ortk95
commented
3 months ago Doing this per-object looks like it'll end up quite complicated (and potentially fragile). I think it will need a whole new target creating dynamically in __init__, with different radii, and all associated kernels, IDs, variables etc. created for each new radius.
It will probably therefore be easier to implement this within each method by adding an alt parameter to any functions which convert to lonlats - i.e. this basically completes the work of #359 (which added an alt parameter for conversions from lonlats). I think the best way to do this will be to define a context manager that temporarily adjusts the kernel variables containing the radii (e.g. BODY599_RADII) when needed. For example, something like:
class AdjustAltitude:
def __init__(self, target_id: int, alt: float):
self.target_id = target_id
self.alt = alt
def __enter__(self):
self._original_radii = spice.bodvar(self.target_id, 'RADII', 3)
spice.pdpool(
f'BODY{self.target_id}_RADII', [r + self.alt for r in self._original_radii]
)
def __exit__(self, exc_type, exc_val, exc_tb):
spice.pdpool(f'BODY{self.target_id}_RADII', self._original_radii)
def radec2lonlat(ra: float, dec: float, *, alt: float = 0.0) -> tuple[float, float]:
with AdjustAltitude(code, alt):
...Would obviously need to test this properly to ensure the variables are changed correctly, in particular ensuring that the radius value in the main kernel pool isn't ever left with an adjusted value. Would probably want to add some short-circuit logic for the alt=0 case.
We should add an optional offset to the altitude of the target body's 'surface' level, e.g. to allow mapping of atmospheric structure above/below the nominal 1-bar 'surface'.
~Should be able to implement this with a new kwarg in
Body, which then adjusts ther_eqandr_polarvalues accordingly in__init__. I think everything should then flow automatically from this (mapping, wireframes, coordinate conversions etc.), though obviously need to check this.~...2lonlatfunctionsSATURN (699) +70kmalt=...in the appropriate methods, but we can always think about making the context manager public at some later point if the need arises.