Shipley Farms · Vilas, Watauga County, North Carolina
Sun on 115 acres
The farm sits in a bowl. The ridge to the east stands eleven degrees above the horizon and takes an eighth of every year’s daylight — and almost none of its energy. Mornings start late here. Evenings run long.
- Sunlight a year kWh per square metre
- Hours of sunshine
- Daylight the ridges take
- Mean skyline degrees above the horizon
- Centroid 36.2466° N, 81.7504° W
- Elevation 2,896 ft / 883 m
- Tract 115 acres / 46.5 ha
- Basis 1991–2020 normal
The bowl
Every number on this page is downstream of one fact about the ground. Standing in the middle of the tract and turning a full circle, the skyline never drops to the flat horizon an almanac assumes. It averages degrees up, and it is not symmetric.
The skyline from the middle of the farm
Horizon elevation in every direction, computed from USGS 3DEP elevation at 10 m out to 10 km and 45 m beyond, with Earth curvature and standard refraction applied. The sun’s track on three dates is drawn over it: where a track runs inside the grey mass, the sun is behind a ridge.
Hover a bearing, or focus the chart and use the arrow keys, to read the ridge behind it.
An eastern wall
The highest point of the skyline sits at bearing °, east-north-east: a ridge km away that rises to m and stands degrees above the farm. Averaged across the whole eastern quadrant the skyline is degrees up.
A western notch
The lowest is at bearing °, almost due west, where Linville Creek drains out of the valley and the skyline falls to degrees. The western quadrant averages degrees — less than half the eastern. Sunrise is late here. Sunset is not early.
Checked against a second DEM and a second algorithm
The European Commission’s PVGIS service computes its own horizon for any coordinate from SRTM elevation at about 90 m. Asked for this one it returns the same bowl: correlation 0.888 across 48 bearings, with the same eastern maximum and the same western minimum. It reads 0.54 degrees lower on average, which is what a 90 m grid does to a sharp ridgeline. Two independent elevation datasets and two independently written algorithms, and the disagreement is half a degree.
Late mornings, long evenings
The bowl costs the farm hours of daylight a year — percent of the hours an unobstructed site would get. That sounds severe. The energy it costs is percent.
Both numbers are correct, and the gap between them is the whole point. The sun the ridges take is the sun within a few degrees of the horizon, arriving through ten or more atmospheres at a glancing angle. An hour of it carries a few per cent of the energy of an hour at noon. What the farm loses is not light. It is schedule.
Minutes of sun taken at each end of the day
How long after almanac sunrise the sun clears the eastern ridge, and how long before almanac sunset it drops behind the western one. Averaged over 1991–2020 at minute resolution.
In midsummer the asymmetry is extreme. On 21 June the almanac puts sunrise at and sunset at . On the ground the sun appears at and disappears at : minutes gone in the morning, in the evening. The farm trades a late start for a long evening.
By midwinter the sun rises and sets to the south-east and south-west, where the skyline is lower and more even, and the loss splits almost in half — minutes at dawn against at dusk. But the day is short to begin with, so December loses the largest share of its daylight of any month.
| Month | Almanac rise | Over the ridge | Almanac set | Behind the ridge | Daylight | Sun on the ground |
|---|
The part that matters is the frost
An hour and a quarter of missing morning sun in June is not an energy problem, it is a drying problem. Dew and frost sit on the east side of this farm after they have burned off an unshaded field at the same elevation, which pushes back when hay can be cut and baled and lengthens the window in which wet grass stays wet. The percent energy figure is the honest answer to “how much sunlight does the farm lose”. It is the wrong answer to “when can we work”.
How much light actually lands
Across the year the farm receives kilowatt-hours of solar energy on every square metre, an average of a day. For a place with this reputation for cloud, that is more than most people would guess — and more than several cities that sell themselves on sunshine.
Annual sunlight, this farm against eight places
Daymet 1 km global horizontal irradiation, 1991–2020, extracted the same way at every location so the comparison is like for like.
The farm gets per cent more annual sunlight than Raleigh and per cent more than Miami. Neither is a trick: Florida and the Carolina Piedmont lose their afternoons to convective cloud through the whole summer, while a high valley at this latitude keeps a flatter, duller, but longer-lasting supply. Against the genuinely sunny west the gap is real but modest — Phoenix beats this farm by per cent, not by double.
Through the year
Bars are the sunlight that lands on a square metre each month. The line is how many hours the sun is actually above the farm’s own skyline, which is not the same as the almanac day length.
The seasonal swing is wide: in the best month against in the worst, a factor of . Year to year, though, the total barely moves. The standard deviation across 1991–2020 is per cent, against 23 per cent for this farm’s rainfall. Sunlight is the most reliable input this farm has.
Every year since 1980
Annual sunlight at the farm pixel. The band is one standard deviation either side of the 1991–2020 mean. The dimmest year on record, 2018, is also the farm’s wettest year on record — 79 inches of rain. The two go together.
Which number to believe
Gridded radiation products disagree with each other here, and one of them is badly wrong. Sorting that out needed a ground instrument, and the nearest ones threw up a problem of their own.
The only reference-grade pyranometers anywhere near these mountains belong to NOAA’s Climate Reference Network, at two stations near Asheville about 115 km south-west. They cannot validate this farm’s number. They can answer the question that matters, which is whether a gridded product is biased in terrain like this at all.
Bias over matched days. Annual totals within 3.6 per cent in all six years.
Bias over the same period at a station 12 km from the first, and growing since 2017.
What 8 SSW reads as a fraction of 13 S on the clearest days, when cloud cannot be the explanation.
Two pyranometers 12 km apart correlate at , so they are seeing the same weather. Yet one reads per cent of the other on the brightest days, and after peaking at the ratio falls away to . No atmosphere does that to one station and not its neighbour. Soiling, sensor drift or vegetation growing into the field of view will, and the station showing it is the one that disagreed with Daymet.
One instrument drifting away from its neighbour
Annual ratio of the Asheville 8 SSW pyranometer to Asheville 13 S, for all days and for the clearest tenth of days. A ratio of 1.0 would mean agreement.
So Daymet is used for the headline figures, unbiased to within a fraction of a per cent against the station that is internally consistent. UNKNOWN Whether Daymet carries a bias specific to this valley cannot be tested, because no instrument has ever measured solar radiation here. The validation shows the product is not biased in this kind of terrain generally; it is not a measurement of this farm.
One number worth flagging, and it is a big one
The ERA5 reanalysis, which many weather sites and API feeds serve as “sunshine hours”, reports hours of sunshine a year at this location: per cent of all available daylight. That would put a farm in the Appalachian cloud belt alongside Denver. It is wrong, and the same dataset says why: it puts the fraction of sunlight arriving as diffuse skylight at , when the clearness measured here implies . ERA5 is sending too much energy down the direct beam, and its sunshine count — which is just a threshold on that beam — inherits the error.
No sunshine recorder has ever operated at or near this farm. Inverting the Ångström–Prescott relation on the measured sunlight gives hours a year, per cent of possible, with a range of to hours across the usual coefficient choices. That is the figure on this page, and it is a derived estimate, not an observation.
What that much light is
A square metre is hard to picture. The tract is 115 acres, or square metres, and the arithmetic from there is startling.
gigawatt-hours, or petajoules
at 10,500 kWh a year each
megawatts, continuously, day and night
megawatts falling on the tract at once
Put another way, the sunlight that lands on this farm in a year carries the energy of about tonnes of crude oil. Essentially all of it is re-radiated as heat, evaporates water, or is thrown back to the sky; the sliver that photosynthesis captures as grass is a fraction of one per cent, and the sliver of that which leaves as beef is a fraction of a per cent again. DESIGN This page does not attempt the grass-to-beef chain, which needs the farm’s own yield and carcass records rather than public climate data.
For anyone pricing a solar array
The obvious worry is that the eastern wall ruins the economics and the panels should be aimed west to compensate. They should not. Searching azimuth and tilt against the real skyline moves the best-aimed direction only degrees west of due south, and the ridges cost a fixed array about per cent of its annual clear-sky yield. Aim it south. DESIGN This is a geometry result from a clear-sky model, not a yield estimate; a real quote needs shading analysis at the array’s own position, not the tract centroid.
Method, and what this cannot tell you
How these figures were built
Solar positions come from NOAA’s solar position algorithm, stepped at one minute through every day of 1991–2020, and check against published sunrise times for Boone to the minute. The horizon was computed by marching outward along 1,440 bearings over USGS 3DEP elevation — 10 m sampling to 10 km, 45 m to 55 km — applying Earth curvature and standard refraction, which matters: the curvature drop alone is 43 m at 25 km.
Irradiation is Daymet V4 at 1 km, validated against NOAA Climate Reference Network pyranometers. Direct and diffuse were separated with the Erbs correlation on the measured clearness index rather than taken from a reanalysis. Blocked beam energy uses a Hottel clear-sky transmittance at this elevation; blocked skylight uses the sky view factor, , integrated from the same horizon.
What is an estimate, not a measurement
Sunshine hours, the diffuse fraction, every energy figure in section 05 and the solar array geometry are modelled, and labelled where they appear. No instrument measuring sunlight has ever operated on this tract.
Known limits
UNKNOWN Horizon from one point. Everything here is the skyline from the tract centroid. 115 acres of Appalachian hillside is not one point: a field on the eastern slope sees a lower eastern ridge and gets earlier sun, and one tucked under the ridge gets less. The spread across the tract is not computed.
UNKNOWN A 45 m grid smooths ridgelines. Beyond 10 km the elevation sampling will slightly under-read sharp peaks, which makes the sun loss here a floor rather than a ceiling. The PVGIS comparison suggests the effect is a few tenths of a degree.
Vegetation is ignored. The horizon is bare earth. Treelines on the near ridges add real elevation to the skyline, and the sun loss is correspondingly understated for fields close to woodland.
The cheapest way to close the gap
A pyranometer on the tract would, over a few years, turn the headline sunlight figure from a validated model into a measurement.
Sources
- Solar radiation
- Daymet V4 R1, Thornton et al., ORNL DAAC, 1 km single-pixel extraction, 1980–2025. doi:10.3334/ORNLDAAC/2129
- Radiation validation
- NOAA U.S. Climate Reference Network, daily SOLARAD, Asheville 8 SSW and Asheville 13 S, 2015–2020.
- Reanalysis comparison
- ECMWF ERA5 via the Open-Meteo historical archive, 0.25° cell, 1980–2025.
- Terrain
- USGS 3DEP elevation, delivered through the National Map ImageServer. Independent horizon check: European Commission PVGIS 5.3, SRTM-derived.
- Position
- Tract centroid as used throughout this site’s weather pages.
Shipley Farms, 1699 Linville Creek Rd, Vilas, NC 28692. Five generations in Watauga County since 1872. All measurements on this page describe the 115-acre home tract at 36.2466° N, 81.7504° W.
No shortcuts since 1872