The Science Behind Your Star
Every star we identify comes from a documented calculation you can check yourself. This page explains, in plain language, how the sky of one moment is reconstructed — the coordinates, the time conversion, the star catalog and the selection rule — including the parts we deliberately do not model.
A Star Named is an astronomical memorial service that reconstructs the sky at the recorded time and location of a person’s or pet’s passing and identifies the star closest to the zenith.
1. Location — where the sky is measured from
The place you enter is resolved to a latitude and a longitude. Latitude fixes which part of the sky passes overhead: at any instant the zenith’s declination equals the observer’s latitude. Longitude shifts local time and the sky’s rotation angle — it decides when that point comes overhead (section 3).
Elevation is not used — and here is why
We do not adjust for height above sea level. Altitude changes sunrise and sunset by a minute or two, but the star we identify lies within a few degrees of the zenith, where elevation has practically no effect. It also means we never ask you for a number you may not know. If a moment falls exactly at dawn or dusk, elevation could in principle shift the pick — we accept that small uncertainty instead of inventing precision.
2. Time — from a clock reading to an instant in UTC
You enter the local civil time at that place — the time that was on the clock. We convert it to UTC using the IANA timezone database, which carries the real history of clock rules for the region, including daylight saving. Every astronomical computation then runs on UTC alone.
- Daylight saving is handled from the database: the same clock reading means a different instant in summer and winter, and the correct one is used for your date.
- Impossible times are refused, not guessed: if the wall-clock time you enter never existed on that date (the hour skipped when daylight saving begins), the calculator asks you to adjust rather than silently computing a different moment.
- Incomplete records are corrected and disclosed: for daylight saving in China, 1986–1991 — a real period that today’s timezone databases omit — we apply a documented correction so the sky is still right.
- Historical dates: before standard time was adopted (the date varies by country; for the United States it was 1883), towns kept local mean time — solar time, not clock time. Databases model that era inconsistently, so on historical pages we record a confidence level for every date and time, and where the time is not documented we say so instead of guessing.
3. The astronomical calculation — sidereal time, hour angle, altitude, azimuth
Given an instant in UTC, a latitude and a longitude, we find the point of the sky directly overhead:
- The instant is converted to a Julian Date, the standard day-count that astronomical formulas take as input.
- Greenwich Mean Sidereal Time (GMST) is computed from the standard IAU 1982 polynomial — the sky’s rotation angle, in degrees.
- Local Sidereal Time (LST) = GMST + east longitude. LST is the right ascension currently crossing your meridian, the north–south line through the zenith.
- The zenith of that instant: right ascension = LST, declination = latitude. This single point is what we match against the catalog.
- Because the catalog is stored in J2000 coordinates, the zenith point is precessed back from the epoch of the date to J2000 with the IAU 1976 precession model — for a 2026 moment that shift is about 0.37°, far too large to ignore.
- Altitude is the angle above the horizon (the zenith is altitude 90°); azimuth is the compass bearing measured from true north through east; the hour angle measures how far the sky has turned since an object crossed the meridian (15° per hour). These are the coordinates we use to show your star’s position and its visibility windows.
- Deliberately not modelled: nutation, aberration and polar motion — together well under 0.02°, an order of magnitude below the typical separation between neighbouring stars. We state the error budget instead of pretending to precision we do not need.
4. The star catalog — HYG v4.1
Candidates come from the HYG catalog, a real public catalog merging the Hipparcos, Gliese and Yale surveys: J2000 positions (RA/Dec), distances from parallax, brightness and spectral type. We include 107,838 stars covering all 88 IAU constellations; every record is traceable to its source IDs. The full accounting lives on our data page.
The catalog spans magnitude 0.03 to magnitude 12 — far below the naked-eye limit of about 6.5. The star nearest the zenith is often a faint one. That is the honest trade-off of choosing by position rather than brightness, and your certificate states the magnitude so you know what to expect.
One more honest note: positions are stored at epoch J2000, and we do not yet propagate each star’s proper motion to the date of observation. For nearly every star that amounts to a few arc-seconds over a human lifetime — far below the selection resolution — but for a handful of very fast nearby stars it can matter. The correction code is in place; it will be switched on as proper-motion values are added to the catalog rows.
5. Choosing the star — the exact rule
Stated the way it is implemented:
- Take every catalog star whose declination is within 15° of the zenith’s declination — a wide-margin speed filter; the nearest star is never near the band edge.
- For each candidate, compute the angular distance from the zenith by spherical trigonometry.
- Choose the smallest angular distance: the real star closest to the zenith.
- Skip stars that are already named — each real star is used only once. On an exact tie in distance, the star closer to Earth wins.
- The same inputs always produce the same star: no randomness, no human touch. The chosen star’s catalog IDs and coordinates are printed on the certificate.
6. Check it yourself
You do not have to take any of this on faith — Stellarium (free and open source) or any planetarium app can confirm the result:
- Set the same date, time and location you entered.
- Look straight up: that point is the zenith. With the equatorial grid on, you can read its coordinates.
- Find the star by the catalog ID printed on your certificate (HIP number or similar) — it should sit nearest the zenith, at the angular distance your certificate reports.
7. Known limitations
- Mean positions only: nutation, aberration and polar motion are not modelled (together well under 0.02°).
- Proper motion is not propagated; it matters only for a few very fast nearby stars.
- Elevation above sea level is not used (section 1).
- Historical timekeeping varies by region and era; historical pages carry a confidence level per date, and we do not publish a star for a historical moment whose time of day is not documented.
- Visibility is computed geometrically (altitude > 0°); atmospheric refraction near the horizon is not applied. The star is identified by position, not brightness — it is usually too faint for the naked eye.
Go deeper
星背后的科学
我们找出的每一颗星,都来自一套你可以亲手复核的计算。这一页用平实的语言讲清楚:某一刻的星空是如何被重建的——坐标、时间换算、星表与选星规则——包括我们刻意没有建模的部分。
A Star Named 是一项天文纪念服务:它重建一个人或一只宠物离世时刻、在其所在地的星空,并找出最靠近天顶的那一颗真实恒星。
1. 位置:从哪里量这片天空
你填写的地点会被解析为一组经纬度。纬度决定天空的哪一部分经过头顶:任意时刻,天顶的赤纬都等于观测者纬度。经度决定时刻与天空转过的角度——也就是那个点何时升到头顶(见第 3 节)。
高程:不使用,以及为什么
我们不按海拔高度做修正。海拔会让日出日落时刻差上一两分钟,但我们找出的星位于天顶附近几个角距之内,在那里海拔几乎没有影响。这也意味着我们不会向你索要一个你可能并不知道的数字。如果某一刻恰好落在黎明或黄昏,海拔理论上可能影响选星结果——我们宁愿接受这点不确定性,也不虚构不存在的精度。
2. 时间:从钟面读数到 UTC 时刻
你输入的是当地民用时间——当时钟上显示的时刻。我们用 IANA 时区数据库把它换算为 UTC;该数据库记录了各地区真实的时钟规则历史,包括夏令时。此后所有天文计算只依赖 UTC,不再做任何时间假设。
- 夏令时由数据库处理:同一个钟面读数在夏季与冬季对应不同的时刻,我们按你给出的日期取正确的那一个。
- 不存在的时刻会被拒绝,而不是猜测:如果你输入的钟面时间在那一天从未出现过(夏令时开始时被跳过的一小时),计算器会请你调整,而不会默默换成另一个时刻。
- 不完整的记录会被修正并披露:例如中国 1986—1991 年的夏令时——真实存在、但今天的时区数据库未收录——我们应用了有据可查的修正,让星空依然正确。
- 历史日期:在标准时间被采用之前(各国时间不同,美国为 1883 年),城镇使用的是地方平太阳时——太阳的时间,而非钟表的时间。数据库对那个年代的建模并不一致,所以在历史人物页面上,我们为每一条日期与时刻标注可信度等级;时刻无据可查时,我们如实说明,绝不猜测。
3. 天文计算:恒星时、时角、高度角与方位角
给定 UTC 时刻、纬度与经度,我们求出天空正头顶的那一点:
- 先把时刻换算为儒略日(Julian Date)——天文公式通用的日数基准。
- 用 IAU 1982 标准多项式计算格林尼治平恒星时(GMST),即天空转过的角度(度)。
- 地方恒星时(LST)= GMST + 东经。LST 就是此刻正经过你天顶子午线的赤经。
- 这一刻的天顶:赤经 = LST,赤纬 = 纬度。这个点就是与星表匹配的目标。
- 星表坐标以 J2000 历元存储,所以天顶点要用 IAU 1976 岁差模型从当日历元岁差回 J2000——以 2026 年的时刻为例,这一位移约 0.37°,大到不可忽略。
- 高度角是目标在地平线以上的角度(天顶 = 90°);方位角是从正北起、向东量的罗盘角;时角衡量天体过中天后天空转过的角度(每小时 15°)。你的星的方位与可见时间窗就是用这些坐标给出的。
- 刻意不建模的部分:章动、光行差与极移——合计远小于 0.02°,比相邻恒星之间的典型角距还小一个数量级。我们把误差预算写清楚,而不是假装需要更高的精度。
4. 星表:HYG v4.1
候选星来自 HYG 星表——一张真实的公开星表,合并了依巴谷(Hipparcos)、格利泽(Gliese)与耶鲁亮星目录:J2000 位置(赤经/赤纬)、由视差换算的距离、亮度与光谱型。本站收录其中 107,838 颗,覆盖全部 88 个 IAU 星座,每条记录都可追溯到来源编号。完整的统计与版本见我们的数据页。
星表的星等范围是 0.03 等到 12 等——远暗于肉眼极限(约 6.5 等)。最靠近天顶的那颗星,常常是一颗暗星。这是"按位置而非按亮度选星"的诚实代价;证书上会写明星等,让你心里有数。
还有一条诚实说明:坐标以 J2000 历元存储,我们目前尚未把每颗星的自行传播到观测日期。对绝大多数恒星,几十年也只有几角秒——远低于选星的角距分辨率——但对极少数高速近邻星会有影响。修正代码已经就位,待自行数据补入星表行后即会启用。
5. 选星:与实现一致的精确规则
规则按实现原样陈述:
- 取所有赤纬与天顶赤纬相差 15° 以内的星——这是留足余量的加速筛选;最近的星绝不会落在带宽边缘附近。
- 对每颗候选星,用球面三角计算它与天顶的角距。
- 取角距最小者:距天顶最近的那颗真实恒星。
- 跳过已被命名的星——每颗真实恒星只使用一次。角距完全相同时,离地球更近的星胜出。
- 同样的输入永远得到同一颗星:没有随机,不经人手。选中的星表编号与坐标会印在证书上。
6. 亲自复核
这些都不需要你凭信任接受——用 Stellarium(免费开源)或任何星空软件即可核对:
- 把日期、时刻与地点设为与你输入的一致。
- 抬头看正上方:那一点就是天顶。打开赤道网格即可读出它的坐标。
- 用证书上印的星表编号(HIP 号等)找到那颗星——它应当距天顶最近,角距与证书所写一致。
7. 已知限制
- 只计算平均位置:章动、光行差与极移均未建模(合计远小于 0.02°)。
- 未传播自行;仅对极少数高速近邻星有可见影响。
- 不使用海拔高度(见第 1 节)。
- 历史计时方式因地区与年代而异;历史页面为每个日期标注可信度等级——时刻无据可查时,我们宁可不发布这颗星,也不猜测。
- 可见性为几何计算(高度角 > 0°),未做地平附近的大气折射修正。选星按位置而非亮度——选中的星通常暗于肉眼可见。