CHANGELOG.md

@neaps/tide-predictor

0.9.0

Minor Changes

• #220 c1f0144 Thanks @bkeepers! - Filter spurious extremes from tide predictions using prominence threshold and minimum temporal gap criteria.

Patch Changes

• #234 6feeca0 Thanks @bkeepers! - Align timeline predictions to clock boundaries based on timeFidelity. For example, with the default timeFidelity of 600 seconds, predictions now fall on :00, :10, :20, :30, :40, :50 past the hour regardless of the requested start time. The start time will always snap to the previous clock boundary, and the end time will snap to the next clock boundary.

• #257 ccb662f Thanks @bkeepers! - Fix incorrect division for milleseconds in JD function (Thanks @dartheditous)

• #253 764b8c0 Thanks @bkeepers! - Update T3, R3, 3N2, and 3L2 constituent definitions from TICON manual

0.8.0

Minor Changes

• #227 b3efa7c Thanks @bkeepers! - Add support for timeline predictions with offsets.

  getTimelinePrediction now accepts an offsets option for subordinate stations, using proportional domain-mapping to interpolate between reference station extremes with time and height adjustments.

0.7.0

Minor Changes

• #213 9f3fdf6 Thanks @bkeepers! - Speed up extremes detection by ~100x, deprecate timeFidelity option on getExtremesPrediction, which will always be <1s now.

0.6.0

Minor Changes

• #208 355f696 Thanks @bkeepers! - Replace constituent definitions with the IHO TWCWG (International Hydrographic Organization Tidal and Water Level Working Group) constituent list.

  There are lot of implementation details that changed, but the highlights are:

  1. Switched from ~60 hand-coded constituents to 395 IHO standard constituents (6.6x increase).
  2. Switched to IHO standard nodal correction formulas described in Annex A, which use simplified Fourier series formulas and are the International standard used by hydrographic offices worldwide.
  3. Implemented the IHO Annex B rules for resolving compound constituent members. This allows for much more comprehensive tidal predictions that include minor constituents that can have significant local effects.

  Benchmark Results

  Most importantly, the new implementation delivers significantly improved accuracy when comparing predictions to NOAA, with a 60% reduction in median height error (22.3mm to 8.7mm) and a 60% reduction in timing error (15 min to 6 min) at the 95th percentile.

  Metric	Before	After	Improvement
  Median height error (MAE p50)	22.3 mm	8.7 mm	↓ 60.9%
  95th percentile height error (MAE p95)	45.4 mm	23.9 mm	↓ 47.4%
  RMSE (median)	25.4 mm	10.3 mm	↓ 59.6%
  Median timing error (p95)	15 min	6 min	↓ 60.0%

• #212 ebb5dc0 Thanks @bkeepers! - Export constituents, deprecate default export

  -import tidePredictor from "@neaps/tide-predictor";
  +import { createTidePredictor } from "@neaps/tide-predictor";

0.5.0

Minor Changes

• Add constituents to support TICON data #186

Patch Changes

• Add MP1 constituent