How to Read a Stargazing Forecast: Cloud, Seeing, Transparency, Moon & Kp

Every useful stargazing forecast is built from the same five inputs, and they are not equally important:

1. Cloud cover (%) — the biggest gate. A clear sky beats a dark sky every time.
2. Moon phase + rise/set window — a full moon at a dark site is still a washed-out sky.
3. Darkness (Bortle / SQM) — how much artificial light pollution sits over your site.
4. Transparency — how clear the air is (humidity, smoke, haze).
5. Seeing — how steady the atmosphere is, which matters most at high magnification.

When you add Kp (geomagnetic activity) you are asking a second question: is there also an aurora worth chasing?

The order above is also the decision order. Most failed stargazing nights die at step one or step two, long before seeing or transparency get a vote.

Cloud cover is usually given as a percentage (0–100%) for each forecast hour. The honest reading is not binary — it is about where the cloud is and which way it is moving.

The practical scale:

• 0–20% — effectively clear; the best-case for deep-sky work.
• 20–40% — mostly clear, but watch for patchy bands; a passing cloud can still cost you a meteor shower hour.
• 40–60% — mixed. Some targets work, others are lost. Only worth it for bright objects or a short, specific window.
• 60–80% — marginal. Most deep-sky targets are gone; aurora chances drop sharply.
• 80–100% — not a stargazing night. No number further down the page can rescue this.

The trend matters more than the current value. A forecast that shows 80% at 9pm and 20% by midnight is a real opportunity if you can wait it out; 10% at 9pm climbing to 70% by 1am is a night that dies after the first hour.

Weather providers such as MET Norway (used by Darkest Hour) publish short-range cloud cover updated several times daily. Because cloud moves in bands — especially in maritime regions like New Zealand — the forecast made at lunchtime can be wrong by dusk. Always re-check within an hour of leaving.

The moon is the brightest object in the night sky by a huge margin. Even at 30–40% illumination it hides faint structure, and a full moon at a Bortle 1 site is still a washed-out sky.

The two questions to ask:

• What is the phase? The week around new moon is the prime deep-sky window every month. Around full moon, plan for planets, the moon itself and bright clusters — not the Milky Way.
• When is it up? Phase alone is not enough. A last-quarter moon that rises at 1am still destroys the deep-sky hours after it appears. Check moonrise and moonset for your exact coordinates, not just the date.

The practical pattern: find your moonless window first, then fit everything else inside it. If the moon sets at 11pm, your window is 11pm–3am; if it rises at 3am, your window ends at 3am. The best nights are those where the moon is down for the entire dark period.

A good forecast tool shows the moon's illumination and its rise/set times for your site. Darkest Hour's tonight card does both, so you can see the window before you commit to the drive.

Once the sky is clear and the moon is down, three subtle numbers decide how good the night really is:

Darkness (Bortle / SQM) is the site's ceiling — how much artificial light pollution washes out faint objects. Bortle 4 is the realistic sweet spot for most observers: dark enough for a real Milky Way, close enough to reach in an hour or two. Bortle 1–3 is where deep-sky work shines; Bortle 5–6 still works for planets, the moon and bright clusters. Treat Bortle maps as comparisons, not measurements — most are modelled estimates, and local terrain or a nearby town can shift the real value by a class or two.

Transparency is how clear the air is. Humidity, smoke, haze and thin cirrus all scatter light and flatten the sky. A high-transparency night makes even a Bortle 5 site perform well; a smoky, humid night ruins a Bortle 2 site. Transparency is the difference between 'the map says dark' and 'the sky actually looks dark'.

Seeing is atmospheric steadiness — how sharply stars twinkle and how stable high-magnification views are. Poor seeing blurs planets and the moon at high magnification and ruins fine planetary detail, but matters little for wide-field Milky Way viewing and most deep-sky astrophotography.

The mistake most people make: choosing a site by darkness alone. A Bortle 3 site under high humidity and cloud loses to a Bortle 5 site on a clear, moonless, transparent night — every time.

The Kp index is a 0–9 planetary measure of geomagnetic disturbance — the headline number in aurora forecasts. It is an extra gate, not a substitute for the first five:

• Kp 0–2 — quiet; aurora stays near the polar regions.
• Kp 3–4 — active; aurora can dip toward mid-latitudes on the poleward side.
• Kp 5 (G1 storm) — visible from southern New Zealand and the northern US/Scandinavia.
• Kp 6–7 (G2–G3) — reaches further toward populated latitudes; photographed from Wellington and southern Australia.
• Kp 8–9 (G4–G5) — severe to extreme; rare, and potentially visible from very low latitudes.

The honest rule: Kp tells you if a storm is likely; your local sky tells you if you will see anything. A Kp 7 with 100% cloud is nothing. A Kp 4 with a clear, dark, moonless sky and an open horizon can be a real glow.

Watch the Bz trend too — a strongly negative (southward) Bz for hours is a much better live signal than a Kp spike in one three-hour window. For the full decoder, read our aurora forecasts explained guide.

Here is a realistic forecast for a southern New Zealand site on a winter night, and how a planner reads it:

Verdict: Go. The four main gates are open, and the aurora number adds upside rather than risk. The plan: be set up by 9pm, point wide-field south for the aurora watch, then sweep the Milky Way core until the cloud band arrives.

Now change one number — cloud at 70% — and the honest verdict flips to Wait, because every other number was conditional on the sky being clear. That is the whole discipline of reading a forecast: the weakest gate decides the night.

This is exactly what Darkest Hour's tonight check automates: it pulls the live provider feeds for your coordinates — cloud from MET Norway, Kp from NOAA SWPC, plus moon position and modelled darkness — and returns a single honest verdict (Go / Worth checking / Wait), with every source and update time labelled on the card.

The workflow that works:

1. Days out — watch the trend and the moon phase. The week around new moon is your planning baseline.
2. A day out — check the cloud outlook for the site, not your living room. If it looks plausible, commit.
3. The night of — re-check within an hour of leaving. Weather moves; a verdict that flips is the tool working.

When you want to compare a few candidate sites or scan the whole week for the best night, the Pro planner ranks nearby sites against the same live forecasts, finds your clear windows and shows the best night of the week. See Pro pricing — the free tier covers a single spot, and you can decide before you pay.