Lattice OS
Engage
Kessler · Lattice OS
L

Lifecycle Atmospheric Risk Service (LARS)

The substrate catalog is the instrument.
v0.5.0 snapshot 2026-11-05 49,998 LEO objects
Two daily-refresh observables. β_rel(h) per-shell storm-response coefficient (200–450 km) plus SWADI F(h) substrate-weighted atmospheric density factor (450–1500 km). The catalog as a distributed in-situ atmospheric instrument. βrel (h) — storm-response coefficient per altitude shell 0.00 0.02 0.04 0.06 0.08 200 250 300 350 400 450 altitude (km) SWADI F(h) — substrate-weighted atmospheric density per altitude shell 0.0 1.0 2.0 3.0 4.0 F = 1 450 600 800 1000 1200 1500 altitude (km) 0.88 2.01 1.34 1.13 1.38 3.61 1.82 1.21 1.46
Top panel: βrel(h) per shell — median (dot) with p25–p75 whisker showing per-shell cohort spread. βrel falls from the 250–300 km shell to the next and rises strictly only over 300–450 km (A17). Bottom panel: SWADI F(h) bars across 450–1500 km, F=1 dashed reference is the Vallado climatology baseline. Two observables, complementary altitude regimes, one subscription.

Methodology

(1) F(h) = Σi wi · ρobs,i  /  Σi wi · ρbase,i SWADI substrate-weighted atmospheric density per shell (2) wi ∝ exp(−σmag,i2 / 2σsmooth2) · ri2 substrate weighting: clean line-integrators get full weight; cascade-coupled noisy objects down-weighted (3) dh/dt(t) = K(h) · [1 + βrel(h) · ap(t)] storm-response decomposition: β_rel measured per shell from cross-object regressions

The substrate catalog is the instrument. βrel(h) regresses 226 cross-object pairs across 5 altitude shells; the strongest pair correlates at r = +0.949 across 159 days, in 300–350 km shared shells, between two unrelated active payloads. Independent objects measure the same atmospheric forcing — the residuals are common-mode signal, not per-object noise.

Definitions: ρobs,i is the King-Hele residual atmospheric density inferred from object i's dh/dt; ρbase,i is the Vallado-climatology baseline at the same altitude. wi weights low-|σmag| dead-mass objects (clean atmospheric line-integrators) over high-coupling cascade participants (whose orbits cross too many wakes to be clean). βrel(h) is the linear sensitivity of normalized drag enhancement to apdaily, fit per-shell from differenced residuals over a 159-day window. Both observables refresh daily against the F2-validated GFZ-Potsdam 95-year archive.

What you receive

βrel(h) cohort — A17

Shell (km) n_obj βrel p25 βrel median βrel p75 positive sign
200–250 10 0.0106 0.0640 0.0681 8/10
250–300 31 -0.0020 0.0128 0.0142 19/31
300–350 32 0.0010 0.0066 0.0066 25/32
350–400 28 -0.0038 0.0092 0.0102 18/28
400–450 23 -0.0024 0.0122 0.0140 16/23

A15 (per-object correlation): 16 of 42 objects pass at r ≥ 0.3, 81% positive correlation. A16 (cross-object differenced-residual correlation): 163 of 226 pairs positive (72%); strongest pair r = +0.949. A17: the analyzer's own rule (more steps rise than fall) passes; read strictly as monotonic, A17 is FAIL.

SWADI F(h) snapshot — 9 populated shells

Shell (km) n_obj F factor departure from Vallado
450–500 12 0.88 -12% below baseline
550–600 11 2.01 +101% above baseline
600–650 24 1.34 +34% above baseline
650–700 6 1.13 +13% above baseline
700–800 34 1.38 +38% above baseline
800–900 8 3.61 +261% above baseline
900–1000 25 1.82 +82% above baseline
1000–1200 21 1.21 +21% above baseline
1200–1500 5 1.46 +46% above baseline

F = 1.0 means observed atmospheric density matches the Vallado climatological baseline at that altitude. The 850 km shell sits at F = 3.61 (3.6× nominal) — the catalog detected a real local thermospheric anomaly the climatology cannot. Coverage gaps where n_obj < 5 are surfaced explicitly per audit claim A9, not extrapolated.

Engagements are direct. Engage by email →

Email with the asset, fleet, or jurisdiction you want assessed and a sample lands in your inbox within one day.

tony@fancyland.net