<a id="recipes-wenzel16jclim"></a>

# Legacy recipe: Multiple ensemble diagnostic regression (MDER) for constraining future austral jet position

#### NOTE
This recipe has been retired and moved to the folder “archive”. This recipe
is only included for documentation purposes and not maintained any more. The
last known ESMValTool version supporting this recipe is version 2.13.0.

## Overview

[Wenzel et al. (2016)](https://journals.ametsoc.org/doi/full/10.1175/JCLI-D-15-0412.1) use multiple ensemble diagnostic regression (MDER) to
constrain the CMIP5 future projection of the summer austral jet position with
several historical process-oriented diagnostics and respective observations.

The following plots are reproduced:

* Absolute correlation between the target variable and the diagnostics.
* Scatterplot between the target variable and the MDER-calculated linear
  combination of diagnostics.
* Boxplot of RMSE for the unweighted multi-model mean and the (MDER) weighted
  multi-model mean of the target variable in a pseudo-reality setup.
* Time series of the target variable for all models, observations and MDER
  predictions.
* Errorbar plots for all diagnostics.
* Scatterplots between the target variable and all diagnostics.

## Available recipes and diagnostics

Recipes are stored in archive/legacy_recipes

> * recipe_wenzel16jclim.yml

Diagnostics are stored in archive/legacy_diag_scripts

> * austral_jet/asr.ncl
> * austral_jet/main.ncl
> * mder/absolute_correlation.ncl
> * mder/regression_stepwise.ncl
> * mder/select_for_mder.ncl

## User settings in recipe

1. Preprocessor
   * `extract_region`: Region extraction.
   * `extract_levels`: Pressure level extraction.
   * `area_statistics`: Spatial average calculations.
2. Script austral_jet/asr.ncl
   * `season`, *str*: Season.
   * `average_ens`, *bool*, optional (default: `False`): Average over all
     given ensemble members of a climate model.
   * `wdiag`, *array of str*, optional: Names of the diagnostic for MDER
     output.  Necessary when MDER output is desired.
   * `wdiag_title`, *array of str*, optional: Names of the diagnostic in
     plots.
3. Script austral_jet/main.ncl
   * `styleset`, *str*: Style set used for plotting the multi-model plots.
   * `season`, *str*: Season.
   * `average_ens`, *bool*, optional (default: `False`): Average over all
     given ensemble members of a climate model.
   * `rsondes`, *array of str*, optional: Additional observations used in the
     plot but not for MDER output.
   * `rsondes_file`, *array of str*, optional: Paths to the additional
     observations Necessary when `rsondes` is given.
   * `rsondes_yr_min`, *int*, optional: Minimum year for additional
     observations. Necessary when `rsondes` is given.
   * `rsondes_yr_max`, *int*, optional: Maximum year for additional
     observations. Necessary when `rsondes` is given.
   * `wdiag`, *array of str*, optional: Names of the diagnostic for MDER
     output.  Necessary when MDER output is desired.
   * `wdiag_title`, *array of str*, optional: Names of the diagnostic in
     plots.
   * `derive_var`, *str*, optional: Derive variables using NCL functions.
     Must be one of `"tpp"`, `"mmstf"`.
   * `derive_latrange`, *array of float*, optional: Latitude range for
     variable derivation.  Necessary if `derive_var` is given.
   * `derive_lev`, *float*, optional: Pressure level (given in *Pa*) for
     variable derivation.  Necessary if `derive_var` is given.
4. Script mder/absolute_correlation.ncl
   * `p_time`, *array of int*: Start years for future projections.
   * `p_step`, *int*: Time range for future projections (in years).
   * `scal_time`, *array of int*: Time range for base period (in years) for
     anomaly calculations used when `calc_type = "trend"`.
   * `time_oper`, *str*: Operation used in NCL `time_operation` function.
   * `time_opt`, *str*: Option used in NCL `time_operation` function.
   * `calc_type`, *str*: Calculation type for the target variable. Must be
     one of `"trend"`, `"pos"`, `"int"`.
   * `domain`, *str*: Domain tag for provenance tracking.
   * `average_ens`, *bool*, optional (default: `False`): Average over all
     given ensemble members of a climate model.
   * `region`, *str*, optional: Region used for area aggregation. Necessary
     if input of target variable is multidimensional.
   * `area_oper`, *str*, optional: Operation used in NCL `area_operation`
     function. Necessary if multidimensional is given.
   * `plot_units`, *str*, optional (attribute for `variable_info`): Units
     for the target variable used in the plots.
5. Script mder/regression_stepwise.ncl
   * `p_time`, *array of int*: Start years for future projections.
   * `p_step`, *int*: Time range for future projections (in years).
   * `scal_time`, *array of int*: Time range for base period (in years) for
     anomaly calculations used when `calc_type = "trend"`.
   * `time_oper`, *str*: Operation used in NCL `time_operation` function.
   * `time_opt`, *str*: Option used in NCL `time_operation` function.
   * `calc_type`, *str*: Calculation type for the target variable. Must be
     one of `"trend"`, `"pos"`, `"int"`.
   * `domain`, *str*: Domain tag for provenance tracking.
   * `average_ens`, *bool*, optional (default: `False`): Average over all
     given ensemble members of a climate model.
   * `smooth`, *bool*, optional (default: `False`): Smooth time period with
     1-2-1 filter.
   * `iter`, *int*, optional: Number of iterations for smoothing. Necessary
     when `smooth` is given.
   * `cross_validation_mode`, *bool*, optional (default: `False`): Perform
     cross-validation.
   * `region`, *str*, optional: Region used for area aggregation. Necessary
     if input of target variable is multidimensional.
   * `area_oper`, *str*, optional: Operation used in NCL `area_operation`
     function. Necessary if multidimensional is given.
   * `plot_units`, *str*, optional (attribute for `variable_info`): Units
     for the target variable used in the plots.
6. Script mder/select_for_mder.ncl
   * `wdiag`, *array of str*: Names of the diagnostic for MDER output.
     Necessary when MDER output is desired.
   * `domain`, *str*: Domain tag for provenance tracking.
   * `ref_dataset`, *str*: Style set used for plotting the multi-model plots.
   * `average_ens`, *bool*, optional (default: `False`): Average over all
     given ensemble members of a climate model.
   * `derive_var`, *str*, optional: Derive variables using NCL functions.
     Must be one of `"tpp"`, `"mmstf"`.

## Variables

* *ta* (atmos, monthly, longitude, latitude, pressure level, time)
* *uajet* (atmos, monthly, time)
* *va* (atmos, monthly, longitude, latitude, pressure level, time)
* *ps* (atmos, monthly, longitude, latitude, time)
* *asr* (atmos, monthly, longitude, latitude, time)

## Observations and reformat scripts

* ERA-Intermin (*ta*, *uajet*, *va*, *ps*)
* CERES-EBAF (*asr*)

## References

* Wenzel, S., V. Eyring, E.P. Gerber, and A.Y. Karpechko: Constraining Future
  Summer Austral Jet Stream Positions in the CMIP5 Ensemble by Process-Oriented
  Multiple Diagnostic Regression. J. Climate, 29, 673–687,
  doi:10.1175/JCLI-D-15-0412.1, 2016.

## Example plots

<a id="fig-wenzel16jclim-1"></a>
![image](recipes/figures/wenzel16jclim/CMPI5_uajet-pos_rcp45_20ystep_FIG1.png)

<a id="fig-wenzel16jclim-2"></a>
![image](recipes/figures/wenzel16jclim/CMPI5_uajet-pos_rcp45_20ystep_FIG2b.png)

<a id="fig-wenzel16jclim-3"></a>
![image](recipes/figures/wenzel16jclim/CMPI5_uajet-pos_rcp45_20ystep_FIG3.png)

<a id="fig-wenzel16jclim-4"></a>
![image](recipes/figures/wenzel16jclim/ta_trop250_ta_DJF_trend.png)

<a id="fig-wenzel16jclim-5"></a>
![image](recipes/figures/wenzel16jclim/uajet_H-SH_c.png)
