Alexander wrote:
> Certanly exponential factor exp(-r/a) should be present.
> I mentioned this before.

Yes you did!  I am following your lead with this 1/sqrt(r)
thing and the exponential factor.  And it is working - it is
giving the right answer for Stefan's signal!

3dB per 1000km (daytime) is mentioned in some literature but
I have to follow that up.    Assuming that, we have the
estimate for daytime flux density at range r:-

   B = sqrt( 9.5e-21 * ERP/r) * exp( -r/2.9e6)

> But this should be calculated. At the earth bondary one can
> use Leontovich bondary condition

Yes.

> But what to do with ionosphere boundary... To neglect dissipation
> in ionosphere looks not reasonable.

Yes too.  The ionosphere boundary is more complicated.  There are
two values for the refractive index - attenuation depends on
direction relative to Earth's magnetic field.  The result is
phase and polarisation changes.   I have some references to follow
up here too.   But it is complicated and time consuming.  It
would be nice to just use average figures for day and night
attenuation - surely good enough for 'amateur' work.

> To solve the problem (rather volumable problem) is nice task
> for physicist.

The hard work was done, decades ago, and appears to be fairly
well established.  But it is a major job to go through the
literature and pick out what is relevant.  It is a heady mix
of plasma physics, EM theory and ray tracing.  There are one
or two textbooks - one I will order today.
--
Paul Nicholson
--