Research page: Simulations of April 19 Outbreak

Summary

8/24/00: G-S PBL and Blackadar higher-res terrain runs done (both at 3 km). Higher-res terrain has interesting differences in initiation (more cells, more linearly aligned at start) but evolution looks even less like the real event, so will document it and set it aside for now. Have started 1 km run now; decided to stick with 250x250 inner grid.

SLS conference preprint done. Snapshot at 26 hours (1560 min) consistent with cold front aloft. Surface drytrough/pacific front look merged, but earlier check suggested they were separate; need to revisit later.

At convective initiation time, ascent at dryline coincident with start of cool air aloft. Notable 500 mb RH gradient appears partly/largely driven by vertical advection with deep pattern of ascent/descent at leading edge of CFA. Whether this is related to dryline or is truly CFA-induced vertical motion consistent with frontogenesis is not yet diagnosed. Convective initiation occurs right after passage of 600 mb theta-e ridge/onset of cooling, though, which can't be a coincidence.

See current work under "Overview" below.

Blackadar run (#398) 3-km results with original terrain are here, and the results with 9-km terrain is here. Full suite of 1-km fields (until model blew up, CFL) from run 405 are not yet online, but partial results are here, including animation.
8/23/00

Overview

1. Blackadar 4-grid run at 3 km (run 398) run out - central IL convection lines out. Run stable, though. 5th grid will likely be run from this, as before, but with half the former time step.
2. G-S PBL 4-grid run blew up, CFL on grid 3. The convective area is so big that the part outside of grid 4 apparently killed off the run. I'll have to turn down the time step (probably pointless) or, better yet, rerun with a larger grid 4 (doable with restart at 18 hours) to continue it. Run made it most of the way: 25.3 hours.
3. Blackadar 4-grid run with more detailed (still only 9 km) topography is still being run; at last check, deep convection not yet started. Need to see this before doing 5-grid case to see if topo/landuse makes any difference with physics being used.

• 470e: 5 grids, 300x300, really use 298, run 6 steps.
• 401n: redo 401 with dt=90. Note normalized runtime = 4:15
• 401o: queued.
• 398 done.
• 401 done unless rerun with larger grid 4 (3 km).
• 399 (terrain) continuing.

Attributes

• Up to 5 grids, with horizontal resolution of 81/27/9/3/1 km.
• Grid layout is shown here. Domain 5 is 250x250, with 1-km horizontal resolution, and is movable.
• 3 grids have been (recently) run; inner dx=9 km, with 66 layers (~60 m vertical resolution near ground). All 3 grids run from start.
• Explicit precipitation (only) used on grids 3,4,5
• Nudging: analysis only, outer 2 grids only.
• Grell CUPA used on outer two grids.
• PBL: Gayno-Seaman or Blackadar (for now)
• Dt: varies.

Comments, run summary to date

• 395n: run 2 steps, G-S PBL, plotting: new grid layout; move G5 X-=25
• 395o: run 2 steps, G-S PBL, plotting: with FDDA enabled; ok
• 395p: run 10 steps, G-S PBL, no plotting: timing for 2 grids, dt=30; elapsed 4:31/5:47 run 25:20/32:28
• 395q: run 12 steps, G-S PBL, no plotting: timing for 3 grids, dt=40; real 9:56 user 2:19:43
• 396a: run 2 steps, BL PBL, plotting: try blackadar, FDDA
• 396b: run 10 steps, BL PBL, no plotting: timing for 2 grids, dt=30; elapsed 2:35/3:19 run 15:30/19:52
• 396c: run 12 steps, BL PBL, no plotting: timing for 3 grids, dt=40; real 8:09 user 1:52:53
• 395: run 4 grids (3 km res), G-S, to 24 hours.
• 396: run 3 grids (9 km res), Blackadar, to 24 hours.
• 397: repeat 395, G-S PBL, with dt=90. Done through 24 hours w/3 grids.
• 398: repeat 396, Blackadar PBL, with dt=90. Done thru 24 hrs w/3 grids.
• 331b looked pretty good!!

Conclusions...so far...

• Real way to compare model fields to observations is to first interpolate model (e.g., sfc) grid data to sao locations, and to do a comparable analysis on that. E.g., the 20 C dewpoint isodrosotherm on Run 395 (G-S) falls between sfc reporting stations, so it may be closer to the 'truth' than we think.
• Need closer look at upperair we are nudging towards at 00z. The mm5 fields look too weak w.r.t. verification and this may be contributing to higher pressures (and less warm frontal troughing?) in the model. Need to examine jet streak structure and implied left-exit region ascent.
• Also plot sfc convg, mconv?
• May need larger (x) grid-4, to allow grid 5 to initially cover convective development and be moved to follow mature cells eastward.
• Known surface problems: too warm (1 deg C or so), not moist enough (close, though, within 1 g/kg), sfc winds too backed (south of warm front), MSL pressures too high (by 4+ mb), winds aloft a bit too weak (5 to at most 10 kts at 850, 700 mb). These comments refer to west-central IL, nne of STL.
• G-S run with 3 km resolution has little convection.