Planning for the 2 August 2027 eclipse

The August 2nd, 02027 eclipse will sweep from Gibraltar across North Africa, passing over Luxor, and onward to Arabia. Near Luxor, Egypt, totality will last for ~6 minutes 23 seconds. This will be the longest total eclipse over land until 2114. The Pyramids of Giza are not in the path of totality.

I've made a planning tool for use in figuring out my own trip. It's got the eclipse path, lines of equal totality, and cloud climatology. Clicking the map shows eclipse stats for that spot: the beginning and end of totality, the Sun's position, the spot's elevation, whether hills hide the Sun, and cloud cover for that spot over previous years. Airports within a six-hour drive of the path are marked.

Major cities

The list below contains every city of 100,000 or more in the path, plus a few points of interest on and off the path. Times are local and totality calculations account for elevation. "Any cloud" is the share of July and August days with any cloud there, thin or thick. "Thick cloud" is the average share of the sky covered by low and medium cloud at the moment of eclipse over the two weeks around 2 August since 1991, "Clearest–cloudiest year" is the range of thick cloud cover from year to year, and "Clear enough" is the share of those days with thick cloud under 30%. "Daily high" is the average high temperature around 2 August. Click a column to sort, or a row to find the place on the map.

Airports within a 6-hour drive

Every airport with scheduled flights from which you can drive into the path in six hours or less, as measured by OpenStreetMap roads. Drive times are for an empty road and may not be entirely accurate: OpenStreetMap's speed data can be sparse and is sometimes imputed by road size. Real-time traffic data isn't likely to have informed the estimates. Drives are constrained to stay inside one country, since several borders along the path are closed (Morocco and Algeria, Libya and Egypt, Saudi Arabia and Yemen). "Best totality" is the longest totality among points within six hours' drive. Click a row to see the airport and that point on the map.

What else to plan for

The map handles the geometry and the climatology. Other things to think about:

• The Sun's altitude. In Spain and Morocco the eclipse will occur mid-morning with the Sun about 38° up in the east, so a hill or a building can block it. The map's red shading marks where the terrain itself hides the Sun during totality: west-facing slopes and valleys in the Rif mountains of northern Morocco, the Tell Atlas in Algeria, and a few other places. Note that the elevation data is relatively coarse, so you'll need to judge local topography yourself. Around Luxor the eclipse occurs in the early afternoon and the Sun is 80° up, nearly overhead. Bring a reclining chair, and point cameras straight up. Clicking the map gives the altitude, the compass direction, and the elevation.
• Heat. The temperature chart and the table's "Daily high" column show the typical high around 2 August. Upper Egypt and inland Arabia run past 104 °F (40 °C), and around Luxor the eclipse comes in the early afternoon, close to the hottest part of the day. Make plans to have shade and water.
• Mountain tops. A summit gives a clear horizon all round, and often clearer air above the haze. The green triangles on the map ("Accessible peaks" in the layer menu) are about 2,000 named summits in the eclipse's path that stand at least 300m above their surroundings and have a road within 500m. Click a mountain for more details. Some of these high points will be private, military, or crowned with antennas - check before you make plans.
• Year-to-year luck. The average values for cloud cover hide a lot. Click any spot for a chart of each year's thick cloud at eclipse time since 1991. In some places one year in five is far cloudier than the rest.
• Dust and haze. A clear sky in the desert can still be hazy with dust. The cloud data don't count dust, and the Sahara and Arabia are dustiest in summer. Coastal Spain and Morocco can also get early-morning sea fog that burns off before the eclipse.
• Distance from the centre line. Totality time falls off slowly near the centre and fast near the edges, so you can go 30–40% of the way to the edge for a few seconds' loss. Near the edges of the eclipse path the lunar mountains and valleys can change the timing by a few seconds, so stay a few kilometres inside the line you're aiming for.
• Mobility on the day. Fixed sites (a hotel roof, a temple) are easy, but problematic if there are clouds. Having a car will help you dodge clouds the day of the eclipse.
• Safety and access. Libya, Yemen, and Somalia are in the path, but most governments advise against travel there. Some of Algeria's interior and border areas carry similar warnings.
• Bookings. Nile cruises, Luxor hotels, and southern Spain will sell out early.

How it works

Totality

All data comes from NASA's Besselian elements for the eclipse¹, which describe the Moon's shadow as a pair of cones whose axis moves across a plane through the Earth's centre. Local circumstances (the contact times, the length of totality, and the Sun's position) follow the method of the Explanatory Supplement to the Astronomical Almanac, as in NASA's own eclipse calculator. The page computes these in your browser when you click. The rest is computed in Python ahead of time.

To draw the path and the lines of equal totality, I take cross-sections of the path every 15 seconds. Each starts where the shadow's axis meets the Earth and runs perpendicular to the path. On each I solve for the edge of the umbra, then for the points where totality is 30 s, 1 m, 1 m 30 s, and so on. The lines of equal totality connect these points.

As a check, the computed central-line durations agree with NASA's path table² to within 0.2 seconds and the computed limits lie within 100 m of NASA's. Durations are for the height of the ground (see below) and a smooth Moon. The Moon's limb profile changes durations by a second or two, more near the edges.

Any cloud

The "Any cloud" layer and charts use the EarthEnv MODIS cloud frequencies of Wilson and Jetz³. They count how often the MODIS instruments on NASA's Terra and Aqua satellites saw cloud at each 1 km pixel between 2000 and 2014. Terra crosses in the late morning and Aqua in the early afternoon, so this samples about the time of day the eclipse happens. I averaged the July and August means, since the eclipse falls between them, and then averaged the result over blocks of 0.05° (about 5 km). The map draws each of those cells as a solid square, and clicks and transects sample the same grid. A transect's mean cloud weights each part by its length.

This is climatology, the long-run chance of cloud, and not a forecast. A 10% cloud frequency means about one day in ten was cloudy, not that the sky is 10% covered.

Terrain and elevation

Elevation comes from the Copernicus GLO-90 digital surface model⁴, at 3 arcseconds (about 90 m). For every pixel in the path I walked 12 km toward where the Sun stands at that place's own mid-totality (not one moment for the whole path; the Sun's position is computed for each 0.05° block, about 5.5 km, at its height) and checked whether any terrain rises above the line of sight, allowing for the Earth's curvature and standard refraction. The map shades each 0.025° cell (about 2.5 km) by the share of its pixels where the Sun is hidden. It's a surface model, so tall buildings and dense trees count as terrain, but features smaller than about 90 m (a single building, a wall) don't show up. The Sun is at least 37° up everywhere on land in the path, so only steep terrain can hide it.

Every length of totality on the page is computed for the height of the ground there: the path's edges and the lines of equal totality, the centre-line charts, the city and airport tables, transects, and clicks (peaks use the summit's own height). The height comes from the same 2.5 km cells, blended between their centres so totality doesn't jump at a cell's edge; over the sea it's sea level. Height shifts the path toward the Sun by about the height divided by the tangent of the Sun's altitude: about 1.3 km per 1,000 m in Morocco, where the Sun is 38° up, and much less near Luxor, where it's nearly overhead. A 2.5 km cell smooths over valleys and summits, so on a steep slope the true height can differ by a few hundred metres, which changes totality near the edges by a second or so.

Thick cloud at eclipse time

Not all cloud matters equally. The MODIS frequencies count any cloud, thin or thick, so a morning of high, thin cirrus counts the same as an overcast one. Thin cirrus dims the corona but still shows it. Low cloud and mid-level cloud are what hide totality. Scattered cumulus is a gamble: in the summer heat it often thins out as the land cools during the partial phase.

To calcualte thick cloud cover I used the ERA5 reanalysis⁵ from the Copernicus Climate Change Service, which gives hourly cloud cover in three layers (low, medium, and high) on a 0.25° grid (about 25 km), from 1991 to 2025. For each grid point I took the cloud at that point's own moment of mid-eclipse, interpolating between hours, on each of the 15 days from 26 July to 9 August, every year. "Thick cloud" is low and medium cloud together, combined as if they overlapped at random but never more than ERA5's own total cloud cover (ERA5 assumes layers partly stack, so the random combination can exceed its total). Its average is the chance that thick cloud is in front of the Sun at a random moment, which is the number that matters for totality. A day is "clear enough" when thick cloud covers less than 30% of the sky. High cloud is reported on its own. A spot's values blend the four grid points around it, weighted by how close it is to each.

The "Thick cloud at eclipse time" map layer (in the layer menu) shows the average, a click shows each year's, and transects show it as a dashed line. This is a model of the atmosphere constrained by observations, not a direct measurement, and at 0.25° it smooths over mountains and coasts, so the "Any cloud" layer is the better guide to the fine detail and ERA5 to what kind of cloud it is and how it varies from year to year.

Also note that in Spain and Morocco the Sun is about 38° up, so you look at it through more of the sky. Therefore, any given cover of scattered cloud is more likely to be in the way than at Luxor where the sun is almost directly overhead.

Two more layers show how much the cloud varies, in violet and magenta so they don't read as more cloud. "Cloud: year-to-year swing" is the standard deviation of a place's 35 yearly means of thick cloud at eclipse time, in percentage points: a place averaging 15% with a swing of ±3 is dependable, and one with ±12 has some very cloudy years. "Cloud: chance of a bad year" is the share of those years in which fewer than half the days around 2 August were clear enough. It mostly follows the average cloud, but it picks out the edges, such as the Sahel's northern margin, the Atlantic coast, and the Atlas and Arabian highlands, where usually good weather sometimes fails for a whole season. A click on the map gives both numbers for that spot.

Temperature

Temperatures are NASA POWER's daily maximum 2 m air temperature⁶, from NASA's MERRA-2 reanalysis, from 2001 to 2025, on a 1° grid. For each cell and year I averaged the 15 days from 26 July to 9 August. A spot's values blend the four cells whose centres surround it, weighted by how close it is to each. Reanalysis smooths over local effects: a valley floor or a city will be hotter than its 1° cell's average, and a coast or a mountain cooler.

Peaks

The peaks are the named peaks, mountains, and hills in GeoNames⁷ inside the path. GeoNames positions are rough, so I moved each to the highest point of the Copernicus elevation model within 600 m. I kept summits at least 300 m higher than the lowest point within 3 km. For road access I used the OSRM routing engine's distance from each summit to the nearest road it would route a car on, and kept those within 500 m. That distance is as the crow flies, so the walk can be longer and steeper. At 90 m resolution the elevation model averages sharp summits down (it puts the top of the Rock of Gibraltar at 361 m against a true 426 m), so summit heights read low. Hand-picked viewpoints appear as purple triangles. The routing engine's map has no roads in Gibraltar, so the Rock (reachable by cable car) is one of these.

Airports

Airports with scheduled service come from OurAirports⁸. Drive times are from the OSRM routing engine⁹ on OpenStreetMap roads. I asked for the time from each airport to points on land inside the path, about every 25 km along it and at five places across it. I dropped points more than 5 km from a road, and drives into another country, except between Gibraltar and Spain.

Cities

Cities come from Natural Earth's populated places¹⁰. Local times on the map use Natural Earth's country outlines¹¹ and each country's UTC offset in August 2027.

References