Usable range is the lowest and highest band that meets the
requirement on that circuit — the practical version of LUF and MUF. It is
narrower than the raw MUF, because a frequency can be below the MUF and still be
too weak to work.
The shaded half is night. The bright curve is the terminator, the
“grey line” operators watch: the absorbing D layer decays quickly after
sunset while the reflecting F layer lingers, so paths along it often run long.
Power and antenna: the grid is computed once at 100 W into an
isotropic antenna, and your station is applied on top. Power is a flat offset —
10 log₁₀(W/100) — which scales the signal while the noise at the
far end is unchanged, so it is exact. Neither power nor antenna requires regenerating
the map.
A modelled antenna's gain varies across the map, and that is the point.
Pick one from the top group and its real elevation pattern is applied per cell: the
distance to a cell sets the take-off angle that path needs, and the pattern gives the
gain at that angle and bearing. A dipole 10 m up shows roughly 13 dB between
the distance it favours and the ones it does not, and swings further still off the ends
of the wire. No P.533 output changes — this is geometry laid over the same grid,
which is why switching antennas is still instant.
The fixed-gain entries are peak gain, applied flat , exactly as
before. Beams stay in that group because their gain does not follow from the thin-wire
model this uses, so a Yagi here remains optimistic on paths whose take-off
angle it does not favour. For a wire antenna, prefer the modelled entry.
Modelled gain assumes perfect ground and stops at 3°. Over a
perfect reflector a horizontal antenna has an absolute null at the horizon, which is an
idealisation, not a fact — so the lookup is clamped at 3° of elevation and no
gain is allowed below −20 dBi. Without those limits a modelling artifact
would erase whole regions of the map. Real ground also costs 1–2 dB that
none of this charges you for, and more for a ground-mounted vertical. The
antenna visualiser shows the pattern behind these numbers
and states its own limits.
Receive antenna gain is deliberately excluded — at HF the
receiver is limited by external noise, so a bigger receive antenna lifts signal and
noise together and largely cancels out of SNR.
The aurora overlay is observation, not prediction. It is NOAA's
OVATION model — where aurora is likely in the next half hour — drawn over
the map. It does not change a single number underneath it. What it adds is the thing a
Kp reading cannot give you: where . A path crossing the oval is the one to
worry about; a path well clear of it is largely unaffected no matter what Kp reads.
The grid is 2° and anything below 2% probability is dropped as noise.
The 24-hour chart is the exact circuit. The map trades precision
for breadth: a 6° cell, and by default four snapshots of the day. The chart is the
other side of that trade — one path, computed at its real coordinates, every hour
and every band. It is a single prediction, so it takes a fraction of a second once the
engine is running, which is why it is on demand rather than precomputed for every
receiver.
Coarse passes are coarse. A 24° sample is painted across the
whole block it stands for, so the early picture is a sketch of its own neighbourhood
rather than a prediction for every point in it. Let the finer passes land before
reading much into a boundary. Receiver figures always snap to the nearest 6° cell
(up to ~330 km).
Engine ITU-R P.533 (WebAssembly).
Space weather: NOAA SWPC (public domain) and KC2G (attribution required), fetched by
our own service so your browser never contacts them.
Coastlines: Natural Earth (public domain). Places: GeoNames (CC BY 4.0).