mesmer.create_emulations.make_realisations#
- mesmer.create_emulations.make_realisations(preds_lt, params_lt, params_lv, params_gv_T, time, n_realisations, seeds, land_fractions)#
Make climate realisations based on pre-calculated MESMER parameters
- Parameters:
preds_lt (dict) – nested dictionary of predictors for local trends with keys
[pred][scen] (1d/ 2d arrays (time)/(run, time) of predictor for specific scenario)
params_lt (dict) – dictionary with the trained local trend parameters
[“targs”] (emulated variables, str)
[“esm”] (Earth System Model, str)
[“method”] (applied method, str)
[“method_each_gp_sep”] (states if method is applied to each grid point separately, bool)
[“preds”] (predictors, list of strs)
[“scenarios”] (emission scenarios used for training, list of strs)
[xx] (additional params depending on method employed)
[“full_model_contains_lv”] (whether the full model contains part of the local variability module, bool)
params_lv (dict) – dictionary with the trained local variability parameters
[“targ”] (variable which is emulated, str)
[“esm”] (Earth System Model, str)
[“method”] (applied method, str)
[“preds”] (predictors, list of strs)
[“scenarios”] (scenarios which are used for training, list of strs)
[xx] (additional keys depend on employed method)
params_gv_T (dict) –
[“targ”] (variable which is emulated, str)
[“esm”] (Earth System Model, str)
[“method”] (applied method, str)
[“preds”] (predictors, list of strs)
[“scenarios”] (scenarios which are used for training, list of strs)
[xx] (additional keys depend on employed method and are listed in train_gv_T_method() function)
time (dict) –
[“scenario”] timepoints (1D np.ndarray) used for training of the scenario (note that hist and scenario e.g. ssp126 are kept separate)
n_realisations (int) – Number of realisations to draw
seeds (dict) –
- [“esm”] (dict):
- [“scenario”] (dict):
[“gv”] (seed for global variability) [“lv”] (seed for local variability)
land_fractions (xr.DataArray) – Land fractions of each cell. Used to convert the MESMER outputs back onto grids.