Logbook
Methods, evidence, and decisions
Dated accounts of substantive work across the platform: data validation, derivation evolution, lift definitions, ensemble interpretation, terrain diagnostics, smoke and radiation, palette design, catalogue growth, and operations.
Why usable lift can sit above the boundary layer
The solid usable-lift line often rides above the dashed boundary-layer line; the two answer different questions, and the trail to why runs through Dr Jack's own caveats, a 1980 aircraft campaign, and a coefficient choice in canadarasp's source
Seven forecast-data failures that passed parsing
HTTP 200, decoded cleanly, values a chart could draw, and wrong anyway; it took an independent witness to prove what each payload actually represented
Column ceilings can hide ensemble spread
Adding 500 hPa headroom exposed spread that a 700 hPa ceiling had concealed
The stability ramp, measured
Colour-vision floors, a defect every reviewing eye missed, and the lineage behind the shipped ramp
Adopting a model two months before its retirement
Four candidates, one retirement notice, the sunset field, and slugs as identity
Retiring 121 metres per degree
Bolton's LCL replaces an inherited constant, and de-capped hours surface
B/S does not transfer to the valley
Why surface-to-boundary-layer shear can fail at mountain sites
Smoke and thermals
How smoke enters the documents, and why the w* correction is gentler than intuition says
The wire told on us
Two days of transport measurement found a serial fetch loop, an unequal mirror, and a builder failing one request in ten
Measured against the sky
GOES-18's downward shortwave and aerosol optical depth arrive as observation models; on their first live day, four sites read 554–623 W/m² under a BC plume that thinned from AOT 3.0 to 1.4 in under two hours