LA-026 · EOS-05 / GISAT-1A · GSLV-F17…
EOS-05 / GISAT-1A · GSLV-F17 — Launch-Day Risk Assessment Sealed Before the Window · Two-Level Conditional Ladder · Eighteen-Window Branch Map · Cryogenic…
Sealed claim: LA-026 — EOS-05 / GISAT-1A · GSLV-F17 (GSLV Mk II) · LAUNCH-DAY RISK ASSESSMENT · A TWO-LEVEL CONDITIONAL LADDER + AN EIGHTEEN-WINDOW BRANCH MAP · SEALED 28 AUGUST 2026, BEFORE THE WINDOW SEALED 28 AUGUST 2026 — six days before the window opens. probAtSeal 0.90 (operator-set, both levels). EXECUTIVE SUMMARY — LA-026 · EOS-05 / GISAT-1A · GSLV-F17 WINDOW NET FRIDAY 4 SEP 2026 IST · 02:30–06:30 IST (opens 3 SEP 21:00 UTC — the night of 3/4 Sep) · SECOND LAUNCH PAD, SATISH DHAWAN SPACE CENTRE, SRIHARIKOTA · GTO INJECTION OVERALL VERDICT: NOT A CLEAN FLIGHT. The risk rides the LAUNCH VEHICLE, never the payload, and the CRYOGENIC UPPER STAGE is the named high-risk subsystem. The structure is a two-level conditional ladder committed before the window — LEVEL 1, P = 0.90: the attempt in this window — 02:30–06:30 IST Friday 4 September (3 Sep 21:00–01:00 UTC) — does not achieve a clean, on-time mission success (any scrub or delay grades it, whatever cause is attributed — scrub is scrub — and any ascent/vehicle anomaly grades it); LEVEL 2, P = 0.90 given launch: a serious ascent-phase anomaly at the cryogenic upper stage, severity up to loss of mission — the CE-7.5 stage whose chill-down and ignition near T+4.6 minutes is GSLV's historical single point of failure and the exact mode that lost the predecessor EOS-03/GISAT-1 on GSLV-F10 in August 2021. If it flies, the realized T-0 selects the operative window from the eighteen-window branch map below: exactly one window's consequence-set activates as that branch's testable prediction; the other seventeen grade VOID. IN EVERY BRANCH THE SATELLITE ITSELF IS ASSESSED SOUND — this is a launch-vehicle call. THE FALSIFIER, STATED AT SEAL: a clean, nominal, on-time success grades this advisory a MISS at full weight, both levels. RECOMMENDATION: do not commit at window-open; if the mission flies, the RECOMMENDED CORE is 03:26–03:46 IST (late W5 through W6) — the recommended slip target, same night, zero external cost — with the flank 03:46–04:00 IST (W6's close into W7); 04:27–05:39 IST (late W9 through W14) is avoided outright, and inside it 04:39–04:48 IST is the CATASTROPHIC overlay: do not be counting down through it. The §08 map states, window by window, what this advisory expects if the team launches there anyway. §01 FRAME & SCOPE. EOS-05 (GISAT-1A) is ISRO's geostationary imaging satellite — rapid, repeat disaster imagery from GEO via a 700 mm Ritchey–Chrétien telescope with multispectral (~42 m) and 158/256-channel hyperspectral imagers — selected-region revisit ~every 5 minutes, the full Indian landmass ~every 30 — ≈2.2-tonne class to GTO (predecessor EOS-03: 2,268 kg at launch, roughly half of it the propellant for the satellite's own GTO→GEO raise) — riding GSLV Mk II flight GSLV-F17 from the Second Launch Pad, Sriharikota, in a four-hour window opening 02:30 IST on Friday 4 September 2026 (3 Sep 21:00 UTC). This advisory addresses the launch and its day; it imputes no fault to ISRO and asserts no non-public information. The reading binds to this window only (the night of 3/4 September 2026) — a slip to a new date is its own address and re-seals as its own reading. §02 THE PUBLIC PRIOR — WHY THE BASELINE IS NOT "ROUTINE". Three public facts set the baseline, and none of them says routine. (a) GSLV Mk II carries ISRO's poorest operational record — approximately 67% career success (18 flights: 12 full successes, 2 partial, 4 failures), the vehicle Indian commentators nickname the "naughty boy," its failures clustered on the cryogenic upper stage. (b) The GISAT payload line is 0-for-3 on its own history: two attempts called off (March 2020, March 2021), then the outright loss of EOS-03/GISAT-1 on 12 August 2021, when the GSLV-F10 cryogenic upper stage failed to ignite at T+4.56 minutes — first and second stages nominal, mission lost. (c) ISRO returns to flight here after a roughly seven-month gap, its first mission since the PSLV-C62 failure of 12 January 2026 — the reliable workhorse itself down. The countervailing narrative is real and is stated: the last three GSLV missions succeeded, including NISAR (GSLV-F16, July 2025) on the newer cryogenic stage. The field would price meaningful slip risk and non-trivial vehicle risk; this advisory prices both higher, names the subsystem, and stakes the difference. §03 THE CONDITIONAL LADDER — COMMITTED BEFORE THE WINDOW. LEVEL 1 (P = 0.90): no clean, on-time mission success in this window (02:30–06:30 IST Fri 4 Sep; opens 3 Sep 21:00 UTC). Any scrub or delay grades this level regardless of the attributed reason — scrub is scrub; attribution is explanatory color, never a grading condition — and any ascent or vehicle anomaly grades it. LEVEL 2 (P = 0.90 given launch): a serious ascent-phase anomaly at the CRYOGENIC UPPER STAGE — expected presentation at or near the stage's chill-down/ignition event, approximately T+4.6 minutes — with a severity arc running from off-nominal performance and insertion shortfall up to loss of mission. BRANCH SELECTION (binding): if liftoff occurs, the realized T-0 selects the operative window from the eighteen-window branch map of §08; exactly one branch's consequence-set activates as that branch's testable prediction, and the remaining seventeen grade VOID — never counted as hits or misses. IF NO LAUNCH OCCURS IN THE WINDOW (scrub or stand-down, any reason): Level 1 grades HIT as called — and Level 2's condition never activates, grading NEITHER hit nor miss (VOID), with all eighteen branches VOID; the cryogenic-stage read goes untested, not vindicated, and the record's own information-yield pricing scores a scrub-only outcome as the field's modal expectation — a low-yield hit, stated at seal. PAYLOAD, WINDOW-INVARIANT: in every branch the satellite is assessed sound; EOS-05 itself is not the locus of risk. §04 RANKED MANIFESTATIONS — THE EXPECTED PRESENTATIONS, STRONGEST FIRST. (1) Cryogenic-stage ignition/chill-down anomaly at ~T+4.6 minutes — the GSLV-F10/EOS-03 mode: thrust fails to build or builds late; mission lost or orbit severely short. (2) Cryogenic propellant-conditioning shortfall — loading-density or tank-pressurization drift (the LH2 tank-pressure path named by the F10 Failure Analysis Committee) presenting as a hold, a recycle, or an underperforming stage. (3) Insertion shortfall — an off-nominal GTO (low perigee/apogee shortfall) requiring the satellite to spend its own propellant, shortening station life without losing the mission outright: roughly HALF of GISAT-1A's launch mass is the fuel for its own GTO→GEO climb, so every meter-per-second the vehicle leaves short converts directly into lost mission years — the time-cost of §07, paid at the payload level. (4) Countdown attrition — holds and last-moment recycles that consume the window and force either a late-window commit or a scrub. (5) In the late-window branches only (W11–W14, with the tail into W15–W16): separation-event discontinuity — stage, fairing, or spacecraft separation — and telemetry/link dropout, the abrupt-event class. The desk states one explicit departure from the modeled norm: the consensus improvement narrative (three consecutive GSLV successes) is expected NOT to carry this day. §05 PRE-LAUNCH ACTIONABLES — WHAT TO INSPECT WHILE INSPECTION IS STILL POSSIBLE. Every named item is available only while the vehicle stands: CE-7.5 igniter-chain and pyrogen verification; full chill-down sequence rehearsal and thermal-conditioning review; LH2 tank pressurization integrity — the vent-and-relief path the GSLV-F10 Failure Analysis Committee named — re-verified under flight-representative timeline; propellant loading-density margins re-derived for the night thermal environment; thrust-buildup monitoring thresholds and the abort/hold criteria at the ~T+4.6 minute event reviewed against the F10 telemetry; separation-system pyro continuity checks for the late-window branches. None of this is exotic; all of it is cheap against the cost of the day. §08 THE EIGHTEEN-WINDOW BRANCH MAP — THE FULL FOUR HOURS, TILED AT FINE GRAIN, WITH THE CONSEQUENCE OF IGNORING EACH. Times IST (UTC). The realized T-0 activates exactly one row; the rest grade VOID. Each row states what this advisory expects IF THE TEAM LAUNCHES INSIDE THAT WINDOW. Two named overlays cross the row boundaries and are sealed as time-ranges: THE RECOMMENDED CORE = 03:26–03:46 IST; THE CATASTROPHIC OVERLAY = 04:39–04:48 IST (read: loss of mission). · W1 · 02:30–02:34 (21:00–21:04) — LAUNCH HERE AND the count is fighting the vehicle from the first second: the commit-at-open cluster — maximum expression of the cryogenic-ignition risk (the F10 mode), hold/abort pressure in the opening minutes, vehicle-body vibration and structural stress, ground-side last-moment discontinuity. Alongside the 04:39–04:48 catastrophic overlay (W10–W11), this is one of the two stretches where the mission is most likely to be lost. VERDICT: NO-GO — DO NOT COMMIT AT OPEN. · W2 · 02:34–02:49 (21:04–21:19) — LAUNCH HERE AND the opening-phase stress is still draining: residual propellant-conditioning irregularities and countdown-timeline friction; toward the close of this window the data-integrity caution of W3 begins. Survivable, not chosen. VERDICT: CONDITIONAL — IMPROVING. · W3 · 02:49–03:03 (21:19–21:33) — LAUNCH HERE AND the failure class is informational: telemetry ambiguity, data-integrity gaps, at least one sensor reading the team cannot fully resolve inside the count — the half-signal that forces a hold decision on incomplete data. VERDICT: CONDITIONAL — DATA-INTEGRITY CAUTION. · W4 · 03:03–03:18 (21:33–21:48) — LAUNCH HERE AND the data-integrity caution persists and margins narrow toward the window's close: expect at least one unresolved reading and hold pressure late in this band. VERDICT: CONDITIONAL — DATA CAUTION, THINNING. · W5 · 03:18–03:32 (21:48–22:02) — LAUNCH HERE AND the band opens with abrupt small-event risk — a sudden hold, a recycle, a short-notice parameter excursion — BUT from 03:26 the night's best configuration begins: the RECOMMENDED CORE (03:26–03:46) opens inside this window's tail. VERDICT: DEGRADED OPEN — THE CORE BEGINS 03:26. · W6 · 03:32–03:47 (22:02–22:17) — THE HEART OF THE CORE. The cleanest configuration of the night: cryogenic-ignition risk at its minimum expression, margins healthiest, nominal-ascent indications strongest through 03:46. If the mission must fly in this window, fly it HERE. VERDICT: RECOMMENDED — THE CORE. · W7 · 03:47–04:01 (22:17–22:31) — LAUNCH HERE AND the configuration is still favorable but thinning: minor thermal-conditioning drift re-enters and margins ease off the core's peak. The acceptable flank of the core, not the core. VERDICT: GOOD — THE FLANK. · W8 · 04:01–04:15 (22:31–22:45) — A brief steadier pocket: LAUNCH HERE AND margins hold, though the recommendation has passed; residual guidance/avionics transients possible. VERDICT: ACCEPTABLE — STABILIZATION POCKET. · W9 · 04:15–04:30 (22:45–23:00) — LAUNCH HERE AND navigation/guidance transients enter and dispersions become possible; from ~04:27 the propulsion/fire anomaly class activates — window quality decays into the avoid band at this window's close. VERDICT: ACCEPTABLE — DEGRADING, AVOID FROM 04:27. · W10 · 04:30–04:44 (23:00–23:14) — LAUNCH HERE AND the propulsion class is active — thrust-buildup irregularities, combustion-adjacent faults — and from 04:39 THE CATASTROPHIC OVERLAY OPENS (04:39–04:48): the read inside it is LOSS OF MISSION, the F10 outcome repeated. Do not be counting down through the close of this window. VERDICT: AVOID — CATASTROPHIC FROM 04:39. · W11 · 04:44–04:58 (23:14–23:28) — LAUNCH HERE AND the catastrophic overlay holds through 04:48 — loss of mission remains the read in the first minutes — then the abrupt-event class takes over: stage-, fairing-, or spacecraft-separation discontinuity; telemetry or communications dropout at a critical event. VERDICT: CATASTROPHIC TO 04:48 — THEN AVOID. · W12 · 04:58–05:12 (23:28–23:42) — LAUNCH HERE AND separation-discontinuity risk is near maximum, with combined staging-anomaly and link-loss presentation. VERDICT: ELEVATED — AVOID. · W13 · 05:12–05:26 (23:42–23:56) — LAUNCH HERE AND the abrupt-event class holds with obstruction-flavored faults surfacing late in the ascent — the worst stretch of the late window. VERDICT: ELEVATED — AVOID. · W14 · 05:26–05:39 (23:56–00:09) — LAUNCH HERE AND the elevated stretch closes; abrupt-event vulnerability persists to 05:32, easing after. VERDICT: ELEVATED — AVOID, EASING LATE. · W15 · 05:39–05:53 (00:09–00:23) — LAUNCH HERE only if the earlier core was forfeited: the telemetry-ambiguity class returns at reduced intensity; acceptable margins, uninspiring configuration. VERDICT: FALLBACK — DEGRADED. · W16 · 05:53–06:07 (00:23–00:37) — LAUNCH HERE AND the abrupt-event tail persists at reduced intensity with short-notice discontinuity risk; flyable, earns nothing. VERDICT: FALLBACK — WEAK. · W17 · 06:07–06:20 (00:37–00:50) — LAUNCH HERE AND late-window fatigue shows in timeline and margins; the closing stretch improves modestly but the window's value is spent. VERDICT: FALLBACK. · W18 · 06:20–06:30 (00:50–01:00) — The last minutes of a four-hour window: commit pressure maximal, timeline fatigue real. LAUNCH HERE AND the operational-error class compounds the residual technical one. VERDICT: FALLBACK — SCRUB PREFERRED. THE WINDOW HEALTH CURVE (sealed companion dataset): the same analysis is also published as a continuous per-minute score across the full 240-minute window — 241 rows at jyotishintelligence.com/dataset/la026-window-curve.json, whose SHA-256 is companion-sealed into the Bitcoin-anchored manifest at seal. The curve is EXPOSITION: it visualizes why each branch says what it says; grading binds to this §08 map and its overlays, never to individual minutes. §08A THE DECISION PACKAGE — THREE COURSES, PRICED FOR THE LAUNCH AUTHORITY. PACKAGE A — FLY AT OPEN (02:30 IST), as currently scheduled: this accepts the sealed NO-GO branch with the commit-at-open cluster as the read; and a count that slips uncorrected downstream drifts toward the avoided 04:27–05:39 stretch, including the catastrophic 04:39–04:48 overlay. Package A is the cost of ignoring this advisory entirely. PACKAGE B — SLIP INSIDE THE WINDOW (RECOMMENDED): same night, same range, zero external cost — move T-0 into the RECOMMENDED CORE, 03:26–03:46 IST; counts recycle by minutes as routine range practice, and launch programs re-target inside an open window without announcement. This is the cheapest risk reduction on the table: the window does not remove the Level-2 signature — the §05 actionables do that work — but it takes the day's best configuration instead of its worst. PACKAGE C — STAND DOWN: scrub, work the §05 actionables on the cryogenic stage while the vehicle can still be touched, and re-baseline; a slipped date is a new address and re-seals as its own reading. The decision is the launch authority's, always: this advisory is an augmenting input, never a go/no-go authority. §06 THE BEHAVIORAL READ (intelligence reasoning, method-agnostic). The conduct around this launch is itself evidence — with the obvious confound conceded first: THE CLOCK IS NOT THE TELL. A GTO injection window is set by orbital mechanics, and GSLV has flown in the pre-dawn hours before (GSLV-F10 itself lifted off at 05:43 IST); no inference is drawn from the hour. The tell is the PUBLICITY POSTURE AROUND the hour: no agency build-up, no national-event staging, no spectacle machinery for a return-to-flight mission — beat coverage exists, but the program whose brand is public triumph (Chandrayaan-3 was a national spectacle) is not staging one. An actor that moves against its own strongest incentive reveals private information it will not state: an agency confident of success maximizes the audience; an agency quietly minimizing the audience privately rates its odds low. The posture is consistent with an institution bracing for a difficult day. TWO FURTHER POSTURE EXPECTATIONS, REGISTERED BEFORE THE FACT: (1) the live telecast is expected to be CURTAILED OR ABSENT — an agency in this posture can decline a live broadcast entirely, citing the hour and the absence of an audience; a full, traditional, live national telecast would cut against this read. (2) If anything goes wrong, official acknowledgment is expected to be MINIMAL AND DELAYED — a nervous, PR-constrained agency does not narrate its own anomaly in real time. Neither expectation is scored; both are stated now so the conduct around the day can be checked against the read. §07 THE STAKES (consequence assessment — not a separate scored claim). With the PSLV workhorse itself just failed (January 2026) and the program politically owned at the top and under heightened national-security attention, a GSLV failure here lands as a political event, not merely a lost satellite — the reading most parsimonious with the muted posture around the day. Public commentary additionally frames the mission as relevant to India's space-based surveillance ambitions — near-real-time GEO imaging is dual-use by nature — a framing (commentators', not ISRO's stated purpose) that widens the stakes of a second GISAT loss beyond disaster response. And the time-cost is already priced by this line's own history: the 2021 failure bought a five-year national capability gap; a second failure prices at another half-decade — the loss class no insurance replaces. This paragraph assesses what a failure would trigger; it is not a separate sealed prediction. §09 RECOMMENDATIONS. Do not commit at window-open (W1). If the mission flies, the RECOMMENDED CORE is 03:26–03:46 IST (late W5 through W6) — the recommended slip target — with the flank 03:46–04:00 IST (W6's close into W7) beside it; the 04:27–05:39 IST stretch (late W9 through W14) is to be avoided outright, and 04:39–04:48 IST is the catastrophic overlay — do not be counting down through it; W15–W18 are a fallback, never a target, and the final minutes prefer a scrub. The §08A decision package prices the three courses: fly-at-open accepts the sealed NO-GO; the in-window slip to the core (03:26–03:46 IST) is the cheapest risk reduction on the table; a stand-down re-baselines to a new address that re-seals as its own reading. The window selection is a mitigation, never a clearance: the Level-2 signature attaches to the vehicle, and the pre-launch actionables of §05 — not the clock — are what remove it. The decision remains the launch authority's, always; this advisory is an augmenting input, graded in public whatever the outcome. §10 GRADING PROTOCOL — PRE-COMMITTED. TIER 1 (T+1 day, from the webcast, ISRO official statements, and tracking): Level 1 grades on the day's facts — scrub, delay, or anomaly grades it; a clean on-time flight grades it MISS. If liftoff occurs, the realized T-0 selects the operative branch of §06 and that branch's named anomaly-set becomes the gradeable prediction; the other seventeen branches grade VOID. TIER 2 (T+days, from public orbital data and ISRO statements): Level 2 grades on the ascent record and the achieved orbit — a cryogenic-stage anomaly, an insertion shortfall against the published target GTO, or a loss of mission grades it; a nominal insertion begins its falsification, completed by the satellite's nominal arrival at station. Anti-hedge clock: grades flip on the live sheet from public evidence and are never revised after the fact. Misses are kept at full weight. OPACITY-PROOF BY CONSTRUCTION (registered at seal): this protocol assumes the agency may be minimally transparent, and it grades from PUBLIC SOURCES ONLY — onboard telemetry is not public, is expected to remain hidden, and is never required. The channels: liftoff and its time are independently observable (range closures/NOTAMs, live public reporting, observers); the achieved orbit is independently measurable from the public satellite catalogs (SATCAT/TLE data — 18th SDS via space-track and public mirrors). Agency statements are corroboration, never the arbiter; where official statements and public orbital data conflict, THE DATA GOVERNS. A curtailed or absent live telecast changes nothing in this protocol and was itself anticipated in §06. Where the public record cannot resolve the failing subsystem, Level 2 grades CONSERVATIVELY — an unresolved attribution never defaults toward the call. BOUNDARY. This assessment is opinion. It rests on no non-public information; it asserts no fact about any organization's internal knowledge or condition; it addresses a mission and its day. The SHA-256 of this sealed text enters the Bitcoin-anchored manifest at seal.
- Sealed on public record
- 2026-08-28
- SHA-256
886dea385fb3e80ad11d4ad210792a2a7324fbff577d4e8f47abfee6112846ec