Harold E. Johnson wrote,
It ends
when the wave front arriving at the receiving antenna becomes planar. ie, to
function efficiently in the far field, the receiving antenna needs to
intercept a planar wavefront. That is, the individual rays need to be
arriving in parallel. If the distance between antennas is very great, that
is very nearly the case.
If the capture area of the receiving antenna is great relative to the
distance to the source, the received energy arrives as non parallel rays
that basically reach the receiving antenna out of phase with each other and
partially cancel. So, the gain of antennas measured in the "near field",
where the received energy is not a planar wavefront, will be in error. The
distance to the end of the near field is highly dependent on the gain of the
antenna and with UHF and SHF antennas often exhibiting very high gain, their
near fields can be and often are very large.

Balanis divides the near-field region into two parts: a reactive near-field
R0.62 square root(D^3/Lambda) where D is the largest antenna dimension,
Lambda is the wavelength, and R is the distance from the antenna surface,
and a radiating near-field region R2D^2/Lambda. The far-field he defines as
anything greater than 2D^2/Lambda. He gives exceptions to these rules, so
take them with a grain of salt.

73,
Tom Donaly, KA6RUH