Panchāṅgam

How our Panchāṅgam is calculated

Last reviewed: 28 October 2026

The daily Panchāṅgam and muhūrta guidance on this site are computed for the San Francisco Bay Area in the spirit of a traditional almanac — no vague predictions, no sensational astrology, no advertising and no tracking. This page explains exactly how the figures are produced, so you can judge them for what they are.

What we compute

For each day we calculate the five limbs of the Panchāṅgam — tithi, nakṣatra, yoga and karaṇa, together with the vāra (weekday) — along with sunrise and sunset. It also shows the traditionally restricted starting periods (Rāhu Kālam, Yamagaṇḍam and Durmuhūrtam), the context-dependent Guḷika Kālam, and the traditional reference periods (Brahma Muhūrta, and Abhijit only when it does not overlap a restricted period), plus festival and vratam dates.

Those periods are not all the same kind of claim, and the Panchāṅgam page keeps them in separate lists rather than one favourable column. A prayer period such as Brahma Muhūrta is a time for prayer and recitation and is not offered as a window for beginning anything. A general commencement window is daylight clear of the periods in which a new undertaking is not begun — it says nothing is in the way, which is not the same as saying it is the right hour. A ceremony candidate is such a window long enough to hold the whole of a particular rite, offered as something to take to Gurukkal. Where a favourable period and a restricted one cover the same minutes, the overlap is stated with the minutes named and is not resolved, because resolving it is his to do and not this page’s.

Convention and ayanāṃśa

We follow the Drik (observation-based) convention, using the Lahiri (Chitrapakṣa) ayanāṃśa — the most widely used sidereal reference in South Indian practice. Since October 2026 it is computed from its official definition — the value the Indian Calendar Reform Committee fixed for 1956, carried forward by the precession of the equinoxes, with nutation added (the “true” Lahiri used by the Indian Astronomical Ephemeris) — rather than from a straight-line approximation. An independent computation of the same definition agrees to 0.01 arc-second. Tithi, nakṣatra, yoga and karaṇa are derived from the true (apparent) longitudes of the Sun and Moon corrected by the ayanāṃśa.

Astronomy

Since October 2026 the Sun and Moon are computed from the planetary theories professional almanacs use: VSOP87 (Bretagnon and Francou) for the Sun, ELP/MPP02 (Chapront and Francou, fitted to JPL’s DE405) for the Moon, IAU 1980 nutation, aberration and light-time, in Terrestrial Time with ΔT from the IERS observations. The planets for the muhūrtham checks use VSOP87 as well. Every build compares all of it with NASA JPL’s DE421 ephemeris — see the figures below. (Before October 2026 the engine used Jean Meeus’ shorter series, good to about 36 arc-seconds for the Sun; that was enough to move the start of a Tamil month by up to fifteen minutes.) Sunrise and sunset are found by stepping through the local civil day in two-minute intervals, computing the Sun’s altitude at each step, and interpolating the moment it crosses the horizon altitude; the result agrees with the NOAA solar-position algorithm to a tenth of a minute. The daily “kālam” periods (Rāhu, Yama, Guḷika) are derived in the traditional way — the daytime from sunrise to sunset is divided into eight parts, and the relevant part is assigned by weekday.

Technical disclosure

• Coordinates: 37.6624, −121.8747 (Pleasanton, CA); timezone America/Los_Angeles.
• Sunrise / sunset altitude: the standard −0.833°, allowing for atmospheric refraction and the solar disc — the sunrise in a newspaper or a phone. Some Indian almanacs define sunrise differently: as the moment the centre of the Sun’s disc meets the true horizon, with no refraction. Here that sunrise comes about four to five minutes later and that sunset about as much earlier. No Tamil Thirukkaṇita publisher we could find states which definition it uses, so this site keeps the standard one, says so here, and has put the question to Sri Kumar Gurukkal. What it can change: the kālam periods start up to about four and a half minutes differently, and a tithi or nakṣatra “at sunrise” changes only if it ends inside those few minutes.
• Ayanāṃśa: true Lahiri (Chitrapakṣa), from its official 1956 definition with precession and nutation; convention: Drik (observation-based), the reckoning Tamil almanacs call Thirukkaṇita.
• Kālam method: the daytime is divided into eight equal parts between sunrise and sunset, and each period is assigned by weekday.
• Abhijit overlap: Abhijit Muhūrta is presented only when it does not overlap Rāhu Kālam, Yamagaṇḍam or Durmuhūrtam; on days where it overlaps, it is withheld rather than shown as usable.
• Published reference range: January 1, 2026 through December 31, 2028 — the span of the festival and observance data. The 2026 and 2027 dates come from a regional Panchāṅgam reference for Northern California. Since October 2026 this site also recomputes them with its own rule engine as a check, and recomputing a date is not the same as verifying it. The 2028 dates are this site’s own computation. Each named festival is computed from its standard definition — the lunar month, fortnight and tithi; or the nakṣatra; or the day of a solar month — written down and tested against every 2026–27 date. Where a written source gives the deciding window or tie-break, the source decides: śrāddha at aparāhṇa (Dharmasindhu), Mahā Śivarātri and Janmāṣṭamī at the middle of the night, Gaṇeśa Caturthī and Rāma Navamī at madhyāhna, Karva Chauth and Sankaṣṭī at moonrise. Where no source gives it, the window is the one that reproduces the 2026–27 dates, and the page says so. A 2028 date is held, not guessed, when its tithi prevails at no sunrise or at two and nothing settles which day; each held date is listed on the festival calendar as to be confirmed with Gurukkal, and each festival page says why. For a sample of 49 major festivals, the 2028 dates were compared with an independent almanac computed for India (prokerala.com): 45 fall on the same day or one day earlier — the expected effect of the Bay Area being twelve and a half hours behind India — one (Karva Chauth) is two days earlier, as the 2026 reference data also is, and three are held. Two published 2027 dates were corrected in the same review: Hariyālī Amāvāsyā (1 August, not 31 August) and Vaikuṇṭha Ekādaśī (moved into Mārkaḻi: 7 January 2028, not 9 December 2027). No date on this site has yet been reviewed by Gurukkal. Beyond 31 December 2028 the astronomy still computes, but no festival dates are published.
• Checked against NASA JPL, on every build (from October 2026): every change of tithi, nakṣatra, yoga and karaṇa, and every saṅkrānti, from January 2026 to December 2028 — 5,622 instants — is recomputed from JPL’s DE421 ephemeris (a planetary ephemeris NASA’s Jet Propulsion Laboratory produced for spacecraft navigation) by a separate program, and the site’s engine must agree with each. It does: the largest difference for any tithi is 1.8 seconds, nakṣatra 1.2 s, yoga 1.2 s, karaṇa 1.8 s, and saṅkrānti — the start of a Tamil month — 3.8 seconds. Daily positions agree to 0.2 arc-second for the Sun and 0.7 for the Moon; the five planets to 9 arc-seconds. (The previous engine was out by up to 103 seconds on a tithi and 25 minutes on a saṅkrānti.) That is a check of the astronomy: it establishes when a tithi ends, not which day a festival is kept, and it is not a review by Gurukkal.
• Independent check, run on every build (from 31 August 2026): sunrise and sunset are compared against a second, separate implementation — the NOAA solar position algorithm, written out from the published equations in tools/verify_panchangam.js rather than shared with the engine. Across nine dates and locations, including both daylight-saving transitions and a leap day, the two agree to better than 0.1 minute. Since October 2026 the Moon is also checked against a second, separate implementation on every build: the instants of new and full moon — where every tithi begins and ends — are computed from Jean Meeus, Astronomical Algorithms, chapter 49, written out from the published formula and sharing no code with the engine. Across 150 new and full moons in 2026–2028, at Fremont and at Delhi, the largest difference is 1.33 minutes. That confirms the arithmetic; it does not establish any festival’s rule. The Moon is also checked against physical constraints: thirty tithis must average one synodic month (29.5422 d against 29.530588 d), twenty-seven nakṣatras one sidereal month (27.3246 d against 27.321582 d), and every tithi length must fall inside the range the Moon’s own speed allows. An earlier hand check on 12 August 2026 against a traditional Drik reference agreed within about ±3 minutes; the automated checks above are what run now. The same date returns identical values whether the device is set to Los Angeles, New York, London, Kolkata, Singapore or UTC.
• The muhūrtham finder’s classical rules (added 28 October 2026): for a wedding, gṛha praveśam, upanayanam, nāmakaraṇam, annaprāśanam or akṣarābhyāsam the muhūrtham finder now also applies the rising sign (lagna) and the houses the planets occupy from it, Jupiter and Venus combustion, eclipses (computed from JPL DE421), the hours around a saṅkrānti, the Tamil months Āṭi, Puraṭṭāci and Mārkaḻi, and the nakṣatras, tithis and weekdays the classical texts give for the rite — and, only if a family enters birth stars, Chandrāṣṭamam, Chandra Bala, Tārā Bala and a lagna in the 8th from a birth rāśi. Each is a rule in the ruleset with the status sourced: taken from Bṛhat Saṃhitā, the muhūrta chapters of the Nārada Purāṇa, Dharmasindhu or the Sūrya Siddhānta, cited rule by rule, with at least two texts agreeing wherever a rule sets a time aside, and not yet reviewed by Gurukkal. Where the texts or families differ on a rule the finder applies, the stricter reading is used and the difference is stated — with one exception, stated: the months follow Tamil practice (no wedding or gṛha praveśam in Āṭi, Puraṭṭāci or Mārkaḻi) rather than the classical lunar-month lists, which set aside more months — whether those also apply is Gurukkal’s to say. The lagna is checked against the IAU’s SOFA library (agreement within a minute of arc). The finder proposes a few dates with every reason shown; Gurukkal fixes the muhūrtham.
• Known limitations: it does not read a personal horoscope or planetary strength, and it does not apply the lunar-month rules, the “childhood” and “old age” of Jupiter and Venus, the navāṃśa, family conditions such as a first-born’s own month, or tables for rites other than the six above. Each of those is recorded as a proposed, inactive rule in the site’s own muhūrtham ruleset, which names every one of them as not evaluated rather than quietly omitting it. The daily panel on the homepage computes for the Pleasanton reference location; the Panchāṅgam page and the muhūrtham page compute for whichever location you choose.
• Computed and shown, but not used to rule on suitability (added 4 October 2026): the lagnam (rising sign, with its real unequal intervals), the twenty-four horai, the sixteen Gowri divisions, and — only if you choose to enter a birth star or birth rāśi — Tārā Balam and Chandra Balam. These are on the Panchāṅgam page under “advanced”. Three distinctions matter here and this site keeps all three. Computed and shown as information is not used to decide whether a time is suitable: on the Panchāṅgam page no window is offered or withheld because of a tārā, a house or a rising sign. (Since October 2026 the muhūrtham finder does use them, as cited classical rules — see below.) Computed is not reviewed: the reading of which positions are favourable is the widely-stated one, it is labelled as such on every row, and it is recorded in the ruleset as a question still open with Sri Kumar Gurukkal. And a birth star alone is not a birth rāśi: nine of the twenty-seven nakṣatras span two rāśis, and for those this site asks for the pāda rather than guessing. Anything you enter stays in your browser tab — it is never added to the web address, never saved, never sent anywhere, and never written into a calendar file.
• The Tamil calendar fields, and the status of each (added 4 October 2026): Mācam is the sidereal rāśi the Sun occupies, so the months run saṅkrānti to saṅkrānti and Cittirai begins around 14 April rather than on 1 April. Ṛtu follows from the mācam, two months to a season. Ayaṉam is where two conventions genuinely disagree by about three weeks: tropically, Uttarāyaṇa begins at the winter solstice around 21 December; sidereally it begins at Makara Saṅkrānti around 14 January, which is also Tai 1 above. This site is a Drik almanac and follows the tropical definition, which is what the published Drik Panchāṅgam makers use; the sidereal date is stated alongside it so that a reader who expected Tai 1 can see both, and neither is presented as the only answer. Varuṣam is the one field here that is not a position — it is a name from the sixty-year cycle on an anchor taken from sources (since October 2026, two calendar publishers agree that Parābhava began on 14 April 2026), so it is marked as not yet confirmed by Gurukkal. The day of the Tamil month (since v25.1): which civil day a saṅkrānti opens is a rule about the hour of the crossing, and for the Tamil calendar the sourced rule is sunset — a crossing before that day’s sunset begins the month that day, after sunset the next (Mishra, The Calendars of India, arXiv:1007.0062; Dershowitz and Reingold, Indian Calendrical Calculations). The month and its day are therefore read at your sunset, so the same crossing can begin Tai on a different civil day in the Bay Area than in Chennai, and both are the rule applied correctly. This is the computed (Thirukaṇita) reckoning; a Vākya almanac can differ by a day. Other regions use other cut-offs — Kerala aparāhṇa, Odisha the next sunrise, Bengal midnight — and the festival calendar applies each to its own festivals; the build checks that this page and the festival engine agree on every day of 2026–2028.
• The Tamil daily-sheet fields (added 28 October 2026): the Gowri periods by name, the Yōgam (Amirtha, Siddha, Maraṇa, Prabalāriṣṭa), Sūlam and its parikāram, Chandrāṣṭamam, Kari nāḷ, Mēl / Kīḻ / Sama nōkku, Vaḷarpirai / Tēypirai, the śrāddha tithi, the Tamil reading of each horai, and Agni Nakṣatram in season. None of these is a position of the sky: each is a table or a rule that Tamil calendar publishers print, applied here to computed times. For each one we compared several publishers, and the page says how far they agree — the Gowri weekday table is printed identically by three; the Yōgam table by one, but nineteen of its entries were checked against four independent daily calendars and all matched; Kari nāḷ agrees across four publishers in nine months and the other three months’ variants are shown; Chandrāṣṭamam reproduces every printed example we checked, to the minute. Where publishers disagree, the disagreement is printed rather than settled: the last Gowri period of Saturday night is printed by every full table as Sōram a second time, and it is shown that way with a flag. None of these tables has been reviewed by Sri Kumar Gurukkal, and each field says so. Every table is checked cell by cell against the research record on every build (tools/verify_tamil_daily.js), and the sources are listed on the Panchāṅgam page.
• Not published, because they are not verified: Nalla nēram (Tamil calendars print it from fixed weekday tables, two different tables are in print, and no publisher states the rule — the named Gowri periods are the computed part of it); Nētram and Jīvaṉ (the one published rule does not reproduce a major publisher’s own printed values); the Vāstu day; and the lunar month names. No value is shown for any of them, because an invented one would be indistinguishable on the page from a computed one.
• Calculation methods: only Drik (Thirukkaṇita) is implemented. Vākya and Sūrya Siddhānta are different computations, not different labels on this one, so they are reported as unavailable rather than offered — a Vākya heading over a Drik result would be a false capability. Choosing Tamil names changes labels and not a single number; choosing Amānta or Pūrṇimānta changes which month a date is named in and never when a tithi ends.
• Searching across 2026–2028: the Panchāṅgam page finds every occurrence of a tithi or nakṣatra by interval, not by day label. This matters because a tithi that begins after one sunrise and ends before the next prevails at no sunrise at all and therefore has no day label — 2026 has eighteen such intervals. A search written over day labels returns nothing for them and looks as though it worked. Which civil day a particular festival is kept on remains a ritual rule, and the search reports the astronomy rather than deciding it. Note the published range above: the search computes across all three years; the 2026 and 2027 festival dates come from a regional reference and the 2028 dates are this site’s own computation, so for any date a family will act on, please confirm with Gurukkal.

Two layers, kept apart

Since 31 August 2026 the astronomy and the tradition live in separate files, because they are different kinds of thing. assets/js/panch.js computes positions: objective, testable, the same for everybody. assets/data/tradition-profile.js holds the interpretation — which periods are set aside, which tithis are avoided for beginning something, how Guḷika Kālam is treated. Every rule in it records where it came from and whether Sri Kumar Gurukkal has reviewed it, and the ruleset carries a version number that is stamped onto every recommendation the site gives. Rules he has not been shown do not run at all.

See it working on the muhūrtham page, which shows the reason for every window it offers and for everything it sets aside. The underlying ideas — what a tithi is, why sunrise decides the day, why two almanacs disagree — are explained in Sacred Learning.

Location and time

• Reference location: Pleasanton, California — latitude 37.6624, longitude −121.8747. Timings shift by only a minute or two across the immediate Bay Area, so other Bay Area cities are close but not exact.
• Timezone: America/Los_Angeles, with daylight saving handled automatically.
• Display: all times are shown in Pacific Time (America/Los_Angeles), regardless of the timezone your device is set to, so the figures always describe the Bay Area day.
• Accuracy, measured rather than estimated: against NASA JPL’s DE421 ephemeris, every tithi, nakṣatra, yoga and karaṇa change in 2026–2028 agrees to within two seconds and every saṅkrānti to within four (details above); sunrise and sunset agree with the NOAA algorithm to a tenth of a minute. A difference of more than that from another almanac is a difference of convention — the place, the definition of sunrise, or Vākya rather than Thirukkaṇita reckoning — not of arithmetic.
• If our times differ from another source by a few minutes, check the location first. Sunrise is local. The twelve places this tool offers span 4.7 minutes of sunrise on the same morning — San Francisco is 2.1 minutes after Pleasanton and Stockton 2.6 minutes before it, and every kālam derived from sunrise moves with it. A published almanac set to a different city, or to a different sunrise convention or ayanāṃśa, will differ for reasons that are not errors on either side.

Guḷika Kālam in Sri Kumar Gurukkal’s tradition

According to Sri Kumar Gurukkal’s tradition and practice, Guḷika Kālam is a context-dependent period rather than a general prohibition. Ordinary poojas and other ceremonies may be performed during this time. Certain death-related rites should not be begun during Guḷika Kālam. Please consult the priest responsible for the rite for final guidance. Applications can differ according to sampradāya, family custom and the nature of the rite. (Sri Kumar Gurukkal does not perform funeral rites; he does perform ancestral remembrance such as Śrāddha and Tarpaṇam.)

Why observances can differ

Two careful almanacs can legitimately differ. Festival dates and observances vary by location, sampradāya and the exact timing of the relevant tithi (for example, whether a tithi is current at sunrise, at midday or at moonrise). Where a date commonly differs, please treat this calendar as Bay Area guidance and confirm important observances with Gurukkal.

Automated guidance is not a personal muhūrtham

This is the most important point. The tool gives preliminary guidance — it is not a muhūrtham personally selected by Sri Kumar Gurukkal, and no date shown here should be read as “approved” by him. For a wedding, gṛha praveśam or any important ceremony, the exact muhūrtham must be confirmed with Gurukkal, who considers the family’s birth details, the ceremony and your tradition.

Specifically, this tool does not:

• select or approve a final Muhūrtham;
• replace a priest;
• perform a horoscope match;
• evaluate all family traditions or ceremony-specific rules;
• guarantee correctness or a result;
• provide medical, legal or financial advice.

Accuracy and corrections

The engine’s festival output has been checked against a traditional Drik reference and matches closely. Even so, an almanac is a serious thing: we warmly welcome review by qualified Panchāṅgam scholars, and we will correct anything found to be wrong. If you notice a timing or terminology discrepancy, please contact SaivaPriest.com with the date, location and the source you compared against. Important ceremony decisions should always be confirmed directly with a qualified priest. Please see Corrections & contributions.

View today’s full Panchāṅgam & festival calendar →

Background reading

• What a Panchāṅgam actually isPañcāṅga means “five limbs”. A Panchāṅgam is not a horoscope and not a list of predictions.
• Why two Panchāṅgams can disagreeAlmost always because they are using different conventions, not because one is wrong.
• Why a muhūrtham changes between India and CaliforniaBecause the almanac is anchored to sunrise, and sunrise is a local event.