Repository navigation
Increment 39: every SVAR sub-catchment of four Swedish SE2 river systems meshed, runner and record - #240
Merged
Conversation
…hment in one night (@architect) Increment 39, re-scoped on Ola's words of 2026-10-12 (no forecast-point list; "each (sub)catchment TINed by tomorrow morning"): every SMHI SVAR 2022 sub-catchment polygon of the Ljungan, Indalsälven, Ångermanälven (which holds Faxälven, Fjällsjöälven and Åseleälven) and Umeälven is meshed by `rasputin mesh` from the polygon as the outline, on GLO-30 with CORINE at 10 m, exactly as the Lagan quick-check case does. No delineation, so the geographic-DEM catchment question does not arise. No production code: a runner script, the fetch script and a README under docs/benchmarks/. Measured today (M1 Max, AC): 5,338 sub-catchments over 98,095 km², median 12.6 km²; 31 coarser BARO areas (1,413 to 4,315 km²); the water-body level is nearly the finest (5,179) and is not a coarser option. One unit at the Lagan settings: 2.35 s for the smallest (0.004 km², 15 triangles), 3.08 s for the median (12 km², 3,247 triangles, .gpkg), 4.22 s for the largest (250 km²), 7.42 s for a 1,413 km² BARO area (119,987 triangles); 2.1 to 2.6 GB peak, fixed. Projection: about 4.6 h serial for the finest level, under 20 min for BARO; the night runs BARO first, then the finest level with a pool of three, resumable, largest first. Data fetched into $RASPUTIN_DATA (not committed): the four systems' SVAR units, 169 MB in 243 s over 54 paged WFS requests with a generic User-Agent (the service breaks chunked responses above about 5 MB, so 100 units a page); the GLO-30 blocks for all four systems, 505 MB in 64 s (cache now 642 MB, 36 tiles needed, 4 were cached). Licence: CC BY 4.0 SE, "Källa: SMHI", checked on SMHI's terms page today. Prior art: SVAR 2022:1.2's delivery description (10 m mean grid from the 1 m national model, snapped to the land-cover grid; MAINDOWN links); the legacy tag has nothing on Sweden or SMHI (grep pasted). ROADMAP row 39 added. Five questions for Ola with defaults (tolerance 10 m, CORINE on, .gpkg only per unit, both levels, GeoPackages stay in scratch). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…, the 5,338 units launched (@Perf) docs/benchmarks/2026-10-13/se2-catchments/se2_batch.py is the runner the design describes (prepare, run --level baro|fine|huvud, report): one GeoJSON per polygon, BARO areas dissolved by BARONR (all 31 are one Polygon), a ledger row appended per finished subprocess, resumable by the ledger, --retry for refused/error rows, a per-run watchdog (--timeout, default 3600 s), peak memory per child from wait4, refusals in mesh's own words with the Rich box stripped, SVG maps from shapely's svg() (no matplotlib in the venv, as the design states) and a per-system GeoJSON for QGIS. Tested on a scratch root: ok rows, a refused row, resume, retry, the watchdog kill and report. Written by @Perf, not @developer as the design names: the harness gives docs/benchmarks/ to @Perf and keeps @developer out of docs/. The venv is this worktree's, built by tools/new_worktree.py --existing (master's C++; src_python, bindings, include unchanged from master). BARO level, one process at a time, AC power, 2026-10-12 16:24:55 to 16:36:30 UTC: 31 of 31 ok, 98,095 km2, 6,149,260 triangles, 366.4 s gpkg + 327.9 s vtk, no refusal, largest max_error_m 9.999995, peak 3.26 GB. Three outputs opened (sqlite3 triangle counts and run layer, vtk headers). Finest level launched detached at 16:37:16 UTC (pid 79982, nohup, launch_fine.sh, fine.log under ../rasputin_scratch/sweden/se2/), three at once, the four whole systems chained after it; 25 units ok in the first 41 s. README has the method, the versions, the BARO table, the launch record and the morning's report command. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
… recorded as Ola's rulings, Status and ROADMAP row to built and running, by the main session Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…ies and overview maps copied into the record, by the main session The detached batch finished 2026-10-12T18:05:23Z: 31 BARO areas, 5,342 fine units and 4 whole systems, 0 refusals. The report step (se2_batch.py report --copy-to) wrote report.md, summary_fine.csv, summary_huvud.csv and twelve overview SVGs (fine, BARO, whole system per river system) beside the README. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…imate reconciled, Status and ROADMAP row to the batch done, by the main session Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
skavhaug
enabled auto-merge
October 12, 2026 20:46
skavhaug
added a commit
that referenced
this pull request
Oct 13, 2026
…P rows 38 and 39 both kept, by the main session Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Increment 39,
docs/increments/39-se2-catchments.md. No production code (count_loc.py: 0); a batch run of the existingrasputin meshon SMHI SVAR 2022 polygons, with its runner and record underdocs/benchmarks/2026-10-13/se2-catchments/.fetch_svar.py(the SMHI WFS download),se2_batch.py(prepare, run per level with a resumable ledger, report),README.md,report.md, three summary CSVs and twelve overview maps. The meshes themselves (about 5 GB of GeoPackages) stay outside git.Source credit: SMHI, SVAR 2022, under the terms the README quotes. Review rounds 1 and 2 recorded in the increment file.
🤖 Generated with Claude Code