Skip to content

SKILL.md

Latest commit

 

History

History
192 lines (150 loc) · 8.16 KB

SKILL.md

File metadata and controls

192 lines (150 loc) · 8.16 KB
Raw
name verto-explorer
description Guided coordinate conversion between Italian reference systems via the official IGM Verto Online service, using the openverto CLI. Use this skill whenever the user needs to transform, reproject, or identify the CRS of coordinates in Italy — even if they don't mention IGM, Verto, or EPSG by name. Triggers include: converting points between Gauss-Boaga, UTM-ED50, IGM95, RDN2008/ETRF2000 or geographic datums (Roma40, ED50, ETRS89); reprojecting a CSV or GeoJSON of Italian coordinates; "what CRS is this file?"; coordinates that "land in the sea" or far away (wrong EPSG / swapped axes); certifying a datum chain is lossless before publishing open data. Guides the user step by step: identify the source system, pick a valid target (different datum), convert, and verify coverage and precision.
license MIT
compatibility Requires the openverto CLI (openverto systems, convert, batch, geojson, inspect, detect, targets, roundtrip, cache, doctor).
metadata
author version
ondata
1.1

Verto Explorer — Guided Coordinate Conversion for Italy

This skill uses the openverto CLI to convert coordinates between every Italian reference system through the authoritative IGM Verto Online service, with no NTv2 grid files installed locally.

Every openverto command supports --help:

openverto --help
openverto convert --help
openverto batch --help

A global -o/--output flag controls structured output: table (default), json, jsonl (one compact object per line — ideal for coordinates), csv. Use it whenever you need to parse results instead of reading a Rich table:

openverto -o json systems
openverto -o jsonl convert --from 4265 --to 6706 12.4924 41.8902
openverto -o csv batch points.csv --from 3003 --to 6707

In structured modes stdout is pure data and errors go to stderr. Exit codes: 0 ok, 2 usage / out-of-grid input, 3 not found, 5 API error.

Conventions to apply (read first)

  • Axis order is east-first, always: e = est (easting OR longitude), n = nord (northing OR latitude). This is the service convention and is the reverse of the EPSG-registry lat/long axis order for geographic CRS like 4265, 4230, 4670, 6706. When a user gives "lat, lon", pass lon as e, lat as n. GeoJSON positions are [x, y] = [e, n], so they already match — no swap when reprojecting GeoJSON.
  • Geographic values are always decimal degrees (gradi sessadecimali), never degrees-minutes-seconds. Convert DMS to decimal first.
  • Same-datum conversions are rejected (e.g. 6706 → 6707, both RDN2008). Always cross a datum family boundary; check with targets.
  • detect recovers the projection/zone, never the datum: magnitude alone cannot tell Roma40 from ED50/IGM95/RDN2008. Take the datum from the dataset's metadata, not from detect.
  • EPSG 4806 (Monte Mario, Rome) measures longitude from the Monte Mario prime meridian (~12.45°E from Greenwich), so a point near Rome has e ≈ 0, not ~12.5. Do not confuse it with the Greenwich-based 4265.

Follow the four phases in order. The goal is correct, verified conversions — not to fire a request blind and risk points landing in the sea.

Phase 1 — Identify the source reference system

Coordinates are useless without their CRS. If the user states the EPSG, confirm its meaning; if not, recover it.

# vague label ("UTM", "Gauss-Boaga") or unknown magnitude → guess the system
openverto -o json detect 2300000 4640000
# → kind=projected, candidate_epsg=[3004] (Gauss-Boaga fuso Est), confidence=medium

# confirm datum family, axis order (e=est/lon, n=nord/lat), unit, zone, false easting
openverto -o json inspect 3004

Key facts to apply:

  • e is always est (easting OR longitude), n is nord (northing OR latitude). Geographic coordinates are ALWAYS decimal degrees.
  • Look-alikes to disambiguate: 3003 vs 3004 (Gauss-Boaga West/East), 23032 vs 23033 (ED50 UTM 32/33), 6707/6708 (RDN2008 TM32/TM33).
  • If detect says the point is outside Italy's bounds, the conversion will be rejected — re-check the EPSG and whether e/n are swapped.

Phase 2 — Pick a valid target

The service rejects conversions between systems of the same datum (e.g. RDN2008 2D geo → RDN2008/TM32). List only valid destinations:

openverto -o json targets 3003   # excludes the Roma40/Monte Mario family
openverto systems                # full list of the 20 supported systems

For Italian open data today, the official target datum is RDN2008 (ETRF2000): geographic 6706, or projected 6707/6708/6709 (TM32/33/34).

Phase 3 — Convert

# one or a few coordinates (e first, n second)
openverto -o jsonl convert --from 4265 --to 6706 12.4924 41.8902

# many coordinates piped as 'e,n' lines
printf '12.49,41.89\n13.0,42.0\n' | openverto -o jsonl convert --from 4265 --to 6706

# a whole CSV → CSV (original columns + e_out/n_out) or GeoJSON.
# catasto.csv has columns id,est,nord (Gauss-Boaga Ovest = 3003);
# points.csv has columns e,n (also Gauss-Boaga Ovest = 3003) — auto-detected.
openverto batch catasto.csv --from 3003 --to 6707 --e-col est --n-col nord --out out.csv
openverto batch points.csv  --from 3003 --to 6706 --format geojson --out out.geojson

# Italian-style CSV (';' delimiter, comma decimal), piped from stdin to stdout
cat comuni.csv | openverto batch - --from 4265 --to 6706 --decimal , --e-col lon --n-col lat

# reproject the geometries of a GeoJSON file
openverto geojson aree.geojson --from 4230 --to 6706 --out out.geojson

batch auto-detects common column aliases (est/nord/lon/lat/x/y) and auto-chunks at the 32000-coordinate service cap. Results are cached for offline replay.

CSV input (read with DuckDB). Default format, no option needed: a header row and a dot decimal separator (12.4924); the delimiter (,, ;, tab, |) is auto-detected — never pass it. For Italian-style CSVs with a comma decimal (12,4924), add --decimal ,. Pass - as the file to read from stdin; omit --out to write to stdout (combine with -o jsonl/-o csv for clean piped data).

Output field names (verified — parse these directly):

  • convert rows: in_e, in_n, out_e, out_n.
  • batch in -o json/jsonl/csv: the original columns plus e_out, n_out (skipped rows are omitted). In file mode (--out, default table) it writes the original CSV columns plus e_out, n_out, or a GeoJSON FeatureCollection.
  • detect: kind, confidence, candidate_epsg, candidate_names, plausibly_in_italy, reason, note.

Phase 4 — Verify coverage and precision

The IGM grid covers Italy and its surrounding seas only. A point at sea near Italy is valid; a point in another country or open ocean is rejected with stato: errore, dove: Proj, "Coordinate to transform falls outside grid".

  • A single out-of-grid coordinate fails the entire batch request. Isolate and skip the offenders instead of failing the whole file:

    openverto batch mixed.csv --from 4265 --to 6706 --skip-invalid --rejects bad.csv

    The rejected rows go to bad.csv; inspect them — they are usually a wrong EPSG, swapped axes, or genuinely-foreign points.

  • Before publishing, certify the datum chain is lossless within tolerance:

    openverto -o json roundtrip --from 3003 --to 6707 1500000 4640000
# reports residual_m per point and max/mean residual in metres (A→B→A)

  • Check connectivity / cache when something looks off:

    openverto doctor
openverto cache --stats
openverto cache --clear   # force fresh conversions

Quick reference — the 20 supported systems by datum family

  • Roma40 / Monte Mario: 4265 geo, 4806 geo (Rome meridian), 3003 Gauss-Boaga Ovest, 3004 Gauss-Boaga Est.
  • ED50: 4230 geo, 23032/23033/23034 UTM 32/33/34N.
  • ETRS89 / IGM95: 4670 geo, 3064/3065/9716 UTM 32/33/34N, 3035 LAEA, 3034 LCC.
  • RDN2008 (ETRF2000) — current official Italian datum: 6706 geo, 6707/6708/6709 TM32/33/34, 7794 Italy Zone (E-N), 6876 Zone 12.

Conversions are only allowed across these families, never within one.