Release Notes — v0.7.2¶
CLAS PPP-RTK accuracy/robustness fixes and mrtk dump --parse¶
Release date: 2026-06-29 Type: Fix + feature (CLAS PPP-RTK accuracy/robustness + CSSR tooling) Branch: release/v0.7.2
Overview¶
v0.7.2 collects two CLAS PPP-RTK correctness fixes and adds a CSSR dump tool for inspecting decoded CLAS L6D / Compact SSR:
- The PPP-RTK adaptive-ionosphere process noise is restored under storm-day conditions — a lost
Qpscatter-back meant the adaptive noise was never actually applied. Storm-day CLAS fix rate improves from 77% to 85% (+8 pp). - CLAS now falls back to global orbit/clock corrections when the close-network partition goes stale, instead of dropping corrections.
mrtk dumpgains a--parsemode that reproduces the upstream claslibdumpcssrper-subtype CSV output byte-identically.
The two fixes are CLAS PPP-RTK behavior changes with an accuracy/robustness effect; the dump tool is a self-contained parser that leaves the production CLAS decoder untouched. F5 (GSI daily-coordinate) absolute-accuracy regression tests are added to guard geodetic accuracy going forward.
1. Bug fixes¶
PPP-RTK adaptive-ionosphere process noise¶
The compacted→full-nx Qp scatter-back had been lost, so the adaptive ionosphere process noise was never applied to the filter. Restoring the scatter-back makes the adaptive iono noise take effect again.
Storm-day CLAS fix rate improves from 77% to 85% (+8 pp) (#225, PR #226).
CLAS stale network partition¶
When the close-network orbit/clock partition goes stale, CLAS now falls back to the global orbit/clock corrections rather than dropping the corrections entirely. This keeps a solution alive through a stale-partition window (#218, PR #228).
2. What's new¶
mrtk dump --parse¶
mrtk dump gains a --parse mode that dumps decoded CLAS L6D / Compact SSR to the upstream-format per-subtype CSV files (parse_cssr_type1..12*.csv and parse_cssr_header.csv).
It is implemented as a self-contained vendored parser (apps/dumpcssr/cssr_parse.c) ported from claslib 082; the production CLAS decoder is left untouched. The mode requires -grid FILE.
mrtk dump -grid/--grid FILE¶
The CLAS grid definition is now passed directly on the command line via -grid (alias --grid), replacing the previous -k/--config file indirection.
3. Changed¶
mrtk dumpgrid input — the grid definition is taken via-gridinstead of the-k/--configfile (which existed indumponly to read the grid path). The redundant config handling is removed.
4. Tests¶
CLAS absolute-accuracy regression tests are added: F5 (GSI daily-coordinate) geodetic-truth checks for all CLAS PPP-RTK/VRS datasets (TSUKUBA3 / station 0627), guarding geodetic accuracy in addition to the existing consistency-vs-claslib checks (PR #227).
5. Validation¶
mrtk dump --parse byte-identity. The --parse output is byte-identical to upstream claslib 082 dumpcssr / ssr2osr -dump on 2026121A.l6 across all per-subtype CSVs (data fields); the only difference is an upstream version delta in one header column label (PR #230).
CLAS PPP-RTK absolute accuracy. 2D 95% ≈ 4.2–4.3 cm vs GSI F5 truth (TSUKUBA3, 2019/08/27).
Storm-day fix rate. CLAS PPP-RTK storm-day fix rate rises from 77% to 85% with the restored adaptive-iono Qp scatter-back.
Compatibility & migration¶
The mrtk dump grid input changes from the -k/--config file to a direct -grid FILE argument; scripts that relied on the config-file indirection for the grid path must pass -grid instead. The PPP-RTK adaptive-iono and stale-partition fixes are corrections to existing CLAS behavior and require no configuration changes.
Issues / PRs¶
- Release PR #231.
- #225 / PR #226 — restore adaptive-iono
Qpscatter-back (storm 77%→85%). - #218 / PR #228 — global orbit/clock fallback on stale network partition.
- PR #227 — F5 absolute-accuracy regression checks.
- PR #230 —
mrtk dump --parsebyte-identity validation.
References¶
- claslib 082 —
dumpcssr/ssr2osr -dumpper-subtype CSV format (parse_cssr_type1..12*.csv,parse_cssr_header.csv). - IS-QZSS-L6 — CLAS L6D Compact SSR message format.
- GSI F5 daily coordinates — geodetic-truth reference for the CLAS absolute-accuracy regression tests.