Best places by clouds @ eclipse time
The strongest candidate sites re-ranked by the live eclipse-hour cloud forecast (selected model), with the 26-year climatology alongside and distances from your spot. Until the forecast horizon opens, the ranking falls back to climatology.
Your observing spot
Your spot and tapped points are kept in the page link — hit 🔗 sync and open the link on your phone/laptop to carry the exact setup across devices.
Cloud coverage — every dataset @ eclipse hour
One row per point — your observing spot plus every point you tap on the map — one column per dataset: the 26-year ERA5 climatology, the repo's auto-refreshed cloud snapshot, and six live forecast models at the eclipse hour (~20:30 CEST). ± is the model spread (small = the models agree); ☁→☀ is how much of your line of sight to the low Sun is blocked where it crosses the low/mid/high cloud decks (selected model).
Diagnostics
When something is not working — a blank map above all — this answers it with facts: which build is running, what the service worker is doing, and which tile host this device can actually reach, tested one by one, right now. A host that fails here is blocked on your connection, not broken in the app.
Offline — for the low-signal day
The app itself already works with no network: the whole interface, the eclipse engine, the 149 towns, the climatology and the relief overlay are stored on your device the first time you open it. Map tiles are not — only the ones you have actually looked at. Pull the ones you will need before you leave, while you still have signal.
Tapped map points
Tap anywhere inside the totality band on the Map tab — the point lands here and joins the dataset matrix above (newest first, up to 9; re-tapping a point bumps it back to the top).
Hourly cloud decks — this spot
Reading the pressure (hPa column above & the line under the decks): ≥ 1022 strong high — sinking air, the clear-evening machine · 1015–1021 weak high — usually fine · 1008–1014 flat — the number means little, trust the trend · < 1008 low — unstable, cloud & storm risk. A value rising run-to-run beats a pretty cloud map; falling pressure means cloud may arrive even if the sky looks good now. See the whole field with (H/L centres marked, scrub 17→23 h, slope arrows pointing Low→High) — or switch the overlay to ◔ hPa trend to see, in one colour, where pressure is rising ▲ (low→high, improving) vs falling ▼. Isobar lines, TV-chart style, are also drawn over the ☁ clouds overlay — full interpretation in the .
Pressure over time — this spot
The single most watchable line in eclipse forecasting, from the best-match model. Until 12 Aug enters the horizon it shows the rolling trend (3 days back + the full 16-day horizon); once the date is in range it zooms to the window that matters: 24 h before totality → 12 h after (11 Aug 20:00 → 13 Aug 08:00 CEST). A line climbing into totality = a ridge building = clearing skies; a dive = a front on the way. Colour = the same bands as everywhere else (teal high → red low).
MEASURED test frame: 0 clipped px · 2.04 stops headroom · disk 113.8 px
SHUTTER resolved: 1/320 — confirmed 8 Aug, does not clip
Correction to the earlier note: the spare margin at 1/320 is 0.32 stops measured on the brightest pixel, not a full stop. The "full stop" figure describes the disk centre (48.6% of full scale, 1.04 stops from saturation). Clipping is set by the brightest pixel, so 0.32 stops is the number that matters — still comfortably clear.
The eclipse Sun sits ~10° lower than the test Sun and shines through roughly twice the air (air mass 6.7 vs 3.2), so extinction can only make it dimmer — the 0.32 stops of margin grows, it does not shrink. 1/320 · f/8 · ISO 100, filter on, for the whole partial phase.
If you re-test and the frame does clip — hazy day, higher Sun, different filter sample — fall back to 1/640 and shift the totality bracket one stop faster. That is the only condition that changes this.
- Filter OFFone motion
- Shootat ISO 100 — diamond ringone press · hold for 3 if AEB is on
- ISO 100 → 800three arrow presses
- Shootat ISO 800 — coronaone press · hold for 3 if AEB is on
- Hands off. Look up.the rest of totality is yours
- FILTER ONbefore anything else. Not after checking the screen, not after turning to anyone
- ISO back to 100three presses back
- Resume partialssame frame every 3 min to C4
AEB ±2 EV optional · set once at T−2 min, or drop it
It only widens the safety margin around one shutter speed. It does not replace the fixed-shutter, ISO-only sequence above — that works identically with or without it.
- Do not set AEB days in advance. Set it at T−2 min — the same moment you start the phone and leave the DSLR alone.
- Immediately after: disable Auto Power Off (or its longest interval), for this window only.
- Do not power the camera off again until totality has ended.
- Glance at the three-mark bracket display just before C2 — cheap insurance against a dial bump.
- After C3, powering off is fine. AEB is not needed again.
- If the camera is powered off inside the window (battery swap), AEB must be re-set before the next shot — it will not have survived.
- Filter OFF
- Fire once at ISO 100 — diamond ring
- ISO 100 → 800
- Fire once at ISO 800 — corona
- Hands off. Look up.
THE FRAMES what each one covers, either base speed
Print it. Tape it to the camera back. Everything below is the same card, explained — open only what you need.
①Why this plan, and the three rulesThe reasoning behind every line on the card
At C3 the order is filter first, then everything else — before the screen, before turning to anyone. That is the miss that damages eyes and sensors, and it happens because people are still celebrating C2.
The section carries an older, wider bracket (f/5.6 · ISO 400 · 1/2000 → 1 s, dial-turned by hand). This card overrides it: fixed f/8, a single tested shutter that never moves, AEB doing the bracketing, six frames instead of a dial sweep.
②What the frames coverMeasured scale, the six frames, and the ISO-100-equivalent arithmetic
Measured, not estimated: the test frame puts the solar disk at 113.8 px across — 1 px = 12,232 km. For scale, Earth is 1.04 px. Fine detail is unavailable at this sampling; what reads is the corona's extent and the shape of the diamond ring.
Reference targets, normalised to ISO 100: inner corona ≈ 1/250; mid corona ≈ 1/15 – 1/2. The table below is given for both candidate base speeds, so it stays valid whichever the test selects.
| Base 1/640 · ISO 100 | Frames as shot 1/2500 · 1/640 · 1/160 — diamond ring. |
|---|---|
| Base 1/640 · ISO 800 | As shot 1/2500 · 1/640 · 1/160 → ISO-100-equivalent 1/320 · 1/80 · 1/20. The inner-corona target (1/250) falls inside the bracket; the mid-corona band (1/15 – 1/2) is missed by ~0.4 stop. |
| Base 1/320 · ISO 100 | Frames as shot 1/1250 · 1/320 · 1/80 — diamond ring. |
| Base 1/320 · ISO 800 | As shot 1/1250 · 1/320 · 1/80 → ISO-100-equivalent 1/160 · 1/40 · 1/10. Reaches into the mid-corona band; the inner-corona target sits ~0.7 stop off the fast end. |
ISO-100-equivalent = as-shot shutter × 8, i.e. the three stops between ISO 100 and ISO 800. This assumes a linear ISO scale — standard, but unverified on the 1200D [ASSUMPTION]. The ±2 EV bracket span is wide enough that small nonlinearity does not change which targets are covered.
Prominences will not be captured. At 12,232 km per pixel, a typical prominence (10,000–50,000 km) is 1–4 px. Measured from the test frame scale, not estimated.
③The timeline, step by stepSame rows as the card, with what each one is actually for
④DSLR — the settings line, unpackedWhat the test frame measured, and the one value still open
- AEB ±2 EV is optional, not load-bearing. It widens the exposure safety margin around one shutter speed — it does not replace the sequence, which is identical with or without it: 6 frames with, 2 without.
- Tested: AEB does not survive a power-off on this body; it resets to a single exposure. Sleep and idle are fine.
- Tested: a held shutter with AEB + continuous drive fires exactly 3 frames and stops. It does not loop, so holding is safe.
- If you use AEB, set it fresh once at T−2 min — never days ahead — then disable Auto Power Off and do not power the camera off until after C3. Glance at the three-mark bracket display before C2. If a battery swap forces a power-off inside the window, re-set it before the next shot.
- If you drop AEB: fire once at ISO 100, ISO to 800, fire once — the lower-complexity option, with nothing that has to survive camera state across the event.
- Only two controls move during totality: the filter, and the ISO button. Everything else on this line is locked from C1 to C4.
- Partials: one frame every 3 minutes, filter on, C1 through C4.
- The exposure is measured, and it works. Test frame 2026-08-08 19:01 (1/640 · f/8 · ISO 100 · 55 mm, through the Baader OD 5.0), analysed at raw sensor level: zero clipped pixels in all four channels (max raw 5736 against a white level of 14274), disk centre at 24.3% of full scale = 2.04 stops of headroom, limb darkening cleanly recorded at a limb/centre ratio of 0.434, residual scatter 2.09%, sky background 1.18 ADU above black.
- Resolved: 1/320. One stop slower than the test — 3 clicks counter-clockwise — putting the disk centre at 48.6% of full scale and the brightest pixel at 80.4%: no clipping, 0.32 stops spare. (The margin is measured on the brightest pixel; the disk centre has 1.04 stops. Clipping follows the brightest pixel.)
- Why the altitude rule points the same way: the test Sun was ≈ 18° up (air mass 3.2); the eclipse Sun is ≈ 8.6° (air mass 6.7). Roughly twice the air, so extinction only makes it dimmer — the margin grows. Eclipse lower than test → 1/320.
- The test to run: shoot the filtered Sun at both speeds on the same day, at a Sun altitude as close as possible to the expected eclipse altitude; check each RAW for clipped highlights and visible limb darkening; pick the slowest speed that shows no clipping. If you cannot test near the right altitude, default to 1/640.
- The rule, kept for the record: eclipse altitude lower than, or within ~10° of, the test → 1/320. Higher by more than ~10° → 1/640. Not checked → 1/640. This case: lower, so 1/320.
- Verify the altitude first. Pull C1–C4 and the Sun altitude at maximum for your exact coordinates from timeanddate.com/eclipse or the NASA eclipse pages, and compare with the Sun's altitude at your test time and place.
- Whichever the test picks, it is one value, set once. The shutter does not move again — not for the partials, not through totality.
- Recentre on every single frame. The Sun drifts ~15°/hour; drifting out of frame is the number-one way the partial phase fails.
- Exposure settings survive power-off and dial bumps. Focus ring, zoom ring and tripod framing do not — tape the first two.
- Focus by hand at 10× Live View on the Sun's edge. Autofocus cannot see a filtered Sun.
⑤Phone — the two presets, partials and totalityWhich app, which settings, foil on or off — and the 1-second iOS ceiling
The stock Camera app reads the dark totality sky as night and "fixes" it: Smart HDR lifts the sky and burns the corona to a white blob, Night mode may start multi-second stacking, auto-exposure hunts every time the light changes. It has no manual shutter or ISO, and shooting ProRAW does not fix it. Hence two apps, both configured at home.
- Foil taped over the whole camera island, and the app locked to 1×. If iOS switches lens, it switches to one the foil is not over.
- Halide → Process Zero, full manual (not the ML auto mode).
- MF → infinity, locked. Through foil the frame is black except the Sun; autofocus has nothing to work with.
- ISO 100 · 1/8000 to start — the same brightness the DSLR sees at f/8 · 1/320 · ISO 100, rescaled for the phone's fast lens. Too dark → 1/4000, then ISO 200. Never slower than 1/1000.
- Target: a defined orange disk on a black frame. A white blob with no edge = overexposed, go faster.
- One frame every 5 minutes, recentring each time. This is a scene record, not a disk record — see the note below.
- Blackmagic stays closed during the partials. Its 1/50 video shutter is ~7 stops too slow to shoot through the foil.
- At C2−2 min: foil off the phone, swing it from the Sun to the wide scene, Blackmagic manual and locked, recording, audio on — then hands off until after C3.
- Blackmagic: ISO 800–1600 · 1/50 (1/60 at 30 fps) · WB 5500 K, not auto · MF ∞ · 4K · horizon and foreground in frame.
- Want a phone still as well as the video? Pick one of the three routes below — now, not on the day.
- The unfiltered phone never points at the Sun — not for a "quick look", not at C3. Its job in this window is the landscape and the audio.
- At C3: stop the video when convenient. If you want the phone back on the partial Sun, foil back on first, then the Halide partials preset.
The constraint that decides this: iOS suspends camera capture when an app goes to the background, so leaving Blackmagic to open Halide ends the recording. One phone cannot roll video and shoot stills at the same time. So the question is not "can I", it is "what am I willing to lose".
| A · FRAME GRAB costs nothing — the default |
Shoot 4K and pull the still out of the video afterwards: a 4K frame is ~8.3 MP (3840×2160), which is a perfectly good wide landscape shot. Zero phone handling during totality, zero risk, and you can pick the exact moment later — the diamond ring, the darkest instant, the reaction. If you only want "a photo of totality", this is the answer. |
|---|---|
| B · ONE HALIDE STILL costs the video from that moment |
Only after the DSLR sequence is finished, mid-totality: swipe to Halide (Process Zero, MF ∞) → ISO 800–1600 · 1/4–1 s for the wide scene, or ISO 1600 · 1/15–1/2 s aimed away at the twilight ring → three frames, 5 seconds, phone down. The video is over at that point; do not try to restart it before C3. Budget it as ~15 s of the ~1 min you have — including the fumble. Any hesitation → skip it; rule 3 wins. |
| C · STILLS ONLY, NO VIDEO costs the video entirely |
If the still matters more than the footage, don't start Blackmagic at all. Halide on the tripod, framed wide, two shots: one at C2 + 10 s, one mid-totality, same settings as B. Simpler than B — nothing to abandon mid-window — but you lose the light collapsing and the audio, which is the footage most people actually rewatch. |
What a phone still of the corona is worth: very little. At 1× the Sun's disk is ~1% of the frame, so a "totality photo" from the phone is a dark landscape under a bright dot — which is exactly the shot worth having, because the DSLR cannot take it. Frame it for the horizon, the 360° sunset ring and the people, not for the corona. The 1-second iOS exposure ceiling applies to every route here.
The recommendation: route A. It gets you a photo and the video, and it is the only one of the three that asks nothing of you inside the totality window. Take route B only if you have rehearsed the app switch blind and can do it in under 10 seconds.
What the phone partials are actually worth. At 1× the Sun is roughly 1% of the frame width — you are recording the crescent in a scene, not the disk. The DSLR owns the disk. If that is not what you want from the phone, skip the partials preset entirely and the plan reverts to phone-for-totality-only, with nothing lost.
Test the partials exposure before the day, exactly as the DSLR frame was tested: same foil, same phone, uneclipsed Sun, and pick the fastest setting that still shows a clean disk edge. The 1/8000 figure is derived, not measured [ASSUMPTION] — it scales the measured DSLR exposure by the aperture difference (f/8 → f/1.78 ≈ 4.3 stops), so it is a starting point, not a verified value.
Process Zero skips iPhone image processing entirely and develops from the raw sensor — the only documented way to switch computational photography off at capture.
- Capture mode → Process Zero
- Switch to full manual (not the ML auto mode)
- Main 1× camera only — no ultra-wide, no digital zoom (and the foil only covers so much)
- Manual focus → infinity, locked (autofocus hunts on a black sky)
- Partials preset, foil on: ISO 100 · 1/8000 → a crisp orange disk
- Totality preset, foil off: ISO 800–1600 · 1/4–1 s (wide scene) · ISO 1600 · 1/15–1/2 s (horizon glow)
- Opening Halide mid-totality stops a Blackmagic recording — see the three routes above before you plan on both
- Turn off any auto-ISO / auto-shutter fallback
- Rehearse both presets until switching between them is muscle memory
Full manual video exposure — Halide is stills-only. This device records the light collapsing, and that is the footage you will actually rewatch. Totality only: it is never used through the foil (1/50 is ~7 stops too slow for that) and it never points at the Sun.
- Manual exposure: ISO 800–1600, shutter 1/50 (24/25 fps) or 1/60 (30 fps)
- Manual focus → infinity, locked
- Manual white balance ≈ 5500 K — not auto, or the colour swims as the light drops
- 4K, tripod, framed with horizon and foreground
- Audio ON. People reacting and the birds going quiet is half the value
| Partials · C1 → C2−2 (foil ON) | Halide, Process Zero, manual · 1× · MF ∞ · ISO 100 · 1/8000 · 1 frame / 5 min. Blackmagic closed. |
|---|---|
| C2−2 min | Foil OFF · swing away from the Sun · start the video · hands off |
| Video · C2−2 min → C3 (foil OFF) | Blackmagic · ISO 800–1600 · 1/50 fixed · WB 5500 K · rolling, untouched, on the landscape |
| Totality still — route A | Frame-grabbed from the 4K video afterwards (~8.3 MP). No phone handling during totality. |
| Totality still — route B (ends the video) | Halide, Process Zero · ISO 800–1600 · 1/4 – 1 s · three frames, 5 s, phone down |
| Horizon glow (aimed away, route B or C) | Halide, Process Zero · ISO 1600 · 1/15 – 1/2 s |
| C3 → C4 | Stop the video when convenient. Back on the Sun only with the foil on first, then the partials preset. |
Hard limit: iOS caps a single exposure at 1 second — no app beats it. Night mode's 1–30 s "exposures" have no third-party API and are frame-stacking, which smears the corona.
⚠Still untested — one item leftExposure, AEB behaviour and the ISO control are all answered now
| 1 · Focus | Verify in Live View at 10× on the limb, then tape the focus and zoom rings. The test frame measured brightness, not sharpness — nothing so far says this lens is focused. This is the only open failure point. |
|---|
Answered by test
| ✔ AEB across a power cycle | Does not survive — resets to a single exposure. Handled by setting it at T−2 min and not powering off until after C3, or by dropping AEB entirely. |
|---|---|
| ✔ Held shutter with AEB + continuous | Fires exactly 3 frames and stops. Does not loop. |
| ✔ ISO control | Cross-key arrow confirmed working; 100 → 800 is three presses. |
⚠️Safety — non-negotiableSeven rules; the first two are the ones that blind people
- Filter ON for every second of the partial phase, both sides of totality. Even a deep partial is full-brightness photosphere. It comes off only once the corona is visible and the sky is dark — if you are unsure, it stays on.
- At C3, filter first — before the screen, before speaking to anyone. The bright sliver on the opposite limb is the cue, and it arrives while you are still elated.
- Live View only — never the optical viewfinder with the Sun in frame. A slipped filter puts a focused solar image on your retina.
- Eclipse glasses (ISO 12312-2) on your eyes for every partial phase. A phone screen is not a safe way to watch.
- Do not point the unfiltered phone at the Sun during partials. Sources genuinely conflict on the sensor risk; avoiding it costs nothing.
- Never leave a camera aimed at the Sun unattended with the cap off.
- Lamp-check every filter before every use. Pinpricks fine; tears or daylight shapes → do not use it.
💥If it goes wrongSymptom → fix, in ten seconds, in the dark
| DSLR frames black, filter on | Shutter too fast. Slow to 1/250, bracket up. |
|---|---|
| Blown white disc, no edge detail | Too slow — or not in M. Confirm the mode dial. |
| DSLR won't focus | The lens switch is not on MF. |
| Sun drifted out of frame | Expected. Recentre before every frame. |
| "Busy" between DSLR shots | Long-exposure noise reduction is on. Turn it off. |
| Phone sky grey and lifted, corona blown | Smart HDR — you are not in Halide Process Zero manual. |
| Phone exposure jumping between frames | Auto-exposure still active. Lock it. |
| Phone video colour swimming | White balance on auto. Set it to ~5500 K manually. |
| Not sure totality has actually started | Filter stays on. Wait for corona + dark sky. Two seconds cost nothing. |
| Totality frames all blown white | ISO never came back down after a rehearsal, or the filter is still off from a test. Confirm ISO 100 before C1. |
| Only one frame fired instead of three | AEB was cleared by a power-off (it does that silently), or the drive is on single. Not worth fixing during totality — one frame per ISO is the drop-AEB plan and it works. |
| Missed the diamond ring | Expected if you changed ISO first. The order is filter off → fire → then ISO. |
✅Night beforeTwelve ticks — it remembers them
🎞Afterwards — editing in Apple PhotosNine steps, in order. The order is the point.
This is the recovery path for the totality CR2s. Do it in this sequence — several steps depend on being able to read the histogram, which the last step destroys.
- Open the CR2 → Edit → Adjust.
- Exposure: +1.4. That's your recovery. Type it rather than dragging.
- Highlights: down until the inner ring stops clipping to flat white — you want the ring to show a gradient, not a solid blob.
- Shadows: leave at zero. Lifting shadows lifts sky noise, and there is nothing in the sky to recover.
- Black Point: nudge up until the sky is near-black but not clipped. Watch the histogram's left edge — you want it touching, not piled against, the wall.
- Definition / Clarity: zero. At 118 px it produces halos around the disk, nothing else.
- Sharpening: minimal or none. Same reason.
- Noise Reduction: light. Heavy NR erases the faint outer corona — which is exactly the signal you are trying to keep.
- Crop last. Never first — cropping first hides the histogram context you need for steps 3–5.
Steps 6–8 are all the same argument: at ~118 px there is no fine detail to enhance, so every "enhancement" slider is only free to invent artefacts. The corona's value here is its extent and gradient — protect those and leave everything else alone.
◍What is not verifiedCheck these yourself — do not trust them
- Eclipse Sun altitude is taken from this app's engine for the Meseta (≈ 8.6°); it shifts a degree or so across the band, which does not change the 1/320 decision.
- The test-frame Sun altitude (≈ 18°) is computed for the Geneva area at 19:01 CEST on 8 Aug [ASSUMPTION — the test location was not recorded]. From Salamanca it would be ≈ 27°, which points to 1/320 even more strongly.
- Extinction is now bounded by air mass (3.2 at the test → 6.7 at the eclipse) rather than the generic 0.5–2 stop range, but the extinction coefficient itself is unmeasured — haze or Calima would darken the Sun further, never brighten it.
- ISO linearity on the 1200D — the ISO-100-equivalent conversions assume it [ASSUMPTION].
- Prominences will not be captured — at the measured 12,232 km/px they resolve to 1–4 px. Measured, not estimated.
- Baader mounting — paraphrased; the instructions shipped with the film win where they differ.
- Phone sensor risk during partials — sources conflict. NASA has said phone lenses are small enough not to be damaged, but that newer f/1.7–f/2.0 lenses could be; others assert flatly that sensor burn is real. Unreconciled. Practical resolution: don't point it.
- Halide Process Zero — mechanism documented by the developer and DPReview; no independent test under eclipse-like light found. Moderate confidence on mechanism, low on how the corona renders.
- Halide pricing — varies by App Store region.
- ~58 px Sun at 1× — arithmetic from a ~24 mm-equivalent field and the 0.53° solar diameter.
- Solar disc diameter — now measured at 113.8 px on the 2026-08-08 test frame, superseding the ~118 px estimate from field-of-view arithmetic.
- The reference targets (inner corona ≈ 1/250, mid corona ≈ 1/15 – 1/2 at ISO 100) — standard-practice figures, not measured. They also assume AEB lands where you expect; confirm the three actual speeds on a test burst.
- "This sensor is not ISO-invariant" — true of the 1200D generation as a class; not verified with a lab test on your specific body.
- The 1200D has no C1/C2/C3 custom mode banks — high confidence, but confirm in the manual if you were counting on them.
Times are computed for your chosen spot from NASA/GSFC Besselian elements (ΔT = 69.2 s) — change the spot in the Live tab and this card follows. Gear-buying and build deadlines live in the .
The one-page brief
Totality crosses northern Spain 12 August 2026, ~20:26–20:31 CEST. In the Castilla y León focus area the Sun will be very low — about 10–11° up toward León, 7–9° over the plains, ~7° in Soria — sinking toward the west-north-west (azimuth ≈ 281–284°). Sunset follows totality by roughly 50–60 minutes. Your #1 job: a completely unobstructed horizon toward WNW, and a sky that is clear at that horizon.
Everything below is grouped by topic — tap a section to open it. The living, spot-specific schedule (contact times, filter alarms, gear deadlines) is in the .
🔗This app — where it lives, and how to keep itThe address, installing it on the phone, sharing your setup, and what to do if something breaks
Eclipse Assistant lives at e-clipse.pages.dev. It is a static page — no account, no server holding your data — which is why everything below is about the device in your hand.
Put it on the Home Screen before you travel
- iPhone (Safari): open the site → Share → Add to Home Screen. Android (Chrome): ⋮ → Add to Home screen.
- Do it after choosing your spot: the address bar carries your spot and tapped points, so the icon you save restores the exact setup — even if the phone later clears the site's storage.
- Open it once online before you leave. That first visit is what stores the whole app, the 149 sites, the 4,392 Castilla y León villages, the relief and the climatology on the device.
- Then test it in airplane mode. Everything except live forecasts and new map tiles must still work. If it does not, it will not work in a field either.
Carrying your setup to another device
There is no cloud sync, and that is deliberate — nothing to log into at 19:00 with one bar of signal. The link is the sync: the address always contains your chosen spot and your tapped points. Use 🔗 sync in the Live tab (or just copy the address) and open it on the other phone or laptop; it lands on the same spot, with the same points, and works offline there after one load.
What still works with no signal
| Always | Every contact time, the timeline, the photo card, the horizon simulator, the 149 sites, the villages, the climatology, the relief — all computed on your device. |
|---|---|
| Last stored answer | The cloud matrix, hourly decks, pressure chart and ranking, each stamped with when it was fetched. |
| Needs signal | Refreshing forecasts, live radar/satellite, and map tiles you have not looked at before. |
If something looks wrong
- The app seems out of date. It updates itself when a new version is published, but the very first time you may need one force-quit and reopen.
- The map is blank. The app now tries all six basemaps in turn and tells you which it switched to. If they all fail, that is your connection: run Live → ⌗ Run diagnostics, which tests each tile host one by one and tells you which are reachable. Everything else keeps working — times and the timeline never needed the network.
- Tiles look stale or broken. The map's message offers Retry and Clear tile cache.
- Pull map tiles before you go. The Live tab's offline section downloads the tiles for a region so the map is there without signal — the one thing a first visit does not store for you.
Source and issues: github.com/Mat-Cn/Eclipse.
⏱Key times & alarmsC1 → C4 in CEST, and the two alarms that protect your eyes and gear
| ≈ 19:29–19:36 | First contact (C1) — partial phase begins |
|---|---|
| ≈ 20:27–20:31 | TOTALITY (C2→C3) — up to 1 m 48 s on the centre line (Melgar de Fernamental); 1 m 40 s+ across most of the León–Palencia–Burgos–Soria plains |
| ≈ 21:12–21:25 | Partial ends (C4) — in the far east the Sun sets still partially eclipsed |
- ⏰ Alarm #1 — C2 −15 s: FILTERS OFF. Camera filter comes off for totality; glasses stay on until the last Baily's bead dies.
- ⏰ Alarm #2 — C3 −10 s: FILTERS BACK ON. This is the one that damages eyes and sensors if you miss it. Set both alarms on your phone the moment you're parked.
Exact contact times shift by a few minutes across the band. The computes them (to the second) for your chosen spot, including the two alarm times ready to copy into your phone.
☁️Weather — maps, models & pressureOne-tap cloud tools · where the good odds are · which model to trust when
Cloud & eclipse tools — one tap
Where the weather is (ERA5, eclipse hour, 26 yrs)
- Ebro valley (Zaragoza plain): 28–31%, ~60% clear. Sun ~5–6°. Flat, huge road network, easy repositioning along A-2/A-68/AP-68.
- South Meseta edge of the track (Guadalajara, E Madrid): ~30–33%, sun ~6–7°, short-ish totality (near southern limit) — trade duration for sky.
- North Meseta (León–Palencia–Burgos–Valladolid): 32–38%, ~50–55% clear, longest durations (100–110 s) and the highest Sun (8–11°).
- Asturias & Cantabrian coast: 50–66% cloud — the longest totality on paper, the worst odds in practice. Only go with a locked-in clear forecast.
Which weather model to trust (and how to read it)
Weather apps blend many models; on eclipse day the differences matter. The number after each name is the grid size — smaller grid = sees smaller clouds, but only close to the event:
| AROME-HD 1.3 km AROME 2.5 km | The day-of king (≤ 2 days out). The only models fine enough to place valley fog, coastal stratus sneaking under the Cantabrians, and single convective cells over the mountains. Available in this app's Live tab and on meteoblue/Windy. |
|---|---|
| ECMWF 9 km | The planning king (2–10 days out). Statistically the best global model — use it to pick your region and watch how its eclipse-hour cloud map shifts run to run. If ECMWF and AROME agree on the day, trust them. |
| ICON-EU 7 km / ICON 13 km / NEMS 4 km / ALADIN 2.3 km | Good second opinions. When they all agree with ECMWF, confidence is high; when they scatter, the pattern is unstable — favour mobility over a "perfect" spot. |
| GFS 22 km | Coarse — use only for the long-range trend (it reaches 16 days first). Never trust GFS alone for cloud detail. |
- Read layers, not the total %: low cloud kills the view; thin high cirrus often doesn't (but smears the low sun). The Map tab's forecast overlay draws all three decks at once along the totality band — low = red, mid = green, high = violet. A single deck paints as continuous cloud; where decks overlap, thin vertical bars alternate their colours. It follows the model selector: pick ECMWF to see what Windy shows, ICON-EU for Ventusky, AROME-HD within ~2 days. Zoom in — or tap a spot to draw its sight-line — and the overlay re-samples just that window at up to ~3 km (dashed outline): on "Best match" the window automatically tries AROME-HD 1.3 km and falls back until its ~2-day horizon reaches the date, so the map sharpens itself as eclipse day approaches.
- Read the right hour: 20:00–21:00 CEST. A "sunny day" forecast is irrelevant if a sea of stratus rolls in at 20:15.
- Read the trend: a spot that improves in each successive run beats a slightly better spot that worsens.
- On the day, do it in the Live tab: the dataset matrix asks every model at once for your spot and every tapped point (the ± column is the spread), and the ☁→☀ column weighs the layers where your sight-line actually crosses them.
Pressure: why "high" is your friend
Yes — surface pressure matters a lot. A high-pressure area (anticyclone, ≳ 1020 hPa) means slowly sinking air, which suppresses convection, dries the mid-levels and typically delivers cloudless evenings — exactly what you want at 20:30. Watch for the Azores High extending a ridge over Iberia in the days before: that is the classic clear-eclipse setup.
How to read the number
| ≥ 1022 hPa | Strong high — subsiding air kills convection and dries the mid-levels. The clear-sky machine; commit with confidence. |
|---|---|
| 1015–1021 | Weak high / neutral — usually fine on the August Meseta, but check the cloud decks; a weak ridge can still let cirrus through. |
| 1008–1014 | Flat — the absolute number tells you little here. Read the trend and the model cloud maps instead. |
| < 1008 hPa | Low — unstable airmass: fronts, spreading cloud shields, evening storms. Favour mobility and the driest region on the track. |
- Trend beats snapshot. Pressure rising day over day = a ridge building = improving odds, even if today's cloud map looks mediocre. Falling = a trough approaching = deteriorating, even under a currently blue sky. The Live tab shows the same-day trend since 14 h next to the 20 h value.
- Highs and lows steer the clouds: air flows clockwise around an H and counter-clockwise around an L — so with an H east of you and an L west, expect southerly flow dragging cloud north. Where your spot sits relative to the centres matters as much as the number.
- Two Iberian caveats: (1) a summer high can still trap low marine stratus ("nortada") against the north coast — another reason to stay on the southern plains; (2) the hot interior breeds a shallow thermal low (~1010 hPa) on August afternoons — that one is heat-driven and mostly harmless, but it can spark isolated evening storms over the mountains.
- Where to see it in this app: Live tab — the hPa column in the dataset matrix (every point, selected model), the pressure + trend line under the hourly decks, and the pressure-over-time curve (3 days back + the full forecast horizon — the line to watch daily); Map tab — the ◯ hPa overlay paints the field hour by hour (scrub/▶ 17→23 h, ecl ±3 h) with the H/L centres marked and slope arrows pointing up-gradient (Low→High; longer/oranger = tighter isobars = windier — and wind blows roughly across them, clockwise around the High); the ◔ hPa trend overlay colours where pressure is rising ▲ (low→high, improving) vs falling ▼ (high→low) over that window; and isobar lines (the TV-chart look) are drawn automatically on top of the ☁ clouds overlay: tightly packed lines = windy, cloud fields moving fast; wide spacing = calm. On eclipse day the Live pressure curve re-fetches itself hourly.
- Best external cross-check for the field: Windy's pressure layer ⚑ — flagged: it is Windy again, but its ECMWF isobar view is the most reliable interactive picture of the pressure field over Iberia (H/L centres, the Azores ridge, isobar spacing). For the professional hand-drawn synoptic analysis, the Met Office surface pressure charts are the reference.
📍Picking the exact spotLow-Sun rules: horizon math, facing WNW, elevation, haze
- Horizon math: at Sun altitude h, an obstacle of height H blocks it if closer than H / tan(h). At 6° a 20 m tree line must be ≥ 190 m away; a 200 m ridge ≥ 1.9 km away. At 3° double all that.
- Face WNW (az ≈ 285°). Choose east rims of wide valleys, west-facing slopes, mesa edges, reservoirs' east shores, or the sea on a west coast.
- Get modest elevation over flat land — 50–200 m above the plain lifts you over haze and heat shimmer near the horizon. Avoid summits above ~1,400 m: August afternoons breed convective cloud on the peaks themselves.
- Avoid north-facing slopes, forest clearings, deep valleys, and city skylines to your west.
- Dust/haze: the low Sun shines through a long slab of atmosphere — a dry, haze-free airmass matters almost as much as cloud. Check aerosol/Calima forecasts on the day.
- Scout at the right time: the map's ⛰ horizon simulator and the ⭐ verdict view flag terrain blocks — but nothing beats standing on the spot at ~20:30 on any prior evening: same Sun altitude, same direction, same haze.
- Read the height directly: switch the basemap (▤, top-right) to Elevation (m). It decodes the DEM in the browser and paints height itself — dark = low, white = high, a contour every 100 m, shaded from the WNW, the direction the Sun will actually set from. It is deliberately grey: every other colour on this map already means cloud or pressure. Use it to answer the only question the ground has to answer tonight — am I above the land to my west?
- Contours and exact metres, outside the app: topographic-map.com — Castilla y León. A full hypsometric map with a readable altitude scale and a click-anywhere elevation readout — better than any basemap here for checking whether a specific field sits on a rise or in a hollow, and for spotting the low ground to the west that keeps your sightline open.
- Second opinion on the terrain: Umbra (Lluís Gassó) scores real viewpoints across the whole Spanish track — including the Balearics — on the same three things that decide a low-Sun eclipse: Sun altitude, horizon clearance ray-cast from the terrain, and totality duration, with ERA5 August climatology as an optional shade. Its dots come from OpenStreetMap peaks and viewpoints, so it finds places this app's 149-site list does not. Different method, same physics: where the two agree, the spot is good.
📷Photography — shooting a 9° eclipseGeneral technique + the full Canon 1200D & iPhone 16 field guide
The low sun is a gift: at ~9° altitude you can frame the eclipsed Sun WITH the landscape — silhouettes, ridgelines, human figures — instead of a dot in an empty sky.
- Exposure bible: use Xavier Jubier's exposure calculator — pick aperture/ISO, read off shutter speeds for every phase (partials with filter, diamond ring, chromosphere, corona 0.1–8 solar radii, earthshine).
- Partial phases: certified solar filter over the lens. Remove it only seconds before C2 — and put it back at C3.
- Totality: no filter. Bracket wildly — corona brightness spans 10+ stops. Automate if you can; you'll want your eyes on the sky.
- Focal lengths: 500–1000 mm for the corona close-up (tripod + no shake), 24–70 mm for the money shot here: eclipsed Sun + glowing horizon + landscape. A second body/phone on a wide tripod shot, started before C2, is the low-effort winner.
- Timer app: Solar Eclipse Timer (iOS) talks you through every contact for your exact GPS spot — "filters off", "diamond ring", "filters on" — so you never miss a phase while shooting.
- Practice the full sequence once at 20:28 CEST on any clear evening: same sun altitude, same light.
Field guide — Canon EOS 1200D + EF-S 18-55 IS II & iPhone 16
Checked and folded in from the kit-specific field guide. Assumed lens: EF-S 18-55 mm f/3.5-5.6 IS II, 58 mm filter thread — verify by reading the ring on the front of the lens. Exposure values are starting points from standard practice, not measured with your filter; nothing here substitutes for a test on the real Sun days before. Dated deadlines (buy → build → test) are wired into the .
🎯 What this kit can and cannot do
| Frame-filling solar disc | No. At 55 mm on APS-C the Sun is ~120 px across on the 18 MP sensor (a 0.53° subject in a 22.9° horizontal field). |
|---|---|
| Wide-field corona + landscape + horizon glow | Yes. Corona out to 2–3 solar radii spans roughly 350–500 px. This is the shot worth planning for — and the low Spanish Sun makes it better than at most eclipses. |
| Filtered partial-phase timelapse | Yes. Easiest win in the whole plan. |
| iPhone corona still | No. The base iPhone 16 has main + ultrawide only, no telephoto. 24 mm-equivalent produces a white dot. |
| iPhone wide video of the light collapse | Yes. Arguably the most rewatched footage anyone gets. |
Plan accordingly: the DSLR shoots the wide-field scene, the phone shoots video, and you spend the majority of totality looking at it with your own eyes.
⚠️ Filter safety — read before anything else
- Full-aperture filter on the front of the optic. Never behind, never at the eyepiece.
- Never use the 1200D optical viewfinder with the Sun in frame, filtered or not. Live View only.
- Filter comes off only during totality. Back on the instant the diamond ring reappears. Partial phases are exactly as dangerous as an ordinary day.
- Eclipse glasses ≠ camera filter. You need both; they are different products.
- Never OD 3.8 ("FOTO" grade) for anything your eye might meet: it transmits ~16× more light than OD 5.0 and is photographic-only. Buy OD 5.0 — or Baader's newer ClearWhite OD 5.1, its direct successor (both meet ISO 12312-2).
- Do not use: stacked ND filters, welding glass below shade 14, exposed film, smoked glass, CDs, sunglasses, polarisers. Standard ND does not block infrared.
🛒 Shopping list (Geneva area)
| Baader AstroSolar film, A4 (20×29 cm), OD 5.0 ×2 | Optique Perret, Geneva (~30 min from Annemasse) — phone ahead to confirm physical stock. CHF 30/sheet is at or below the EU benchmark (EUR 29 FR – 36 DE): fair, unusually good for Swiss retail. |
|---|---|
| Eclipse glasses (ISO 12312-2) ×1–2 | Same shop — they stock Bresser. |
| Wired remote — Canon RS-60E3 or clone | Any photo shop, ~EUR 8–25. 2.5 mm sub-mini jack (fits the 1200D). |
| Tripods ×2 | Own / borrow — one per camera. |
| Black card, craft knife, matte black tape, adhesive foam strip | Any DIY shop — for the filter cells. |
- Buy two sheets. The second is insurance against the pinhole you find during inspection — CHF 30 buys more risk reduction than any price-shopping, and you cannot reorder in the last week.
- Fallbacks if Perret is out: Optique Unterlinden (Colmar, ~EUR 29, usually in stock) → Astroshop.de / Teleskop-Express.de (3–5 days to France). Pierro-Astro's small ECO sheet is cheapest (~EUR 15.50) but shows hors stock, 1–4 semaines — too slow to rely on.
🔧 Building the filter cells (Canon + iPhone)
One A4 sheet (20×29 cm = 580 cm²) covers both builds several times over. Cut generously. Baader's own mounting instructions ship with the film — follow theirs over these where they differ.
Universal rules
- Film mounts loose, slightly wrinkled — never drum-tight. Tension distorts the optical wavefront; a visible ripple is correct and does not degrade the image.
- Handle by the edges only. Fingerprints and creases in the coating are defects.
- Friction fit over the outside of the optic, never a screw thread — a slip-on cell cannot unscrew itself mid-shot.
- Tape everything matte black on the inside to kill internal reflections.
- Inspect against a bright lamp before every use. Any pinhole, scratch or crease that transmits light → discard that piece and cut a fresh one.
Canon lens cell (58 mm thread, ~60 mm outer barrel)
- Cut two card rings: outer Ø 80 mm, inner aperture Ø 62 mm. The aperture must clear the full front element with a few mm margin — you are not stopping the lens down.
- Cut a film square ~85×85 mm.
- Sandwich: ring — film (loose) — ring. Tape the perimeter. Do not stretch.
- Cut a card strip ~25 mm wide, long enough to wrap the lens barrel; line the inside with foam. Roll to a snug slip fit — firm enough not to fall off pointing down, loose enough to pull off in one motion.
- Tape the collar to the ring sandwich.
- Safety tether: a loop of string from the cell to the tripod head — if it drops during totality it must not roll away in the dark.
- Test: point at a bare lightbulb and check Live View — you should see the filament and near-black surround. Stray light around the collar → add foam.
iPhone 16 plate
- The phone build is a plate covering the entire rear camera island, not one lens. Measure your actual island with a ruler — both lenses and the flash — not a spec sheet.
- Cut a card plate ~20 mm larger than the island on every side; cut one single window spanning both main and ultrawide with a few mm margin — one window, no alignment risk.
- Mount the film loose behind the window, same sandwich method.
- Attach with removable tape or an elastic band around the body. Nothing adhesive touches the lens glass.
- Why cover both lenses: the phone switches to the ultrawide automatically at certain zoom levels — an uncovered ultrawide means an unfiltered Sun, blown exposure, wrecked shot.
Reality check, kept: even filtered, the partial Sun at 24 mm-equivalent is a tiny bright dot — roughly 1% of the frame width. The plate buys you the crescent in a scene, not a disk shot, and the DSLR owns the disk. The phone's main job is still the unfiltered wide video of the landscape between C2−2 and C3, aimed away from the Sun. Both uses are on the field card: partials preset (foil on, Halide ISO 100 · 1/8000) and totality preset (foil off, Blackmagic rolling).
⚙️ Canon 1200D settings & exposures
Setup (at home, not in the field)
- Mode M, format RAW — you need the latitude for the corona.
- Manual focus. AF fails on a filtered Sun. Focus at 10× Live View on a sunspot or the lunar limb, then tape the focus ring — the 18-55's infinity stop is not reliable.
- White balance Daylight (~5200–5500 K), manual. Auto WB drifts badly as the light collapses.
- IS off on the tripod. Wired remote or 2 s self-timer.
- The 1200D has no built-in intervalometer, and AEB is limited to ±2 EV / 3 frames — insufficient for totality. Bracket manually by turning the shutter dial.
- Drift: at 88 mm-equivalent the Sun moves ~1 px/second on this sensor. Exposures up to ~2 s trail negligibly on a static tripod — you are not shutter-limited.
Partial phases — FILTER ON
f/8 · ISO 100 · 1/125 – 1/500 s as the starting bracket. Test on the real Sun days beforehand with your actual filter — transmission varies by an order of magnitude between products. Bracket ±2 EV, read the histogram, write the working value on tape stuck to the camera.
Totality — FILTER OFF
Dynamic range spans roughly 12 EV. Fix aperture and ISO, vary shutter only: f/5.6 (wide open at 55 mm) · ISO 400.
| 1/2000 – 1/500 | Diamond ring, chromosphere, prominences |
|---|---|
| 1/250 – 1/60 | Inner corona |
| 1/30 – 1/8 | Mid corona |
| 1/4 – 1 s | Outer corona, earthshine on the lunar disc |
- Add ~1 EV when the Sun is below ~10° altitude (atmospheric extinction) — in Spain it will be, everywhere on the track.
- Rehearse the dial sequence blind. Spanish totality is ~1 m 40–48 s (the eclipse's global maximum, ~2 m 18 s, happens near Iceland — not here).
📱 iPhone 16 plan
- Second tripod, locked off, wide, started before C2, then left completely alone.
- Lock AE/AF before totality (long-press on screen). Otherwise the phone auto-brightens and destroys the darkness you're trying to record.
- Point at the landscape and horizon, not the Sun: the 360° twilight ring, the light collapse, the crowd reaction, shadow bands on a white sheet.
- Filter on only for a partial-phase novelty frame; off for totality video.
- Never digital-zoom; grab stills during totality only if it doesn't cost you the view.
✅ Pre-eclipse checklist (tap to tick — it remembers)
Now (by 26 Jul)
Within 48 h of the film arriving
Two weeks out (by 1 Aug)
Day before (11 Aug)
On the day
💥 Known failure modes
| Filter falls off mid-totality | Snug foam collar + string tether to the tripod |
|---|---|
| Forgot to remove filter at C2 | Alarm set to C2 −15 s |
| Forgot to replace filter at C3 | Alarm set to C3 −10 s. This one damages equipment and eyes. |
| Focus drifted | Tape the ring after setting. Verify at 100% on a test frame, not on the rear screen. |
| Phone auto-brightened the darkness away | AE/AF lock before C2 |
| Lost the whole event to fumbling | Rehearse blind. Accept fewer frames. |
| Battery died | Spares in a warm pocket |
| Horizon obstruction at low Sun | Site scout at the correct time of day |
🚗Getting there & moving on the dayRoutes north from Salamanca onto the track, day-of driving, and the post-eclipse jam
You are leaving from Salamanca — which sits at 99.7% partial, just outside the southern limit, so you must drive north onto the track. The good news: from here the northern Meseta — the best-odds and longest-duration zone — is your nearest totality, 1–2.5 h up the A-62 "Autovía de Castilla", which threads Salamanca → Tordesillas → Valladolid → Palencia → Burgos, running straight into the path.
Getting to the track (drive times from Salamanca)
- Palencia — closest full-value totality: A-62 via Valladolid, ~150 km / ~1h45, ~102 s, Sun ~8.6°. Valladolid is nearer (~110 km / ~1h10) but sits near the southern edge (~87 s) — fine as a bad-weather fallback, but push on to Palencia for the real thing.
- León — highest Sun, longest: A-66 "Ruta de la Plata" via Zamora → Benavente → León, ~180 km / ~2 h, ~105 s, Sun ~9.6° — the best altitude anywhere on the track.
- Melgar de Fernamental — the centre line / max duration: A-62 to Palencia then north-east toward Melgar, ~200 km / ~2h15, ~106 s. Burgos is a little further on the same A-62 (~225 km / ~2.5 h, ~104 s).
- Aranda de Duero / Duero valley (eastern CyL): A-62 to Valladolid then A-11 "Autovía del Duero" east, ~180 km / ~2 h, ~102 s — the drier, more continental side if the western Meseta clouds over.
- Ebro valley (driest climatology, but far): ~500 km / ~5 h east (A-62 to Burgos, then A-1). Only worth it if the whole Meseta forecast collapses and the Ebro is locked clear — decide the night before, it is a long haul.
- Skip the Cantabrian coast from here: over the mountains and the worst cloud odds on the track — only with a locked-in clear forecast.
- Let the forecast pick the field. Salamanca as a base keeps the entire western + central track (León ↔ Burgos) reachable in a morning's drive, so you don't have to commit to a town until the day.
On eclipse day (out of Salamanca)
- Full tank in Salamanca the evening before — stations near the line will queue. Water (2 L+/person), food, hat, sunscreen: August Meseta, 35 °C+ in the afternoon; bring a warm layer for after sunset.
- Leave early — be on the track by T−3 h (≈ 17:30). Your spine is the A-62; the A-231 León–Burgos corridor hugs the centre line and will be the busiest road in Spain that evening.
- Download offline maps (and screenshot your Timeline): mobile networks near the centre line may collapse under the crowd.
- Never watch from the road shoulder — illegal, dangerous, and where eclipse-day accidents happen. Park in a field entrance, village, or mirador; park facing your exit.
- Check DGT live traffic before every leg — it's linked from each Timeline step too.
- Plan the drive back south to Salamanca after dark. The 2017/2024 US eclipses jammed roads for hours the instant totality ended. Watch the partial to sunset, eat where you are, then drive home late.
👀What to look for during the eclipseSafety rules, and the show minute by minute — before, during, after totality
Safety, always
- Eclipse glasses (ISO 12312-2) for every partial phase; naked eyes and unfiltered optics only between C2 and C3. Binoculars on the corona are stunning — but only inside totality.
The last hour (partial phase)
- Crescent projections: gaps in foliage, a colander, or crossed fingers project hundreds of little crescent Suns that thin as C2 approaches.
- The light goes wrong: shadows sharpen along one axis, daylight turns silvery-metallic, temperature drops noticeably, and a faint "eclipse wind" often picks up.
- Venus (mag −4) pops out ~15–20 min before totality, ~46° upper-left of the Sun. Animals hush; swallows go to roost.
- The umbra wall: in the last minutes the WNW horizon visibly darkens — that is the Moon's shadow arriving at ~2–3 km/s. From any elevated spot you can watch it race at you.
- Shadow bands: in the final 1–2 minutes faint rippling grey bands may slither over light ground and walls — rare and subtle; spread a white sheet to catch them.
Totality (~1 m 40 s)
- Diamond ring → Baily's beads as the last sliver breaks through the lunar valleys — glasses off as the last bead dies.
- The corona: 2026 comes just after solar maximum, so expect a full, petal-rich corona with streamers all around the disc, plus the red chromosphere rim and, very likely, prominences.
- Planets appear: Venus blazing high upper-left, Jupiter ~10° lower-right of the Sun, Mercury ~15° right and very low, the star Regulus ~10° upper-left. The ⛰ horizon simulator draws them all in place for your spot.
- Look down and around too: a 360° sunset ring on every horizon (this low Sun makes it exceptional), eerie landscape colours, the crowd. Budget your ~100 seconds: corona → landscape → corona.
After C3
- Second diamond ring (filters straight back on), a second chance at shadow bands, then the whole partial sequence in reverse — with the Sun sinking toward a normal sunset ~50–60 min later, still crescent-scarred in the east of the track.
🧭Day-of playbookThe four decisions, T−48 h to T−90 min
- T−48 h: compare every model in the Live tab's dataset matrix. Pick a primary region and a fallback ≥150 km along the track.
- Morning of: re-run the best-places ranking in the Live tab. Watch real clouds on the map's live satellite overlay (updates ~every 10 min).
- T−3 h: commit. Convective cloud builds over mountains in the afternoon — if in doubt, move toward the flat, dry side (from Salamanca: the Duero valley toward Aranda, or the Ebro basin as a far fallback — not the Cantabrian mountains).
- T−90 min: be parked. Set up and focus during the partial phase with time to spare; set the two filter alarms. Totality waits for no one.
The full dated version — with computed times for your spot — is the .
🔗Sources & further toolsOfficial maps, climatology, exposure calculators, live satellites
- Eclipse Assistant — this app · source — see for installing it, sharing your setup and what works offline
- topographic-map.com — Castilla y León altitude map — hypsometric tint, altitude scale, click for the elevation at any point
- Umbra (Lluís Gassó) — viewpoint scoring across the whole Spanish track — sun altitude + ray-cast horizon clearance + duration per OSM viewpoint, ERA5 climatology optional, EN/ES/CA
- Eclipsophile — climatology deep dive (J. Anderson)
- NASA GSFC — interactive path map & Besselian elements
- ESA — official map of totality in Spain
- TheSkyLive — per-town circumstances
- Tres Eclipses (ES) — official Spanish eclipse app
- timeanddate — city times
- Junta de CyL — official observation points
- Eclipse Soria — province guide (Starlight skies)
- Hosteltur — best miradores round-up
- eclipse-solar.es — location guide
- WeatherAway — eclipse cloud forecast with low-Sun cloud layers (multi-model)
- X. Jubier — solar eclipse exposure calculator
- Solar Eclipse Timer — voice countdown app for contacts (iOS)
- Windy — visual cloud forecast maps
- Ventusky — animated cloud-cover maps (total + low/mid/high, ICON/GFS)
- AEMET — official Spanish forecasts
- EUMETSAT — live satellite of W Europe
Data: eclipse geometry computed from NASA/GSFC Besselian elements (F. Espenak), ΔT = 69.2 s; verified against the NASA path tables to <0.1 s in duration. Cloud climatology: ERA5 reanalysis via the Open-Meteo archive API, mean total cloud at 18–19 UT on 10–14 Aug 2000–2025. Live forecasts: Open-Meteo (ECMWF/GFS/ICON). Live satellite & radar overlays: RainViewer. Elevations: Copernicus DEM via Open-Meteo. This tool is for planning; always sanity-check against official sources on the day.