Reframe the Science section headings and add study scope

[pkjha-aero]

Summary

Two wording changes in Science with MLAP, each with the surrounding prose updated so the text matches the new framing rather than only the title.

"Not yet studied" → "Potential future studies"

Applied to both the overview and the physical quantities page. The bodies read as an inventory of gaps — "have no simulation results yet", "the runs have not been made" — which undercut the new heading, so they were rewritten as a research agenda.

Each candidate now states the question it would answer and points at the mechanism that already supports it, making clear these need runs rather than new code:

Study	Question it would answer
Terrain ruggedness	Does slope and variability carry signal beyond height alone?
Month subsets	Do season-specific models beat one year-round model?
Region subsets	Does a model tuned to one landscape outperform a statewide one?
Cross-application in time	How well does a model trained on one period predict another?
Cross-application in space	Does a California-trained model transfer elsewhere?

The two cross-application studies are flagged as most consequential, since the case for MLAP as a cheap surrogate rests on generalization beyond fitted conditions — currently argued from design rather than measured. That is an editorial judgment, easily softened if you would rather not foreground it.

The physical quantities section now carries the supporting evidence: elevation's +0.007 gain held to within 0.0002 across all four datasets, which is what makes ruggedness worth testing. It also notes that PREC looks marginal year-round but California rain is winter-concentrated, so its contribution may be larger in a seasonal subset — linking the two future studies together.

"What is assessed" → "Assessment performed"

The new title claims work done rather than questions posed, so the table gained a Scope column to back that up:

Study	Scope
Data Sampling	7 datasets, 2–15 million rows
Historical Data	9 datasets; 32–48 h max history, 1–8 h interval
ML Parameters	12 hyperparameter studies, each across 3 datasets
Physical Quantities	3 feature studies across 4 datasets

Also states the reproduction rule the pages already follow but never declared up front: studies were repeated across multiple datasets, and effects that do not hold across all of them are reported as unestablished rather than as findings.

Verification

Scope figures were counted, not estimated — dataset identifiers were taken from each page's own tables, and 27 evaluation collections plus the 1–15 million row range were confirmed against the source data.

Test plan

• mkdocs build --strict passes with anchor validation active
• Scope figures cross-checked against the individual science pages
• Elevation +0.007 / ±0.0002 claim re-verified against its own table
• No internal link targeted either renamed anchor

🤖 Generated with Claude Code