on-the-fly, or for writing results to a file or database. The regular output of run_LWFB90 can be suppressed by setting rtrn.output = FALSE, for exclusively returning the output of such functions. To provide full flexibility, the names of the current soil, param_b90, and climate are automatically passed as additional arguments (soil_nm, param_nm,clim_nm) to run_LWFB90 and in this way become available to functions passed via output_fun. In order to not overload the memory with climate input data, it is advised to provide a function instead of a list of climate data.frames, and specify its arguments for individual sites in climate_args, in case many sites with individual climates will be simulated.

Progress bar

This function provides a progress bar via the package progressr if show_progress=TRUE. The parallel computation is then wrapped with progressr::with_progress() to enable progress reporting from distributed calculations. The appearance of the progress bar (including audible notification) can be customized by the user for the entire session using progressr::handlers() (see vignette('progressr-intro')).

Examples

# \donttest{
data("slb1_meteo")
data("slb1_soil")

opts <- set_optionsLWFB90(budburst_method = "Menzel", enddate = as.Date("2002-12-31"))

# define parameter sets
param_l <- list(spruce = set_paramLWFB90(maxlai = 5,
                                         budburst_species = "Picea abies (frueh)",
                                         winlaifrac = 0.8),
                beech = set_paramLWFB90(maxlai = 6,
                                        budburst_species = "Fagus sylvatica",
                                        winlaifrac = 0))

soil <- cbind(slb1_soil, hydpar_wessolek_tab(slb1_soil$texture))

# define list of soil objects
soils <- list(soil1 = soil, soil2 = soil)

# define list of climate objects
climates <- list(clim1 = slb1_meteo, clim2 = slb1_meteo)

# run two parameter sets on a series of climate and soil-objects
res <- run_multisite_LWFB90(param_b90 = param_l,
                      options_b90 = opts,
                      soil = soils,
                      climate = climates)
names(res)
#> [1] "clim1 soil1 spruce" "clim2 soil2 beech" 

# set up and run individual parameter sets for individual locations

# set up location parameters
loc_parm <- data.frame(loc_id = names(climates),
                       coords_y = c(48.0, 54.0),
                       eslope = c(30,0),
                       aspect = c(180,0))

# create input list of multiple param_b90 list objects
param_l <- lapply(names(climates), function(x, loc_parms) {
  parms <- set_paramLWFB90()
  parms[match(names(loc_parm),names(parms), nomatch = 0)] <-
    loc_parm[loc_parm$loc_id == x, which(names(loc_parm) %in% names(parms))]
  parms
}, loc_parm = loc_parm)

names(param_l) <- c("locpar1", "locpar2")

res <- run_multisite_LWFB90(param_b90 = param_l,
                      options_b90 = opts,
                      soil = soils,
                      climate = climates)
names(res)
#> [1] "clim1 soil1 locpar1" "clim2 soil2 locpar2"
# }