Propagation
The one part of LogRunner that touches nothing. It fills no field, changes no QSO and sets no frequency. It answers the question that comes before reaching for the tuning knob: is anything happening at all – and if so, where?
The figures come from hamqsl.com (N0NBH). LogRunner fetches them at most once an hour and keeps the answer locally. With no connection the last fetch stays on screen with its timestamp; after three hours without fresh figures the panel marks them out of date.
| Card | What it is |
|---|---|
| SFI | Solar flux at 10.7 cm in SFU – the figure everything else hangs on |
| SN | Sunspot number |
| A | A index, daily average of geomagnetic disturbance |
| K | K index, the last three hours |
| MUF | Maximum usable frequency, in MHz |
| X-Ray | Solar X-ray flux, in the usual class notation (C1.8) |
| Geomag | State of the geomagnetic field (QUIET, UNSETTLED, ACTIVE, …) |
| S/N | Noise level in S units (S1-S2) |
The abbreviations stay the same in every language – a solar flux localised into something else would be a figure nobody recognises. The full name is in the tooltip.
The cards say how conditions are. The question anybody actually acts on is a different one: are they getting better or worse?
A K index of 3 that has come down from 5 means the disturbance is dying away and the bands are coming back. A K index of 3 that has come up from 1 means a disturbance is arriving. The same number, the opposite decision. A single reading cannot tell you which.
Four of the eight figures can be carried forward over time. Each of them therefore has an arrow, a difference and a small line:
SFI 152 ▲ +6 ╱‾ SN 87 ▲ +4 ╱‾ A 7 ▼ −2 ‾╲ K 2 ▼ −2 ‾╲
For solar flux and sunspot number that is the rising arrow; for the A and K index it is the falling one. So a green arrow points up on one line and down on the next. That is not a bug, it is the whole design:
- The shape of the arrow says what the number did.
- The colour says what that means.
The wording follows the colour rather than the arrow too: a falling K index reads improving.
Each figure has its own threshold. Below it, the figure counts as steady.
| Figure | Movement needed for an arrow | Rising means | Falling means |
|---|---|---|---|
| SFI | more than 3 SFU | better | worse |
| SN | more than 3 | better | worse |
| A | more than 2 | worse | better |
| K | any movement at all | worse | better |
The K index gets no margin: nine whole-numbered steps, and every one of them is a different picture of the bands. Solar flux, by contrast, wanders by two or three points between readings without meaning anything.
The feed likes to jump a point back and forth between readings. An arrow that flipped on every one of those jumps would be noise dressed up as information.
A direction is therefore only shown once the previous reading gave the same direction. Otherwise the panel says steady. The price: a genuine turn is reported one reading late. For a figure that arrives hourly, that is the right side to be wrong on.
One exception: while there is no reading before last to ask, the single difference on hand is shown – a history two readings deep would otherwise sit at steady whatever the sun was doing.
Below the cards is one line for somebody who reads nothing else on the panel. It has three parts, and they answer three different questions:
Overall trend: [▲ improving +7] ▓▓▓▓▓▓▓░░░░░ 54 ╱‾ Period [3h] 6h 12h 24h direction level shape
1. The coloured chip – which way it is going. Arrow, word and difference, by the same rules as the individual figures: green means better. The threshold here is 5 points – a weighted average of four figures moves less than any one of them, so it gets its own, larger threshold. The two-reading confirmation applies here as well.
2. The bar and the number – how good it is right now. A value between 0 and 100. It stays deliberately in the accent colour and does not borrow the green of the trend: "how good" and "which way" are two different statements, and drawn in one colour neither would be readable any more.
3. The small line – what shape the recent history had.
Each of the four figures is first put on a scale of nought to one, with nought always the worse end. The ranges are not the theoretical extremes but the span between "as bad as it gets" and "as good as it gets":
| Figure | Range | Direction | Weight |
|---|---|---|---|
| SFI | 60 – 220 SFU | higher is better | 35 % |
| SN | 0 – 200 | higher is better | 15 % |
| A | 0 – 10 | lower is better | 25 % |
| K | 0 – 9 | lower is better | 25 % |
Values outside the range are clamped: a solar flux of 300 is not three times as good as one of 220, it is simply the top of the scale.
Solar flux carries the most weight because it decides which bands exist at all. The sunspot number carries the least because it is largely the same statement made a second time. The two geomagnetic figures together weigh as much as the flux: they cannot open a band, but either of them can shut every one of them.
Worked example – SFI 152, SN 87, A 7, K 2:
| Figure | Normalised | × weight |
|---|---|---|
| SFI 152 | (152 − 60) / 160 = 0.575 | 0.201 |
| SN 87 | 87 / 200 = 0.435 | 0.065 |
| A 7 | 1 − 7/10 = 0.300 | 0.075 |
| K 2 | 1 − 2/9 = 0.778 | 0.194 |
| Total | 0.536 → score 53.6 |
If a figure is missing from the fetch it drops out and the remaining weights carry the rest of the way – a document without a sunspot number is scored on the other three (55.4 in this example) rather than being marked down for a number that was never there.
It is meant to be compared with itself and with nothing else. Four readings do not combine into a measurement of anything; there is no scale in propagation science on which "54" means something. What the number can do is answer the question is this better than it was three hours ago? That is the entire reason it exists.
Accordingly it is not comparable with figures from other programs, not to be arithmetically set against yesterday in absolute terms, and not a forecast.
To the right on the same line: 3 / 6 / 12 / 24 hours, three by default. This choice decides both things:
- what every difference is measured against – every arrow, every number behind an arrow and the overall trend;
- how far back the small lines reach.
The comparison is made against the most recent reading that is at least as old as the period. If that particular reading is missing the figure, an older one is used – but only up to one more period back. Past that, no comparison is shown at all, rather than one taken over quite a different stretch of time than the one on the button.
Three hours is roughly three readings and the shortest span over which the feed has moved at all. A day is as far back as a trend still says anything about today.
Lines and nothing else: no axes, no scale, no grid. What is being asked is not "how high is the solar flux" – the number is right beside it – but "what shape has it had lately", and in a line barely twenty pixels tall every additional marking makes that harder rather than easier.
Each line scales to its own extremes. A rise in the A index is therefore visible even on a quiet day; the height says nothing absolute in exchange, which is exactly why the number stands next to it. The timestamp of a point is in the tooltip.
Only readings inside the chosen period are drawn – the older ones are kept, but only so that the differences have something to be measured against. Beyond 32 points the series is thinned evenly rather than cut off at the old end: a line drawn from half a period would show a shape the period did not have.
Below the overall trend is the table the panel is really about: four band groups × day and night, each cell green, amber or red. These are N0NBH's verdicts, passed through unchanged.
Below that, smaller, is what the feed says about VHF – aurora and sporadic E, by region. On most days it reads Band closed everywhere; the one day it does not should be on screen without having to be asked for.
The fourth column, Trend, comes from LogRunner's own history instead. Which figure governs which group of bands is not this program's invention but stands in every propagation primer: the high bands need ionisation and live or die by the solar flux, the low bands have all the ionisation they need and are ruined by geomagnetic disturbance instead.
| Band group | Governed by | Second say |
|---|---|---|
| 80 m – 40 m | K index | A index |
| 30 m – 20 m | K index | SFI |
| 17 m – 15 m | SFI | K index |
| 12 m – 10 m | SFI | K index |
If the governing figure has nothing to say, the second one speaks.
This is not a forecast and is not offered as one. The column says: the figure this group of bands hangs on is improving – which is something entirely different, and a great deal smaller, than the claim that the band will be open.
Nobody publishes the history of these figures in a form that could be fetched – so LogRunner keeps its own. Every successfully fetched reading goes into a small database of its own (data/propagation.sqlite), 30 days rolling. Thirty, because the sun turns once in twenty-seven days: the recurrence of a coronal hole is the longest pattern anybody reads off these four numbers.
The key is the feed's own timestamp, not the moment of the fetch. The feed is rebuilt hourly; two fetches within the same hour are therefore the same reading and overwrite each other.
What follows from that in practice:
- The trend is never read over the network. It comes off the disk and is therefore there without a connection.
- A deleted cache (
solar.json) costs the cards, not the lines. - A deleted history costs the lines, not the cards.
- The file may be deleted at any time; what is lost is a month of small lines and nothing else.
- The foot of the panel says how many readings are stored, and the tooltip gives the path.
While too few readings are on hand, the panel says so in words. Four cards without arrows would otherwise look like a broken display rather than a waiting one.