Distinguish the two ways of asking for daily data

[chross22]

The section introduced frequency = "daily" and dates one after the other without saying they answer different questions, so it read as two routes to the same place with dates merely cheaper.

The question that prompted this — whether a date range still returns every day — is the one the section failed to answer. It does, and now says so.

Want	Use	Gives
Every day in a period	frequency = "daily" with years and months	a continuous series
Particular days	dates	only those dates

Each gets a heading. Every day in a period is for a continuous series — a time series at one station, an animation, anything where the gaps between days would matter. Particular days is for matching observations.

Thinning a long record

Documented for the first time. It was implied by "only the days that matter are downloaded" but never shown, and it is the answer for anyone who wants a decade without 4,017 downloads:

# Weekly through a decade: 574 downloads rather than 4,017
accessEnvDat(vars = "SST", bounding_box = bb,
             dates = seq(as.Date("2005-01-01"), as.Date("2015-12-31"), by = "week"))

The counts are checked, not estimated

seq(..., by = "week") over 2005–2015 gives 574 dates against 4,017 days, and parse_dates() takes the Date vector seq() returns unchanged. I had first written 4,018 — an off-by-one on an inclusive date range — and corrected it after running the arithmetic.

Also verified the underlying claim with a live fetch before writing any of this: frequency = "daily" with years = 2015, months = 2 returns all 28 days of February, 1,344 rows, detected as day.

Verification

R CMD check — Status: OK. Full suite green. All README anchors re-checked including the two new subsections.

🤖 Generated with Claude Code