Martial Systems · convective environments

AlleyForge

Theil-Sen slopes of annual significant tornado parameter (STP) and SPC tornado reports, 1979-2017, 28-48°N, 110-80°W. Environment climatology on that window.

Research note (PDF)

1979-2017 result

Mean STP slope about 95°W

Gensini and Brooks (2018) reported that accumulated STP declined west of 95°W and rose east of it. The same sign is recovered here on NARR-derived annual STP (archived 2026-08-16).

West of 95°W	−0.154
East of 95°W	+0.432
Sign	west < 0 < east

Figures are the archived 2026-08-16 run.

Method

STP product and day definition

Product: STP = (SBCAPE/1500) × ((2000−SBLCL)/1000) × (SRH/150) × (BWD/20) (Thompson / Gensini-Brooks)
CIN: STP set to 0 when |CIN| is at least 50 J kg⁻¹
Day: 12-12 UTC diurnal maximum, then annual sum
Grid: native NARR Lambert, bin-average to 1°, 1° Gaussian

Limitations

NARR is a proxy, not the paper's AGL layers

Public NOAA PSL NARR does not publish the paper's exact 0-1 km storm-relative helicity or 0-6 km AGL bulk wind. This run used:

SRH slot: NARR 0-3 km hlcy in place of 0-1 km SRH (Bunkers right-mover)
Shear slot: 10 m to 500 hPa in place of 10 m to 6 km AGL bulk wind

Those substitutions are not calibrated against true-layer SRH01 or BWD06. Low-level helicity and deep-layer shear are not interchangeable. The maps on this page remain that frozen proxy.

Accurate AGL layers are being built from NARR 3D pressure-level wind and geopotential height: height AGL is Z_level minus Z_sfc, then 0-1 km storm-relative helicity (Bunkers right-mover) and 10 m to 6 km bulk wind. The 3D files are pulled for that construction. ERA5 true-layer cubes for 1979-2017 are on disk. Until ERA5 recovers the same 95°W sign, the archived NARR slopes stay the public stand-in.

Maps

Theil-Sen 1979-2017

Theil-Sen slope of annual accumulated NARR STP, 1979-2017 — Annual STP (environment)

Theil-Sen slope of all-rating SPC tornado reports, 1979-2017 — SPC reports, all ratings

Theil-Sen slope of EF1+ SPC tornado reports, 1979-2017 — SPC reports, EF1+

Kansas (37-40N, 102-95W): environment −0.065, EF1+ reports +0.008. Arkansas-Louisiana (31-34.5N, 94-89.5W): environment +0.664, EF1+ reports +0.016. Kansas reports slope up while the environment slopes down. That residual is kept visible. It is not averaged away.

Stage B incomplete

ERA5 true-layer (1979-2017)

ERA5 0-1 km SRH (Bunkers RM) and 10 m to 6 km AGL shear, same STP product, CIN rule, and 12-12 UTC day. Cubes are on disk. This block is a status write, not a promotion. The maps above stay the frozen NARR proxy.

Status	cubes on disk; sign not recovered
Climatology	east about 2x west; spatial r = 0.80 vs proxy
Theil-Sen	west +0.0099, east +0.0050
Stage B	incomplete

Outbreak size

GPD diagnostic

Generalized Pareto fit to daily CONUS tornado-count exceedances (Tippett-style). Not the 95°W test.

Window	n	ξ	σ
1979-2017	631	0.112	12.22
1979-1997	307	0.024	10.13
1998-2017	350	0.133	13.62

GraphForge checks

Analysis contract (pass/fail)

GraphForge ran the cases below against the written 1979-2017 STP result and the analysis contract. Allow cases must pass. Refuse cases must block. This is a software check, not forecast skill.

Loading verification JSON…

Case	Expect	Result
…

GraphForge overview

Scope

What this study does not do

Reports, environments, and climate-model fields are kept separate. This page does not estimate future tornado counts, does not publish an operational 2050 tornado map, and does not attribute the eastward STP slope to greenhouse forcing. Gensini and Brooks left that attribution open.