Homogenize validation data.sets as arguments to cost_likelihood_pmodel() and cost_likelihood_biomee() #236
Description
After the inputs to the running of the model is homogenized (#228),
also the inputs to the calibration would need to be homogenized.
Some work has already been done to reduce code redundundancy in the likelihood/RMSE functions in PR: #233
Either within that PR or after it, we should also homogenize the validation data sets, i.e. rsofun::biomee_validation, rsofun::p_model_validation_vcmax25, and rsofun::p_model_validation.
- Update the definition of the validation data (see below code snippet that compares the current status)
- Update the
rda-files and create their generation script similarly to: Moving generate_biomee/pmodel_driver_data.R from FluxDataKit to rsofun #232 (comment)
# TODO: homogenize format rsofun::biomee_validation with rsofun::p_model_validation_vcmax25
# for pmodel it is a wide data structure, and for biomee it is a long data structure
# rsofun::p_model_validation |> tidyr::unnest(data)
# # A tibble: 2,190 × 4
# sitename date gpp gpp_unc
# <chr> <date> <dbl> <dbl>
# 1 FR-Pue 2007-01-01 2.21 0.0108
# 2 FR-Pue 2007-01-02 2.23 0.00475
# 3 FR-Pue 2007-01-03 2.48 0.00727
# 4 FR-Pue 2007-01-04 1.71 0.00516
# rsofun::p_model_validation_vcmax25 |> tidyr::unnest(data)
# # A tibble: 4 × 3
# # Groups: sitename [4]
# sitename vcmax25 vcmax25_unc
# <chr> <dbl> <dbl>
# 1 Reichetal_Colorado 0.0000339 0.0000136
# 2 Reichetal_New_Mexico 0.0000757 0.0000163
# 3 Reichetal_Venezuela 0.0000472 0.0000164
# 4 Reichetal_Wisconsin 0.0000502 0.0000147
# rsofun::biomee_validation |> tidyr::unnest(data)
# sitename variables targets_obs
# <chr> <chr> <dbl>
# 1 CH-Lae GPP 1.86
# 2 CH-Lae LAI 6.49
# 3 CH-Lae Density 296.
# 4 CH-Lae Biomass 44.5