Debunks Reality Denial Earth Curvature

Testing “You Can See Too Far for Earth Curvature”

One of the most persistent modern flat-Earth claims is built from long-distance photography: a skyline is visible from dozens of miles away, a mountain can be photographed from hundreds of kilometers away, a lighthouse remains in view beyond a claimed “curvature limit,” or a zoom lens recovers a distant ship that supposedly should be hidden. The conclusion is then presented as decisive: if the Earth had the accepted curvature, the object would be behind it; because some part of the object is visible, the Earth must be flat.

Sometimes the photographs themselves are genuine and impressive. The problem is usually the calculation. “Earth curvature” is not one universal number that can be subtracted from every line of sight. The frequently quoted “8 inches per mile squared” approximation gives the departure of a spherical surface below a tangent plane near the observer. It does not directly tell you how much of a distant target must be hidden from an elevated camera. To calculate visibility, one must know the observer elevation, target elevation, Earth radius appropriate to the geometry, exact surface distance or geodesic distance, intervening terrain, the height of the object above local ground or water, and the atmospheric refractive conditions along the ray path.

When those terms are handled correctly, ordinary long-distance visibility follows the globe model remarkably well. More importantly, the same model predicts a fact that simplistic curvature calculators often ignore: the atmosphere can bend light. Under common conditions the ray curves slightly downward, extending the optical horizon modestly. Under strong temperature inversions over water or ice, looming, superior mirages, and ducting can reveal portions of objects that are geometrically below the unrefracted horizon. NOAA explicitly warns navigators that looming can bring normally hidden objects into view, and professional geodetic surveying manuals include both curvature and refraction corrections because both are measurable.[1][2]

The central mistake: “curvature drop” and “hidden height” are not the same quantity. A target 50 miles away is not viewed from a tangent line located at sea level unless the observer’s eye is literally at sea level. Raise the camera and the horizon moves outward. Raise the target and its upper portions can remain visible after its base is hidden. Then refract the ray through real air and the geometric horizon can move farther still. Most viral “too far” claims fail because one or more of these terms is omitted.

Contents

1. What Exactly Is the Claim?

The claim usually takes one of several forms:

The photographs may be real. The distances may be approximately right. The conclusion still does not follow until the correct geometry is used.

A globe visibility problem is a line-of-sight problem. The ray from the observer to the visible point on the target either clears the Earth’s surface or it does not. If it passes through an atmosphere, the ray is generally not perfectly straight, so one must also determine whether refractive bending materially changes the path.

At minimum, a quantitative analysis requires:

  1. observer eye or lens height above the relevant reference surface;
  2. target-base elevation above that reference surface;
  3. target height;
  4. surface distance between observer and target;
  5. Earth radius or ellipsoid appropriate to the location and accuracy required;
  6. intervening topography, waves, structures, and local water level where relevant;
  7. temperature, pressure, and humidity structure along the path if refraction is significant;
  8. camera focal length, crop, sensor resolution, and image scaling if pixel heights are used.

Many online curvature arguments instead use only one number: distance. They enter it into “8 inches per mile squared,” compare the result with the full object height, and declare contradiction. That procedure answers a different geometrical question from the one being observed.

2. What Shape Does Geodesy Actually Use?

For rough horizon calculations, treating Earth as a sphere of radius about 6,371 km is usually sufficient. For precision work, professional geodesy uses an oblate ellipsoid. NASA gives an equatorial radius of 6,378.137 km and a polar radius of 6,356.752 km.[3] NOAA’s National Geodetic Survey explains that Earth’s reference shape is modeled as an ellipsoid of revolution and that more detailed height work uses the geoid, an equipotential surface related to mean sea level.[4]

This distinction is worth making because “the globe model” is sometimes caricatured as requiring every point at sea level to lie on a perfect sphere. It does not. Earth is slightly oblate, has topography, has an irregular gravity field, and has ocean surface variations associated with that field, tides, currents, weather, and dynamic oceanography.

For an observation across 20, 50, or 100 miles, however, these refinements are normally small compared with basic errors such as using the wrong observer height or ignoring refraction. A spherical approximation is often perfectly adequate for demonstrating the geometry.

The mean-radius sphere is therefore not a dogma. It is a calculation tool whose error can be estimated. When greater accuracy is needed, geodesists move to ellipsoidal and geoidal models. NOAA’s geodetic tools explicitly compute distances and azimuths using standard ellipsoids such as GRS80 and WGS84.[5]

3. The Famous “8 Inches per Mile Squared” Rule

For a sphere of radius R, the surface falls below the tangent plane at the observer. If d is a relatively short horizontal distance compared with Earth’s radius, the drop below that tangent is approximately:

drop ≈ d² / (2R)

Using Earth’s radius and expressing d in miles gives the familiar approximation of about:

drop ≈ 8 inches × d², with d in miles

This rule is useful. Surveying manuals contain the same small-angle geometry in more precise units. NOAA’s geodetic leveling manual gives curvature correction proportional to the square of the sight distance divided by twice Earth’s radius.[2]

The problem is not the formula. The problem is what people think it means.

If d = 10 miles, the tangent drop is roughly 66.7 feet. At 20 miles it is roughly 267 feet. At 50 miles it is roughly 1,667 feet. These numbers sound dramatic, which makes them rhetorically powerful. But they describe where the curved surface lies relative to a tangent line originating at the observer’s local surface point. They are not automatically the amount of a distant building that must disappear below the observer’s horizon.

The distinction becomes enormous when the observer is elevated. A camera 100 feet above the surface can see much farther than an eye at 6 inches. A target 1,000 feet tall can have its upper portion visible from far beyond the geometric horizon of its base. The tangent-drop number does not include either fact.

4. Drop Is Not Hidden Height

This is the single most important correction to most “too far” memes.

Imagine standing at the exact surface of a perfectly spherical Earth with your eye at zero height. Your horizontal line tangent to the surface immediately leaves the globe, while the ground falls away beneath it. In that special case, tangent drop and geometric obstruction are closely related.

Real observers are not at zero height.

If your eye is 6 feet above the surface, the line from your eye tangent to Earth touches the surface several miles away. Everything before that tangent point lies below your eye but is not hidden by curvature. Beyond the horizon, increasingly large lower portions of targets become hidden.

Therefore the hidden portion of a target depends on the observer’s height. A curvature calculator that gives only “drop at distance” can badly overstate how much target should be obscured.

There are several equivalent ways to perform the exact geometry. One can use Earth-centered coordinates, construct the observer tangent, calculate line-of-sight clearance, or calculate the horizon distance from each endpoint. For practical intuition, the two-horizon method is especially useful.

5. Distance to the Geometric Horizon

For an observer at height h above a spherical Earth of radius R, the exact straight-line geometric horizon relation is:

d = √[(R + h)² − R²]

When h is very small compared with R, this reduces to:

d ≈ √(2Rh)

For Earth, this gives a surface-distance horizon of roughly 3 miles for an observer whose eyes are around 6 feet above sea level, with exact values depending on the precise height and model used.

Raise the observer and the horizon expands as the square root of height. At 100 feet, it is on the order of a dozen miles. At 1,000 feet, it is several dozen miles. From an aircraft, much farther.

This simple fact immediately defeats many casual calculations. A photograph taken from a bluff, upper hotel floor, bridge, drone, observation deck, dune, or shoreline whose elevation was guessed from Google imagery may have a horizon far beyond the number assumed by the claimant.

Camera elevation must mean the actual lens center above the relevant surface, not the nominal elevation of the city, beach parking lot, or GPS waypoint. Ten or twenty feet of unnoticed vertical difference can materially change a marginal visibility prediction.

6. Observer Height Plus Target Height

If the observer is elevated and the target is elevated, their geometric horizon distances can be added.

For observer height ho and target height ht:

dmax ≈ √(2Rho) + √(2Rht)

This approximates the maximum line-of-sight separation for the top of the target in a no-refraction spherical model, assuming no terrain obstruction.

The result is intuitive. The observer can see to their horizon. The elevated top of the target has its own horizon. If the two horizon arcs overlap, the top can be mutually visible even when the target’s base is hidden.

This is why tall skylines can be seen at distances where street level cannot. It is why mountain summits can be visible across extraordinary distances while their lower slopes disappear behind Earth. It is why lighthouse lantern rooms are positioned high above the water.

Flat-Earth memes often compare the entire height of the object with the tangent drop and ignore the observer’s horizon contribution. That is a category error. The correct comparison is between the ray connecting the actual observer and the actual visible feature and the intervening Earth surface.

7. A Worked Long-Distance Example

Suppose an observer’s camera is 20 feet above the water and a distant building is 800 feet tall with its base near water level. The building is 50 miles away.

A simplistic calculation says:

8 inches × 50² ≈ 20,000 inches ≈ 1,667 feet of “curvature”

Then comes the viral conclusion: “The entire 800-foot building should be hidden by more than 800 feet, yet I photographed it.”

But this compares the building against tangent drop from a surface-level tangent. It ignores the 20-foot observer elevation.

The observer’s geometric horizon is roughly:

do ≈ √(2Rho) ≈ 5.5 miles

The 800-foot building top has a geometric horizon of roughly:

dt ≈ 34.6 miles

Added together, the top would have a no-refraction visibility range on the order of 40 miles. At 50 miles, some additional height or refractive extension would be required. But notice what happened: the relevant discrepancy is not “1,667 feet minus 800 feet.” That entire original calculation was the wrong geometry.

Now add ordinary atmospheric refraction, which generally bends near-horizontal rays downward to some degree and increases the effective optical horizon. Under stronger inversion conditions, still more of the building can be visible. If the observer was actually 40 feet high rather than 20, or the building base was elevated, or the line-of-sight distance was overestimated, the prediction shifts again.

This does not mean every claimed observation is automatically explained by refraction. It means the correct model has variables that can be measured. A serious test measures them before announcing a contradiction.

8. Seeing the Top Is Not Seeing the Base

Long-distance images are often captioned “the entire skyline is visible” when inspection shows that only upper portions are actually clear. Shorelines disappear, lower buildings merge into haze, bases are hidden, or the apparent lower boundary is a refracted image rather than the true waterline.

This matters because curvature predicts bottom-up occultation. The lowest parts become geometrically inaccessible first. Tall objects remain visible longer.

If a 1,400-foot skyscraper is visible but the first several hundred feet are missing, that is not anomalous. It is exactly the kind of geometry expected on a curved surface.

One must therefore identify a physical feature of known elevation on the target. “I can see the building” is insufficient. Which floor? Which rooftop structure? Is the visible lower edge the actual base or merely the skyline merging with haze? Is the shoreline itself visible? Is there a miraged duplicate beneath the real structure? Is the target sitting above water level on elevated ground?

A high-quality analysis labels known elevations and compares those specific points with the predicted line of sight. Without that, the image may be visually striking but quantitatively weak.

9. Zoom Cannot Restore Geometry—but It Can Restore Resolution

Another common demonstration films a distant boat apparently disappearing and then zooms in until it becomes visible again. The claim is that zoom has “brought the boat back from behind the curve,” proving the disappearance was merely perspective.

Optical zoom cannot bend a blocked light ray around an opaque geometric obstruction. If the hull is physically below the geometric horizon and no refracted ray from it reaches the camera, magnification cannot create that information.

But zoom can restore angular resolution.

A distant boat may become too small for the camera sensor, display, compression algorithm, or human eye to resolve. Contrast can also be destroyed by haze and glare. Magnification enlarges the image and may reveal features that were always optically visible but unresolved.

This distinction is easy to test. If a boat merely became too small, zoom should restore the whole silhouette. If curvature is hiding the hull, zoom may sharpen the visible mast and upper structure while the geometrically hidden lower portion remains absent. Atmospheric refraction may alter the exact boundary, but magnification alone does not alter ray geometry.

Digital zoom creates an additional trap because it may enlarge interpolation and compression artifacts rather than recover genuine optical detail. Any experiment intended to test geometric occlusion should preserve original-resolution footage, record the optical focal length, and avoid treating display scaling as new information.

10. Atmospheric Refraction Is Not an Excuse; It Is Measured Physics

Light travels at slightly different speeds through air of different density and refractive index. When the refractive index changes with height, a nearly horizontal light ray bends.

NOAA explains this explicitly for weather radar: atmospheric density changes with temperature, moisture, and pressure, causing beams to refract. Under ordinary conditions the beam curves somewhat toward Earth; under abnormal conditions it can bend less, more, or even become ducted along the surface.[6]

Visible light obeys the same underlying optics. The exact refractive index and its wavelength dependence differ from microwave radar, but the principle is the same: vertical gradients in temperature and density bend rays.

Surveying manuals cannot simply “ignore” this because millimeter-level elevation work would fail. NOAA’s geodetic leveling guidance requires attention to atmospheric conditions and includes refraction corrections. It notes that refraction varies with temperature gradients, solar heating, wind, and sight height above the ground.[2]

Therefore, when a long-distance photograph is taken through tens or hundreds of kilometers of near-horizontal atmosphere, asking about refraction is not special pleading. Failing to ask about refraction would be the physically unrealistic choice.

11. “Standard Refraction” and Effective Earth Radius

A common engineering shortcut treats a downward-curving refracted ray as though the ray were straight but Earth were slightly larger. This is called the effective-Earth-radius method.

For some optical surveying approximations, ordinary terrestrial refraction is often represented as reducing the apparent curvature by roughly one-seventh, corresponding to an effective radius around 7/6 of the geometric Earth radius. NOAA surveying guidance gives a typical refraction effect around 14 percent of curvature under certain leveling conditions.[7]

Radar engineering frequently uses another conventional value, the “4/3 Earth” approximation, for standard radio propagation. That does not mean visible light always uses 4/3 Earth or that refraction is constant. It is a modeling convention for a particular propagation regime.

The important principle is not the exact default coefficient. It is that a refracted horizon is farther away than the unrefracted geometric horizon when rays curve downward.

For a rough spherical model, one can replace R by an effective radius Reff in the horizon equations:

d ≈ √(2Reffh)

If Reff is larger than R, visibility extends farther. Under extreme superrefraction, the effective radius can become very large. Under ducting, a simple single effective-radius approximation may cease to describe the full ray path adequately.

12. Refraction Is Variable, Especially Near Water

The phrase “refraction is only a small correction” is true only under limited atmospheric conditions.

Near a water surface, especially a cold lake or sea beneath warmer air, sharp temperature inversions can develop. The refractive index can then change rapidly with height. Rays traveling only a few feet above the water may curve significantly over long distances.

NOAA’s navigation guidance explicitly says that abnormal temperature structure can produce looming, towering, superior mirages, and other phenomena. It notes that looming can cause objects below the horizon to appear in view.[1]

This is why near-water tests are simultaneously attractive and dangerous. Water provides a seemingly smooth reference surface and an unobstructed path, but it also creates exactly the thermal layering that makes near-horizontal optical propagation difficult to model.

Solar heating, nighttime cooling, seasonal water temperature, fog, humidity, evaporation, and wind can all alter the vertical refractive-index profile. A photograph taken at sunrise after a cool night may behave very differently from one taken on a sunny afternoon.

Thus it is scientifically weak to take one extraordinary long-distance photograph over water, apply a fixed “standard refraction” correction, and declare that anything beyond it disproves the globe. Extraordinary refraction is itself an observed atmospheric phenomenon.

13. Looming Can Reveal Geometrically Hidden Objects

Looming is particularly important because it directly addresses the phrase “you can see too far.”

NOAA’s U.S. Coast Pilot describes looming as an optical condition in which objects appear higher than their true elevation and states that objects below the horizon may actually be brought into view.[1]

That is almost a word-for-word description of what many “impossible visibility” posts present as evidence against curvature.

The mechanism is straightforward. Under a temperature inversion, cold dense air near the surface can sit beneath warmer less-dense air. Light rays tend to bend toward the denser layer. If the ray curves downward enough, a light path from a geometrically hidden object can reach the observer.

The observer then sees the object along the arriving ray direction and mentally extrapolates that ray backward as a straight line. The image appears elevated. Depending on the refractive profile, it may be vertically stretched or otherwise distorted.

Importantly, looming does not require a spectacular upside-down “mirage.” An object can simply appear displaced upward. Therefore the objection “the image looks normal, so it cannot be refraction” is invalid.

14. Superior Mirages Can Distort, Duplicate, Invert, and Elevate Images

Stronger or more structured inversions can produce superior mirages. These may create multiple images of the same object, including upright and inverted components. Objects can appear to float above the horizon, stretch vertically, or merge into strange bands.

NOAA’s Coast Pilot describes superior mirages as images appearing above the real object and discusses related looming and towering phenomena.[1]

This matters when analyzing skyline photographs. A refracted skyline need not preserve true vertical proportions. The lower apparent edge of a building may not correspond to its physical base. Refracted water or shoreline can appear beneath structures. Multiple ray paths can compress or stretch features.

A still image without meteorological context can therefore be deeply ambiguous. The stronger the claim—“this image violates Earth curvature by hundreds of feet”—the more important it is to determine whether the image geometry itself has been altered by atmospheric optics.

Time-lapse photography is extremely valuable here. If a distant skyline changes apparent height, shape, or completeness over minutes or hours while the observer and buildings remain stationary, the Earth has not changed radius. The atmosphere has changed its refractive structure.

15. Ducting Can Carry Rays Far Beyond the Normal Horizon

In extreme superrefraction, ray curvature can approach or exceed the curvature of Earth. The atmosphere can then form a duct in which electromagnetic energy travels unusually long distances.

NOAA documents this for weather radar. Under ducting, radar beams can bend more strongly than Earth’s curvature and strike distant ground targets, producing anomalous propagation echoes.[6]

Radio engineers have exploited tropospheric ducting for long-range communication, and similar refractive layering influences optical paths over water.

Ducting is decisive against a simplistic “maximum visible distance” concept. There is no single immutable optical horizon independent of weather. The geometric horizon is well-defined. The apparent horizon can move because light does not always follow straight Euclidean chords through a stratified atmosphere.

This does not make globe geometry unfalsifiable. It means a correct experiment either measures the refractive profile, chooses conditions and sight heights that minimize refraction, or uses a method that cancels or constrains it.

16. Professional Radar and Navigation Already Account for This

If curvature and atmospheric bending were post-hoc inventions used only to answer internet photographs, they would be absent from operational engineering. The opposite is true.

National Weather Service documentation explains that radar beams rise relative to the ground with distance because of Earth curvature and that refraction alters their path. Under standard refraction, the beam still rises because its curvature is less than Earth’s. Under superrefraction, it stays closer to the surface. Under ducting, it can bend back toward the ground.[6]

The World Meteorological Organization likewise instructs that radar-derived heights need corrections for Earth curvature and atmospheric refraction.[8]

Marine navigation manuals warn of looming and mirages because sailors can see objects displaced from their geometric positions. These are practical hazards, not philosophical arguments.

The same framework explains both successful ordinary predictions and exceptional visibility. That is what a strong model should do.

17. Surveyors Measure Both Curvature and Refraction

Geodetic leveling is one of the clearest empirical checks because the profession has to recover accurate elevations over real terrain.

NOAA’s National Geodetic Survey states that Earth curvature causes the level surface to depart from the horizontal line of sight and provides curvature corrections. It separately discusses atmospheric refraction and its dependence on temperature gradients.[2]

Surveyors minimize errors by balancing backsight and foresight distances, keeping sight lines appropriately above the ground, avoiding problematic heating conditions, and applying corrections when necessary. If Earth were flat, systematic curvature corrections over long baselines would be unnecessary or would introduce growing errors. Instead, geodetic networks close when Earth geometry and observational corrections are handled correctly.

This evidence is more powerful than a single skyline photograph because it is repeatable, instrumented, and embedded in a system of thousands of interlocking measurements.

NOAA defines geodesy as the science of accurately measuring Earth’s geometric shape, orientation in space, and gravity field, using both ground and space-based techniques.[9] The resulting coordinate systems support mapping, engineering, navigation, property surveys, infrastructure, and satellite positioning.

18. Why Reciprocal Observations Are Much Stronger Tests

One-way observations are vulnerable to refraction because the ray may bend in a way that is difficult to infer from a photograph alone. Reciprocal observations improve the experiment.

Suppose two stations of known elevation observe each other simultaneously across a long distance. Each records the apparent vertical angle to the other. With synchronized measurements and meteorological data, one can constrain curvature and refraction far better than from a single camera.

Classical geodesy used reciprocal zenith-angle observations for precisely this reason. Atmospheric refraction often affects the two directions similarly enough that simultaneous reciprocal measurements reduce its influence.

An even stronger geometric test uses three stations or markers at controlled heights. If the endpoints establish a line of sight, the apparent displacement of the midpoint can reveal convexity. That is the principle behind the better-designed Bedford Level experiments discussed below.

The methodological lesson is general: a strong curvature test should be designed to distinguish curvature from refraction, not merely hope the atmosphere behaves ideally.

19. Why “Over Water” Is Not Automatically the Cleanest Test

Water appears perfect for curvature tests because its surface is visually smooth and approximates a level gravitational equipotential. Long lakes and canals provide unobstructed sight lines.

Unfortunately, an optical ray skimming just above water is also exposed to one of the most refractively unstable environments available.

Water temperature changes more slowly than air temperature. Warm air can move over cold water, creating an inversion. Daytime solar heating can produce sharp near-surface gradients. Evaporation modifies humidity. Light rays only centimeters or meters above the water may therefore bend substantially.

The lower the sight line, the more strongly local boundary-layer conditions can matter. NOAA leveling guidance specifically notes that refraction increases as the sight line approaches the ground and that thermal gradients can cause measurable errors.[7]

Therefore a low camera looking six miles across a canal is not automatically a superior experiment to geodetic surveying. It may be a particularly sensitive atmospheric-refractivity experiment.

A better design raises the optical path, records temperatures at multiple heights, repeats observations under different conditions, and uses multiple markers or reciprocal measurements.

20. The Bedford Level Lesson

The Bedford Level experiments are historically important because modern flat-Earth literature descends directly from them.

Samuel Rowbotham conducted observations along the Old Bedford River in the nineteenth century and argued that a boat or target remained visible along a six-mile canal when curvature supposedly should have hidden it. His observations became foundational to Zetetic Astronomy.

In 1870 Alfred Russel Wallace accepted a wager challenging anyone to demonstrate Earth’s curvature. Wallace arranged three markers at equal height above the water over the canal. The central marker appeared elevated relative to the line between the two end markers—the pattern expected on a convex surface. A contemporary report in Nature described the setup and result.[10]

The episode also demonstrates the importance of experimental geometry. Rowbotham’s near-water sight line was highly vulnerable to terrestrial refraction. Wallace’s marker arrangement was designed to make curvature visible as a relative displacement rather than relying only on whether a distant object could still be seen.

The Bedford Level therefore does not show that “water was tested and found flat.” It shows why uncontrolled refraction and experimental design matter. The same canal can produce misleading results when one asks only whether a target is visible and clearer results when one measures relative marker geometry.

21. Distant Skylines: What Must Be Measured

Skyline images are among the most visually persuasive “too far” examples because viewers recognize the buildings and distances can be large.

A rigorous analysis needs to identify:

  1. the exact camera coordinates;
  2. the camera lens height above the lake or sea at the moment of capture;
  3. the exact target building coordinates;
  4. ground or shoreline elevation at each building;
  5. building roof and antenna heights;
  6. surface distance, not just a rounded map estimate;
  7. water level or tide if relevant;
  8. which building features are genuinely visible;
  9. temperature and weather conditions over the path;
  10. whether the skyline is vertically distorted;
  11. whether the horizon line in the photograph is physical water, haze, or a mirage boundary.

Without those inputs, statements such as “400 feet too much is visible” are rarely justified.

Another common mistake is measuring building height in pixels and assuming the photograph is an undistorted orthographic projection. Telephoto lenses compress apparent distances, camera tilt changes vertical mapping, atmospheric distortion changes image scale, and cropping removes the original field of view. Pixel analysis can be powerful, but only after calibration.

22. Chicago Across Lake Michigan

The Chicago skyline viewed from Michigan is a classic internet example because some photographs show tall buildings at distances of roughly 40 to 60 miles depending on the location.

Under ordinary geometry, the tallest Chicago towers can remain partly visible from elevated shoreline positions while their lower portions are hidden. Under strong Lake Michigan temperature inversions, the skyline can appear dramatically elevated or more complete than normal.

This is not hypothetical. Atmospheric specialists identify looming and superior-mirage conditions across the lake, and observers report that the amount of skyline visible changes strongly with weather. A 2025 Chicago Magazine discussion quoted atmospheric-optics expertise explaining that unusual full-skyline views from across Lake Michigan are produced by refractive looming under inversion conditions.[11]

The decisive clue is variability. Chicago’s buildings do not change height, and Earth does not flatten on cold-water days. Yet the skyline’s apparent elevation and completeness can change. A variable atmosphere is the changing component.

Some photographs show only the upper parts of the tallest towers. Others show distorted lower structures. Still others produce an apparent floating city. Grouping all of these images under “Chicago is visible, therefore no curvature” throws away the very information that distinguishes geometric line of sight from refractive optics.

23. Long-Distance Mountain Records

Mountain visibility records can reach extraordinary distances because both observer and target may be thousands of meters above sea level.

This is exactly what the two-horizon equation predicts. A 3,000-meter mountain has a geometric horizon hundreds of kilometers away. Put the observer on another mountain and the potential mutual line-of-sight distance becomes much larger.

Thus saying “a mountain was photographed 300 km away” is meaningless by itself. The relevant question is: from what elevation, to what summit elevation, across what intervening terrain, and under what refractive conditions?

Curvature does not say “nothing can be seen beyond 3 miles.” That number applies only to an eye a few feet above a spherical surface. Elevated observers can see vastly farther.

Long mountain sight lines often occur under exceptionally transparent and refractively favorable weather, which is why record observations are rare. Haze, aerosols, clouds, intervening ridges, and refraction usually limit visibility before pure geometry does.

A correct spherical/ellipsoidal terrain model can predict which peaks should be potentially visible. Geographic-information-system viewshed tools perform essentially this calculation using elevation data and Earth curvature. Such models are routinely used in radio planning, surveying, and landscape analysis.

24. Ships Disappearing Hull-First

The old observation that departing ships disappear hull-first is sometimes dismissed as perspective. Perspective certainly makes the whole ship smaller. But simple perspective does not selectively occult the bottom behind a distinct horizon line while leaving the top visible.

On a curved surface, the effect is natural. As distance grows, the hull intersects the geometric horizon first. Upper decks and masts remain visible longer because they are higher.

OpenStax uses precisely this observation in its discussion of evidence for a spherical Earth: a departing ship’s hull disappears while the mast remains visible longer.[12]

Atmospheric refraction complicates the exact sequence. Near water, inferior or superior mirages can compress, stretch, duplicate, or obscure lower parts. Waves can hide small hulls. Haze reduces contrast. But these are corrections to the baseline geometry, not replacements for it.

A useful test is to observe the same route from two camera heights simultaneously. The higher camera should recover more of the lower vessel and delay its disappearance. This is a direct prediction of globe geometry and is difficult to explain with perspective alone, because both cameras can use identical optics while differing mainly in height.

25. Waves, Tides, and Observer Elevation

For low targets near water, seemingly trivial vertical factors matter.

A camera listed as “2 feet above the water” may actually vary by several feet with waves, tide, lake level, tripod placement, dune slope, or the photographer’s location. The target may also sit above the nominal waterline. Over tens of miles, these differences change the expected hidden amount.

A wave only a few feet high can temporarily block the lowest part of a distant vessel because the sight line grazes the water. Surface chop makes the optical horizon irregular.

Tides can change observer and target height relative to the instantaneous sea surface. Storm surge and seiches can alter lake or coastal water levels. A claim precise to a few feet cannot rely on generic “sea level” elevations taken from a map months later.

This is why professional surveying controls benchmark elevations rather than eyeballing them from photographs.

26. Camera Height Errors Are Devastating

Because horizon distance scales with the square root of height, camera-height errors can transform the predicted visibility.

Consider an observer assumed to be 6 feet above water when the actual camera is on a 30-foot bluff. The assumed geometric horizon is only a few miles; the actual horizon is more than twice as far. The difference propagates directly into the hidden-height calculation for a distant target.

Common sources of error include:

If a claimed anomaly is 50 feet but camera elevation is uncertain by 30 feet, the conclusion is not robust.

27. Distance Errors Matter Quadratically

The tangent-drop approximation grows as d². Therefore a modest error in distance can produce a much larger error in the claimed curvature.

For example, confusing 30 miles with 35 miles changes the tangent-drop term by the ratio:

35² / 30² ≈ 1.36

That is roughly a 36 percent increase.

Long-distance photographs are frequently assigned distances using approximate place names: “Chicago from Michigan,” “mountain from city,” “platform from beach.” But coastlines are large, viewing points move, and target structures may be kilometers apart.

Proper analysis uses exact coordinates and geodesic distance. NOAA’s National Geodetic Survey provides tools for computing ellipsoidal distance and azimuth between geographic coordinates.[5]

Distance should therefore be measured, not inferred from a caption.

28. Sea Level, Ellipsoid, Geoid, and Terrain Are Not Interchangeable

Precision visibility analysis becomes complicated because “height above Earth” has several meanings.

GPS receivers naturally estimate height relative to a mathematical ellipsoid. Surveyed elevations are often expressed as orthometric heights relative to a geoid model approximating global mean sea level. Local water surfaces vary with tides, currents, atmospheric pressure, winds, lake hydrology, and gravity-field variations.

NOAA explains that the geoid is an equipotential surface used as a model of global mean sea level and that it differs from the smooth ellipsoid.[13]

For ordinary 20-mile photography, confusing these surfaces by a meter may not matter. For a claimed few-foot anomaly, it can.

Likewise, Earth terrain is not a smooth sphere. A distant building may stand on elevated ground. A mountain’s base may be hidden by an intervening ridge rather than curvature. A lake surface is not necessarily at the same elevation at every instant because winds and pressure can set up small slopes and seiches.

The point is not that calculation is impossible. It is that a serious geometric claim needs appropriate inputs.

29. Why a Photograph Alone Rarely Proves the Geometry

A photograph records angular information projected through a lens after light has passed through the atmosphere. It does not automatically encode absolute distance, elevation, or ray curvature.

Two very different physical scenes can generate similar images. A geometrically visible target under clear air, a partly hidden target under looming, and a more distant target under a superior mirage can all produce recognizable silhouettes.

To turn a photograph into a curvature measurement, one must calibrate it.

Useful calibration elements include a known horizon, stars, known building elevations, surveyed landmarks, camera orientation, sensor geometry, and known focal length. Raw image files are preferable because they preserve metadata and avoid social-media rescaling.

Multiple photographs at different times are vastly better than one. If the apparent target height changes while geometry remains fixed, refraction is immediately implicated.

A photograph can be strong evidence when embedded in a controlled measurement. It is weak evidence when paired only with a screenshot from an online curvature calculator.

30. Pixel Measurements and Image Calibration

Pixel analysis can quantify hidden height, but only if the image scale is known.

Suppose a building of known height spans 200 pixels. It may be tempting to declare that each pixel equals a fixed number of feet. That works only if the building is approximately vertical in the image, lens distortion is small or corrected, the relevant features are at nearly the same distance, the image has not been nonuniformly resized, and atmospheric refraction has not vertically stretched the scene.

Perspective projection also means angular scale, not physical size, is the natural image quantity. A 100-foot feature at 10 miles spans twice the angle of the same feature at 20 miles.

Better analysis converts pixels to angular elevation using the camera’s calibrated field of view. One then compares the observed angular altitude of known points with predicted geometry.

Even then, refraction may shift apparent angular elevation. A precise optical curvature experiment is therefore closer to surveying than casual photography.

31. What a Flat-Earth Visibility Model Would Predict

Critiquing globe calculations is not enough. A flat model must make its own quantitative predictions.

On a genuinely flat plane with a transparent atmosphere and no intervening obstacles, there is no geometric horizon caused by the surface falling away. Distant objects should become progressively smaller and fainter, but their bases should not be hidden from the bottom upward behind a fixed surface horizon solely because of distance.

To reproduce observed hull-down disappearance, the flat model must invoke perspective, atmospheric extinction, refraction, waves, or some other mechanism. Those mechanisms then need equations.

If perspective is the explanation, what focal geometry predicts the exact altitude at which the hull disappears for cameras at different heights? Why does raising the observer restore lower portions of the target in the amount predicted by spherical geometry?

If atmospheric refraction is invoked, why should it systematically create an apparent convex-Earth obstruction under ordinary conditions while occasionally revealing more under inversions?

If the horizon is an optical vanishing zone, why does radar beam height need curvature correction? Why do geodetic leveling networks require curvature terms? Why does the horizon dip increase predictably with observer altitude?

A replacement theory must answer the same measurements, not merely identify uncertainties in individual photographs.

32. Repeatability Is the Key Test

The strongest long-distance observations are repeatable across weather conditions and observer heights.

Suppose a target is photographed every day for a month from a fixed station. On some days only its top appears. On other days more of the lower structure is visible. On rare days it seems elevated or vertically stretched. The target and Earth geometry are constant. Atmospheric refraction is variable.

Now repeat the observations from several controlled heights. Globe geometry predicts that higher observers should see farther and recover more of the target’s lower portions. A flat plane with only ordinary perspective does not naturally produce the same height dependence.

Record temperature at multiple heights, not just from a weather app. A refractive gradient is determined by the atmosphere along the path, especially near the ray. Conditions at a distant airport may not describe the marine boundary layer above a lake.

The objective should be to map the whole response surface: visible height versus camera height, time, temperature profile, and distance. That transforms an internet argument into an experiment.

33. Independent Evidence from Geodesy and Radar

Even if every controversial skyline photograph were discarded, Earth curvature would still be measured by independent methods.

NOAA defines geodesy around measuring Earth’s shape, orientation, and gravity field. Modern coordinate networks use satellite positioning, precise leveling, gravimetry, very-long-baseline interferometry, satellite laser ranging, and other techniques.[9]

Radar systems incorporate Earth curvature because beams rise relative to the surface with range. National Weather Service training materials explicitly discuss the radar horizon and explain that distant low-level storms can be overshot because the beam becomes increasingly elevated relative to the ground.[14]

Surveyors apply curvature correction because a level line is tangent while the equipotential surface curves away. NOAA manuals quantify the effect.[2]

These systems are operational. Bridges, tunnels, pipelines, property networks, flood maps, aircraft navigation, radar meteorology, and satellite systems depend on them. They cannot all be explained by one ambiguous photograph.

A valid “too far” observation would be interesting, but it would have to outperform this entire measurement network, not merely disagree with a misused online calculator.

34. Common Objections Tested One by One

“At 50 miles there are more than 1,600 feet of curvature.”

That is approximately the tangent drop from the observer’s local surface tangent, not automatically the amount hidden from an elevated observer. Hidden height requires observer elevation and line-of-sight geometry.

“The curvature calculator says the object should be hidden.”

Which calculator? Does it calculate tangent drop or hidden height? Did it include observer height? Target elevation? Refraction? Correct distance? A calculator is only as good as the quantity and inputs being used.

“Refraction cannot explain hundreds of feet.”

Under normal conditions refraction is often a modest correction. Under strong inversions, looming and superior mirages can produce much larger apparent vertical displacements and can reveal objects geometrically below the horizon. NOAA navigation guidance explicitly warns that this occurs.

“If it were a mirage, the object would be upside down.”

No. Looming can elevate an image without creating an obvious inverted duplicate. Superior mirages may contain upright, inverted, stretched, or multiple components depending on the refractive-index profile.

“The skyline looks perfectly normal.”

A visually normal-looking image does not prove negligible refraction. Weak or smoothly varying refraction can shift the apparent altitude without dramatic distortion.

“Standard refraction is already included and the object is still too visible.”

“Standard” is an approximation, not an atmospheric law. Near-water temperature inversions can produce superrefraction well beyond the standard correction.

“Refraction is just an unfalsifiable excuse.”

No. Refractive index depends on measurable temperature, pressure, and humidity profiles. Surveying, radar, astronomy, and navigation routinely predict and correct refraction. Time-varying distortions can also be observed directly.

“Water is level, so a laser a few feet above water should stay the same height.”

“Level” in geodesy means perpendicular to gravity and follows an equipotential surface, not an infinite Euclidean plane. NOAA defines the geoid as a model of global mean sea level shaped by Earth’s gravity field.[13]

“If Earth curved, surveyors would constantly account for eight inches per mile squared.”

They do account for curvature where the precision and geometry require it. In differential leveling, balanced short sight lengths can largely cancel curvature and refraction, so one does not necessarily apply a giant accumulated correction at every setup. NOAA manuals explicitly discuss the curvature term and how procedure minimizes it.

“I can zoom a ship back into view.”

Zoom restores angular resolution and contrast if the ship was merely too small to resolve. It cannot recover light rays that do not reach the camera. Controlled tests ask whether the hidden lower hull reappears or only the already-visible upper structure becomes clearer.

“The horizon always rises to eye level.”

Geometrically, the horizon dips below the true horizontal for an elevated observer. The dip is small at ordinary heights, often below casual visual discrimination, but it grows with altitude and is measurable with appropriate instruments.

“Mountains can be seen hundreds of miles away, impossible on a globe.”

Mountain summits can be kilometers high, and observers may also be elevated. Their horizon distances add. Long-distance records therefore do not contradict curvature merely because the raw distance is large.

“There is no curvature visible from an airplane.”

At airline altitude, Earth’s radius is still enormous compared with the observer’s field of view. Wide-angle lens distortion, window shape, camera tilt, and limited horizon arc complicate photographs. The absence of visually dramatic curvature in a casual image does not override measured geodesy.

“The Lake Pontchartrain power lines prove curvature because they lean, but that is just perspective.”

Perspective certainly changes apparent spacing and orientation, so a photograph of poles alone is not a precision curvature measurement. The better evidence comes from surveyed elevations and line-of-sight modeling. One should not replace one simplistic photo argument with another.

“The Black Swan photo proves the horizon is behind an oil platform farther than globe geometry allows.”

Oil-platform images over cold water are taken in exactly the environment where strong refractive layering and looming occur. Any such claim requires platform dimensions, observer height, exact range, visible structural features, and refractive conditions. An apparent horizon in a miraged image need not be the geometric sea horizon.

“Chicago is fully visible across the lake.”

Sometimes unusually complete or elevated views occur under strong inversion conditions; on other days only upper portions are visible. The variability itself is diagnostic of atmospheric refraction.

“Atmospheric refraction would have to bend light exactly with Earth curvature.”

No. Under ordinary conditions it bends less than curvature, extending the horizon somewhat. Under ducting it can approach or exceed Earth curvature temporarily. The atmosphere does not have one fixed ray curvature.

“If light curves downward, why do we not see around the whole Earth?”

Because ordinary refraction is far weaker than required to follow the surface indefinitely, and strong ducts occupy limited altitude ranges and distances. Absorption, scattering, changing gradients, terrain, and duct termination also limit propagation.

35. A Rigorous Curvature-Observation Protocol

If the goal is genuinely to test Earth curvature rather than produce an ambiguous photograph, the experiment should be designed so that globe and flat models make different numerical predictions.

Step 1: Survey the endpoints

Determine exact coordinates and orthometric elevations of the observer and target using reliable survey data or high-quality GNSS/geodetic methods. Measure the actual optical center of the camera and target marker.

Step 2: Use multiple target heights

Place vertically separated lights or markers on the far target. This allows the image to reveal which elevations are visible and which are hidden.

Step 3: Raise the optical path

Avoid rays only inches above water. Several meters or more of elevation reduces sensitivity to the strongest boundary-layer gradients.

Step 4: Use at least three stations

A midpoint marker at the same surveyed height as two endpoints can reveal convexity directly. Multiple stations also expose local elevation errors.

Step 5: Make reciprocal observations

Observe simultaneously from both ends with synchronized cameras or surveying instruments. This helps constrain asymmetric refraction.

Step 6: Record atmospheric profiles

Measure temperature at several heights at both endpoints and, ideally, intermediate locations. Record pressure, humidity, wind, water temperature, and time of day.

Step 7: Repeat across conditions

Repeat at dawn, midday, evening, different seasons, and different water-air temperature contrasts. A true geometric term remains constant; refraction changes.

Step 8: Preserve raw data

Keep uncropped images, metadata, timestamps, calibration frames, survey logs, weather data, and analysis code. Avoid social-media screenshots as primary evidence.

Step 9: Pre-register predictions

Before looking at the result, calculate what a spherical/ellipsoidal model predicts under zero refraction and a range of measured refractive gradients. Separately state what the flat model predicts.

Step 10: Report uncertainties

Give error bars for camera height, target height, distance, water level, and refractive correction. A claimed 20-foot anomaly is meaningless if the combined uncertainty is 60 feet.

Such an experiment can genuinely test geometry. If repeated, controlled observations systematically violated spherical/ellipsoidal predictions after atmospheric effects were constrained, geodesists would have a real scientific problem. But uncontrolled “I can see it” photographs do not meet that standard.

Conclusion

Yes, people can sometimes see farther than a simple no-refraction horizon calculation predicts. That does not mean Earth lacks curvature. It means the simple calculation was not the complete optical model.

The “too far” argument is powerful because it combines genuine observations with an intuitive formula. The formula is usually the problem. “Eight inches per mile squared” approximates the departure of Earth’s curved surface below a tangent plane. It is not automatically the amount of a distant object hidden from an elevated camera. Hidden height depends on both observer and target elevations. Tall structures and mountain peaks possess their own horizon distances, which add to the observer’s.

Once the correct geometry is used, many supposedly impossible sightings become ordinary. Others require atmospheric optics—which is exactly what one expects for near-horizontal rays traveling tens of miles through the lower atmosphere.

Refraction is not an ad hoc escape hatch invented for flat-Earth photographs. NOAA navigation guidance warns that looming can raise normally hidden objects into view. NOAA weather-radar documentation shows beams bending under normal, superrefractive, and ducting conditions. Geodetic leveling manuals apply curvature and refraction corrections because both have measurable effects. The World Meteorological Organization includes curvature and refraction in radar-height calculations. These phenomena exist independently of the debate.

Long-distance water observations are particularly vulnerable. Cold water beneath warm air can establish inversions that bend rays downward, creating looming and superior mirages. A skyline can appear higher, more complete, stretched, duplicated, or even partly inverted. The apparent horizon in such an image need not coincide with the true geometric horizon.

The historical Bedford Level controversy illustrates the same lesson. A low sight line across miles of water can be strongly affected by refraction. Better-designed multi-marker observations reveal curvature rather than merely asking whether a target remains visible.

The globe model also does something the “too far” objection rarely attempts: it makes quantitative predictions across independent fields. Geodetic networks, leveling, radar horizons, satellite positioning, terrestrial navigation, mountain viewsheds, and line-of-sight engineering all use Earth curvature operationally. These systems continue to work because the geometry is not derived from one photograph or one assumption.

A genuine anomaly would require much more than an object being visible. It would require exact observer height, target elevation, distance, terrain, calibrated image geometry, and measured refractive conditions—and it would need to remain inconsistent with the ellipsoidal Earth model across repeated observations. The strongest test would use simultaneous reciprocal measurements and multiple known-height targets.

When that standard is applied, the claim “you can see too far for Earth curvature” loses its force. What is actually observed is a predictable combination of spherical/ellipsoidal geometry, elevation, optical resolution, terrain, and a variable refracting atmosphere.

For the broader set of claims in which these observations are usually presented, see Testing Flat Earth. For two closely related physical claims, see Testing “Gravity Is Just Density and Buoyancy” and Testing “Atmospheric Pressure Requires a Container”.

Sources and Further Reading

  1. NOAA, U.S. Coast Pilot 8 — Refraction, Looming, Towering, and Superior Mirages. NOAA navigation guidance notes that looming can raise objects apparently above their true elevation and can bring objects below the horizon into view.
  2. NOAA National Geodetic Survey — Geodetic Leveling, Manual NOS NGS 3. Gives Earth-curvature corrections and procedures for accounting for atmospheric refraction in precise leveling.
  3. NASA/NSSDCA — Earth Fact Sheet. Gives Earth’s equatorial, polar, and mean radii and other physical parameters.
  4. NOAA National Geodetic Survey — The Geopotential Surface. Explains ellipsoidal Earth models, the geoid, and physical level surfaces.
  5. NOAA National Geodetic Survey — INVERSE/FORWARD Geodetic Computation Utilities. Computes geodesic distance and azimuth on reference ellipsoids including GRS80/WGS84.
  6. NOAA JetStream — Radar Beams and Atmospheric Refraction. Explains normal refraction, subrefraction, superrefraction, ducting, and how Earth curvature affects radar propagation.
  7. NOAA — User’s Guide to Vertical Control and Geodetic Leveling. Describes curvature error, atmospheric refraction, temperature gradients, and the typical magnitude of terrestrial-refraction corrections.
  8. World Meteorological Organization — Guide to Meteorological Instruments and Methods of Observation. Includes corrections to radar-derived heights for Earth curvature and atmospheric refraction.
  9. NOAA National Geodetic Survey — What Is Geodesy?. Describes measurement of Earth’s geometric shape, orientation, gravity field, and geodetic coordinate systems.
  10. Nature — “The Rotundity of the Earth” (1870). Contemporary report on Alfred Russel Wallace’s Bedford Level curvature experiment.
  11. Chicago Magazine — Why Is Our Skyline Visible From Across the Lake?. Discusses Chicago skyline visibility across Lake Michigan and atmospheric looming under inversion conditions.
  12. OpenStax Astronomy 2e — Ancient Astronomy. Discusses disappearing ships and other classical observations of Earth’s curvature.
  13. NOAA National Geodetic Survey — What Is the Geoid?. Explains the geoid as a gravity-defined model of global mean sea level.
  14. National Weather Service — Radar Sampling Issues. Describes the radar horizon and the increasing beam height with range due to Earth curvature under standard refraction.
  15. NOAA National Geodetic Survey — Geodesy for the Layman. Explains why small-area surveys may use plane approximations while large-area geodesy uses an ellipsoid.
  16. NOAA/AFRL Handbook — Optical and Infrared Properties of the Atmosphere. Discusses refractive-index gradients, ray curvature, looming, mirages, and optical propagation through the atmosphere.