Skip to content

Latest commit

 

History

History
133 lines (43 loc) · 4.27 KB

TODO.md

File metadata and controls

133 lines (43 loc) · 4.27 KB
Raw

  • 320km => test decrease of Qangmin !!!

  • Maintenant que "le netoyage des valeurs exteme" est retire de deformation.py, etre sur que tout ce qui utilise les deformations pour faire des stats applique le netoyage des valeurs exteme! En particulier le scaling (PDFs fait!)

  • Le pb, c que dans tous les cas, le tracking se base sur ce qu'il y a dans les fichiers quads Q*.npz qui eux n'ont pas l'info des quads qui vont disparaitre parce-que valeurs extremes de deformation => trouver un fix!!!!

  • refaire les PDFs de base pour 10km (min area max and min for quads, take +1 -1 km2 at 10km!) pour RGPS et modele (avec la nouvelle method seeding / quads!!! pour le modele)

  • control where deformations are cleaned (tiny values) when doing P90

  • control ZDiv<-zmin_div in mkpdf.py !!!!

  • check why NANs in qud generation logs about stat on quads!!!

  • double-check how are really canceled quads with inconsistent time across vertices, and in particular for quad at t=t0 and the one at t=t0+dtNom !!!

    • the one at t0+dtNom should be checked, and the two canceled if any time fuckup at vertices !!!

  • review the use of areaR in lTisOK() !

  • review the rscore in lQisOK()

  • add the reading and use (inside 01_selection_xy.py) of the new array jtBinN saved in 1st npz file by 00_scan_batches.py

  • in TEST_trackin/SI3_only, apparently proble for: [GetTimeSpan]: will start using model record ... says 1 record too late!!!

  • Use the nearest U and V point in ice part tracker! And not 0.5(U(i-1)+U(i)) as for now !!!

  • Use A at t and not A at centered time t+0.5*dt, when estimating partial derivatives??? (Anton and Pierre say I should!)

  • Fix the output file name of PDF npz files created by statistics: must include first and last stream, at least the year, and resolution !!!

  • deformation.py rejects often because some buoys' time position is too far away from the mean time of the whole ice plate, this ctiterion should be applied in the 1st scrit: 01_selection_xy.py !!! And these buoys removed!

  • Commencer les PDFs plus loin du 0 que maintenant (y a encore le decrochage du shear!)

    • comme 0.0015 plutot que les 0.001 actuel??

  • PDFs: ne pas utiliser les tiny values depuis le debut, pour que le PDF soit bien correctement normalise!

    • aussi: utiliser des bins croissants (ou decroissant) en taille (log ou exp) pour eviter les problemes en log-log

  • Ajouter la partie sous-sampling du nuage de point qui marchait dans RGPS => ajouter a generate_quad_mesh !!!

  • Ajouter l'option de stop a une date donnee dans ice_part_track !!!

  • In quadrangle class, make a function that gives the nominal resolution of all the quadrangles....

    • then can be used,for example, in deformation.py (where it is now infered from file name...)

  • pour la conversion degree-> km => utiliser uniquement fonction and util !!!

  • trouver pourquoi dans deformation.py le UM4 ne resemble pas au UMc, cela devrait seulement etre une moyenne spatiale !!!

  • coder le trajectory (et son seeding initial a n'importe quelle location) direct from scratch

  • utiliser l'interpolation en temps pour RGPS (et seeder le modele avec donnees interpolees au 1er janvier)

In recycle quadrangles: maybe add constraint on quadrangle Area??? Cause it can increas a lot in 3 days...

Retain mainly the constraint on quadrangle area in selection (be uber tolerant on H/L ration and min and max angle...)

In Tri2Quad() remove the pnam argument and instead use the names of points defined in the class!!! (that might be strings of the point indices if explicit names given) !!! => so in the classes add the possibility to provide the vector of names: shape = (nP) of string

Polygon

Class Quadrangle

  • should not contain all the points from original triangles

    • QuaPointIDs is based on the IDs from the cloud point on which triangles were constructed => so from 0 to nT-1
    • .QuaPointIdx is the equivalent but in the reference of only the points used for Quads => so from 0 to nQ-1 (nQ<nT) I guess that should be the 2nd we use as default as in the future use of this class we won't min about all the disregarder triangle points...

    So QuaPointIdx should become QuaPointIDs and QuaPointIDs should be something like QuaPointIDsTri