SeaRise NZ

Goal

Read the official NZ SeaRise dataset published on Zenodo: vertical land movement rates for 8,173 coastal sites and sea-level rise projections to 2300 for 8,179 site ids spaced every 2 km along the NZ coastline, by SSP scenario and percentile.

Use this when

• A task needs location-specific NZ sea-level rise projections (e.g. "how much sea-level rise should Petone plan for by 2100 under SSP2-4.5?")
• A task needs coastal subsidence/uplift (vertical land movement) rates
• A task needs relative sea-level rise that accounts for land movement

Do not use this for

• Live tide levels or storm surge (use nz-tides-surf, gwrc-hilltop-nz)
• Coastal sensitivity classification layers (use niwa-coastal-nz)
• Regional inundation map overlays (use wcc-arcgis-nz for Wellington)

Commands

python3 skills/searise-nz/scripts/cli.py record --json
python3 skills/searise-nz/scripts/cli.py sites --near "-41.29,174.78" --limit 5
python3 skills/searise-nz/scripts/cli.py vlm 2503 --json
python3 skills/searise-nz/scripts/cli.py projections 2503 --scenario SSP2-4.5 --confidence medium --json
python3 skills/searise-nz/scripts/cli.py projections 2503 2504 2505 --year 2100 --json

Site 2503 is the projection site nearest Wellington CBD (0.48 km); always resolve an id with sites --near lat,lon rather than assuming one.

• record [--json] — Zenodo DOI, version, publication date, title, creators, licence, and file inventory with byte sizes, checksums, and download URLs
• sites [--near lat,lon] [--limit N] [--json] — projection sites, nearest-first when --near is given (downloads a ~674 KB CSV)
• vlm SITE_ID [--json] — vertical land movement rate, uncertainty, and quality for one site
• projections SITE_ID [SITE_ID ...] [--scenario SSPx-y.z] [--confidence low|medium|all] [--year YYYY] [--no-vlm] [--json] — sea-level projections for a 2005 baseline plus decadal 2020–2300 steps, with 17th/50th/83rd percentiles

Notes

• Licence: CC BY 4.0 — cite the dataset as Hamling, I., Naish, T., Levy, R. et al., New Zealand Vertical land movement and sea rise projections (Zenodo). The associated paper is Naish, T. et al. (2024), "The significance of vertical land movements at convergent plate boundaries in probabilistic sea-level projections for AR6 scenarios: the New Zealand case", Earth's Future — cite the dataset, not just the paper
• The dataset straddles two Zenodo versions on purpose. Projections come from v4 (14722058, Jan 2025), the current release; site details come from v3 (11398538, May 2024), the newest version that still publishes the site/VLM file. record reports both.
• 8,173 sites carry location and VLM detail, but the projection tables cover 8,179 site ids — a handful of ids have projections with no site-details row, so vlm can legitimately fail for an id projections accepts
• projections downloads and scans one ~58 MB CSV per invocation (about 20–35 s), regardless of the number of site IDs. The CSV is iterated once and is not persisted or cached; all downloaded data remains ephemeral.
• Scenario, confidence, year, and --no-vlm apply to every requested site. Filters reduce output, not transfer size.
• One requested site preserves the original response shape. Multiple requested IDs return requested_site_ids plus ordered sites results. Duplicate IDs are repeated in input order. Each result has status: ok, empty when the site exists but filters select no rows, or not_found when the source has no rows for that ID.
• Scenarios: SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5; values are metres relative to the 1995–2014 baseline. --scenario accepts either spelling (SSP3-7 and SSP3-7.0 both work)
• Low-confidence projections exist only for SSP1-2.6 and SSP5-8.5; the other three scenarios are medium-confidence only. --confidence low on those returns an explicit empty result, not a silent zero
• Negative VLM is subsidence and adds to relative sea-level rise; the default projections file already includes VLM (--no-vlm for climate-only)
• v4 values carry 3 decimal places against v3's 2, and add a zeroed 2005 baseline row; the underlying projections are otherwise unchanged
• Invalid input and upstream/schema failures are structured JSON on stderr with a non-zero exit code. Missing sites in a multi-site request are data results, not command failures.

Resources

• CLI entrypoint: scripts/cli.py
• Contract test: scripts/test_contract.py; smoke test: scripts/smoke_test.py
• Dataset provenance, column layout, and the Takiwa map relationship: references/source-notes.md