p.rank.cross ingests multiple p.rank JSON reports — one per region or
per mission variant — and produces a single ranked table of the best
candidates across all of them. It is the natural top-of-pipeline step when
the same toolkit is run independently on N candidate regions (e.g. the
NASA Artemis III shortlist of ~13 polar regions) and a unified ranking is
needed for mission planning.
The composite score per candidate is a weighted combination of three normalised components:
composite = w_suit · norm(suit_mean)
+ w_area · norm(transform(area_km2))
+ w_borda · norm(borda_score)
where borda_score is an intra-region rank-based score (top candidate in
a region with K candidates gets K points, …, last gets 1), optionally
multiplied by the candidate's Monte-Carlo rank1_probability (from
p.rank) so that weight-uncertainty information from the per-region
analyses survives the cross-rank step.
The area transform defaults to log(1+area) which compresses the
typical orders-of-magnitude spread between small inter-crater patches
and ellipse-sized safe regions; linear and sqrt are also supported.
| Parameter | Default | Description |
|---|---|---|
reports |
required | Comma-separated list of p.rank JSON report files |
region_ids |
report stems | Override region labels (must match reports order) |
weights |
suit:0.6,area:0.25,borda:0.15 |
Composite-score component weights; renormalised to sum=1 |
area_transform |
log |
linear, sqrt or log (log1p) |
n_top |
20 |
Maximum candidates retained in the cross-ranked output |
output |
cross_ranking_report.json |
Output JSON path |
The output JSON has the schema tag p.rank.cross/v1 and contains:
weights, area_transform, n_regions, n_candidates, n_returnedregions[]: per-region metadata (id, source report path, candidate count)candidates[]: top-n_top entries, each carrying cross_rank,
region, intra_region_rank, composite_score, the original
suit_mean, suit_std, area_km2, rank1_probability, and the
three normalised component scores under components.
p.rank.cross reports=region_a_report.json,region_b_report.json,region_c_report.json \
output=cross_ranking.json
p.rank.cross reports=r1.json,r2.json,r3.json \
weights=suit:0.4,area:0.5,borda:0.1 \
area_transform=linear output=area_first.json
# Run p.landing once per region with separate mapsets, then aggregate p.rank.cross \ reports=$(ls ~/grassdata/Moon_SouthPole_5m/artemis_*_report.json | paste -sd,) \ region_ids=connecting_ridge,de_gerlache,malapert_massif,nobile_rim,mons_mouton \ n_top=30 output=artemis_shortlist.json
All input JSON files must have been produced by the same version of p.rank because the schema is validated for required keys. Per-region normalisation is applied before cross-region aggregation, so absolute suitability values need not be comparable across regions. The output JSON can be passed directly to g.gui.landing for visualisation.
p.rank, p.landing, p.mcdm.score
Yann Chemin
Available at: p.rank.cross source code (history)
Accessed: Sunday Nov 15 11:06:05 2026
Main index | Paint index | Topics index | Keywords index | Graphical index | Full index
© 2003-2026 GRASS Development Team, GRASS 8.5.1dev Reference Manual