Skip to content

Latest commit

 

History

History
47 lines (44 loc) · 2.2 KB

todo.md

File metadata and controls

47 lines (44 loc) · 2.2 KB
Raw

Development tasks

todo

  • Inversion:
    • Reduce east/west regularization towards the pole
    • Explore Michael's method to determine regularization parameters
    • penalize deviation from previous timestep? Add prior model vector to run_inversion()
    • transition from east-west to plain gradient or amplitude from 70-80 mlat?
    • minimize gradient
    • boundary condition: HMB
    • Take into account cell distortion in regularization (minimize volume charge density, not sum of line charge densities)
    • Minimize E dot J instead of charge?
  • Conductance ideas:
    • use Knight relation on FACs from previous timestep?
    • use electric field (Cousins / Merkin)
    • Use ovation prime in addition to EUV?
    • Find optimal example with SSUSI etc.
    • Empirical orthogonal functions based on auroral images?
  • Data handling:
    • add capability to save
    • find out how to design (xarray? pandas?)
    • make checks on input data
  • SECS:
    • Implement Fukushima's correction for inclined field
    • Handle induced magnetic field in a better way (mirror current, Liisa's method, ...)
  • User friendliness:
    • include magnetic coordinates in interactive display
    • Plotting tool for cubed sphere projections - like polplot
    • Helper function to solve M-I coupling Poisson equation
  • Code organization:
    • find a better solution for the return_shape keyword (it is handled in decorator now)
  • Run for Nina's dates: 2014-01-21, 2014-12-18, 2014-12-19
  • Mapping of convection data in data.py or dataloader.py (with apexpy mapping functions?)
  • Handle radius coordinate from satellite measurements/magnetometers in a better way (geocentric/geodetic)
  • Calculate resolution with resolution matrix
  • Allow for FAC at any position (not just gridded) as input
  • Add option to specify neutral wind

long-term todo

  • Take into account inclination of magnetic field (but still height-integrated)
  • model residuals from empirical model to reduce boundary effects (HV's idea)
  • nonlinear solver (solve simultaneously for conductance and electric field)
    • Possible approach: Use principal component analysis for conductance, or scaling parameter for conductance patterns
  • add option to solve for neutral wind
  • expand to 3D