Changes between Version 77 and Version 78 of Pirates2022/Tuning2023


Ignore:
Timestamp:
11/15/23 09:29:27 (8 months ago)
Author:
jmignot
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • Pirates2022/Tuning2023

    v77 v78  
     1=== 15 novembre 2023 === 
     2GGac, Juliette, Brady 
     3 
     4* Outil dash:  
     5  * GGas a implémenté les régressions dans l'outil (lineaire, polynomial, exponentiel) + intervalle de confiance (stat.t de scipy). Cela necessite de mettre a jour l'environnement conda. Ggac doit mettre sa doc a jour.  
     6  * Param-param avec metrique couleur, pour se rapprocher des diag de Fred.  
     7  * '''Juliette : demander login forge Ggac''' + ecrire Myriam AMIP VLR.  
     8 
     9* En parallèle du boulot de Brady:  
     10  * Explorer métriques de glace de mer: quelles corrélations, quelles paramètres? cf [4c] 
     11  * Implémenter metriques strato Brady pour verifier lien avec autres metriques 
     12  * automatiser l'identification des 10 metriques les + correlees. ou faire des PCA? -> apres les petits exercices nn_etau et glace de mer à la main.   
     13  * Explorer effet specifique de nn_etau=1 [5d] 
     14 
     15 
    116=== 8 novembre 2023 === 
    217GGac, GGas, Juliette, Brady, Julie 
    … …  
    4191. on the topic glob.rt [Ggas] 
    520 
    6     slope between SST and glob.rt different in VLR whether we use glob.rt from AMIP LR vs VLR -> implies to redo preconditioning AMIP in VLR 
    7 (edited) 
    8  
    9     warning : conclusion drawn on only 5 points yet 
    10  
    11    [1.a] but crash test of preconditioning as done this summer not so wrong for VLR -> implies that it is the preconditioning AMIP in LR that should be redone ! -> first we need more members AMIP VLR to make this point solid 
     21    * slope between SST and glob.rt different in VLR whether we use glob.rt from AMIP LR vs VLR -> implies to redo preconditioning AMIP in VLR 
     22    * warning : conclusion drawn on only 5 points yet 
     23   * [1.a] but crash test of preconditioning as done this summer not so wrong for VLR -> implies that it is the preconditioning AMIP in LR that should be redone ! -> first we need more members AMIP VLR to make this point solid 
    1224 
    13252. stratosphere vs troposphere metrics (what should be added to preconditioning ?) 
    1426 
    15    [2.a] LR AMIP vs CMIP pretty well aligned for some metrics (150hPa, mid lat) -> encouragning 
    16  
    17    [2.b] LR AMIP vs VLR CMIP not so well aligned for those metrics -> needs to confirm with more VLR AMIP ? (edited)  
    18  
    19    [2.c] 150hPa or 200hPa ? seem to have different impacts on sea ice in arctic, depending on LR or VLR 
    20    
    21    [2.d] adding stratosphere metrics, what for ? need a synthesis of climate variables affected by stratosphere metrics - and make sure that including troposphere metrics directly would not make a better job 
    22  
    23    [2.e] contact Francois Lott to imagine other metrics describing stratosphere ? 
     27   * [2.a] LR AMIP vs CMIP pretty well aligned for some metrics (150hPa, mid lat) -> encouraging 
     28   * [2.b] LR AMIP vs VLR CMIP not so well aligned for those metrics -> needs to confirm with more VLR AMIP ?  
     29   * [2.c] 150hPa or 200hPa ? seem to have different impacts on sea ice in arctic, depending on LR or VLR 
     30   * [2.d] adding stratosphere metrics, what for ? need a synthesis of climate variables affected by stratosphere metrics - and make sure that including troposphere metrics directly would not make a better job 
     31   * [2.e] contact Francois Lott to imagine other metrics describing stratosphere ? 
    2432 
    25333. overall tuning of VLR and LR 
    2634 
    27    [3.a] southern ocean warm bias systematic in VLR, much larger than in LR 
    28  
    29    [3.b] stratosphere warm bias more dramatic in VLR than in LR (cf progress from Myriam in reducing stratosphere bias via change in vertical levels ?)  
     35   * [3.a] southern ocean warm bias systematic in VLR, much larger than in LR 
     36   * [3.b] stratosphere warm bias more dramatic in VLR than in LR (cf progress from Myriam in reducing stratosphere bias via change in vertical levels ?)  
    3037 
    31384. sea ice 
    3239 
    33    [4.a] adding atmospheric metrics in preconditioning could help increasing sea ice - can we prove that ? 
    34  
    35    [4.b] very different behavior in VLR vs LR in southern hemisphere - because not enough ice in VLR (cf strong warm bias in t2m) ?  
    36  
    37    [4.c] which metrics / parameters  have the most impact on sea ice - north and south ? 
     40   * [4.a] adding atmospheric metrics in preconditioning could help increasing sea ice - can we prove that ? 
     41   * [4.b] very different behavior in VLR vs LR in southern hemisphere - because not enough ice in VLR (cf strong warm bias in t2m) ?  
     42   * [4.c] which metrics / parameters  have the most impact on sea ice - north and south ? 
    3843 
    39445. upper ocean variables (the blue ones…) 
    4045 
    41    [5.a] southern ocean warm bias, already mentioned in 4.b and 
    42  
    43    [5.b] what else ? 
    44  
    45    [5.c] does the 60yr long simulations provide a different picture from the 20yr long ones ? 
    46  
    47    [5.d] can we say something interesting about the 5 LR1 simulations ? (edited)  
     46   * [5.a] southern ocean warm bias, already mentioned in 4.b and 
     47   * [5.b] what else ? 
     48   * [5.c] does the 60yr long simulations provide a different picture from the 20yr long ones ? 
     49   * [5.d] can we say something interesting about the 5 LR1 simulations ?  
    4850 
    49516. atmospheric variables : can we identify where there is an added value in running in CMIP rather than AMIP, when tuning the atmosphere parameters ?