What Should South Dakota Know?

The challenge

Find a real environmental decision and identify information that is usually missing from the map used to make it. Your project should combine at least three CubeDynamics nouns from at least two noun families, use reusable verbs, show a source-level QA figure, and end with one decision-oriented figure.

Start with the decision

Write one question that a planner, land manager, conservation practitioner, water manager, rancher, emergency manager, or environmental scientist could act on. Then write what a conventional single-layer map fails to show.

Do not begin by browsing for the most dramatic-looking data.

Check the vocabulary before coding

from cubedynamics import data

data.list_sources()          # every noun/source flavor implemented today
data.describe("temperature", "prism")

If a noun is absent from data.list_sources(), it is not a public loader. Add it to your dependency list instead of writing a pretend call. At publication time, the implemented families include gridded climate/weather nouns and Sentinel-2 surface observations. Combining families may require an explicit, scientifically justified alignment step.

An API-current starter

This compact two-noun climate sentence uses actual current interfaces. It is a starter, not a complete three-noun submission:

from cubedynamics import data, pipe, verbs as v

temperature = data.temperature(
    source="prism", statistic="maximum", bbox=aoi,
    start="2024-07-01", end="2024-07-31",
)
precipitation = data.precipitation(
    source="prism", bbox=aoi,
    start="2024-07-01", end="2024-07-31",
)

warm = pipe(temperature) | v.quantile_state(
    quantile=0.75, direction="above"
)
dry = pipe(precipitation) | v.threshold_state(
    threshold=0.1, direction="below"
)

result = (
    pipe(warm.unwrap())
    | v.overlap(dry.unwrap())
    | v.mean(dim="time", keep_dim=False)
).unwrap()

v.overlap is for already aligned state rasters and refuses silent coordinate alignment. It is not a vector-intersection function. A possible third public noun is data.vegetation_index(..., source="sentinel2"), but you must resolve its different grid, observation dates, clouds, and scale explicitly before combining it. Do not paste it into v.overlap and hope the coordinates match.

Your analytical sentence

Keep the logic short enough to read without surrounding prose:

NOUN A→GENERAL VERB→NOUN B→SUMMARY

Explain verbs scientifically: what state, threshold, alignment, change, or summary did each one define? Avoid narrating ordinary Python mechanics.

QA · Check the nouns before trusting the sentence

Show at least one source-level QA figure before the result. Good evidence includes a point time series plus a spatial slice for climate, early/recent panels for change, category maps and counts for polygons, or footprint/network coverage and density sanity checks. Add numerical checks for dimensions, units, missingness, physical ranges, dates, CRS, and exact alignment.

Standard submission pattern

QUESTION↓NOUNS + SOURCE FLAVORS↓PIPE↓VISIBLE QA↓DECISION MAP↓INTERPRETATION + LIMITATIONS

A strong submission answers: What could a real decision maker know or do differently after seeing this? It also says what the result cannot establish.

Completion checklist

Decision question and why someone might care
Missing information
At least three public nouns and their source flavors
At least two noun families
Short pipe using reusable verbs
Visible source QA and numerical sanity checks
One decision-oriented figure
Careful interpretation and explicit limitations
Reproducible environment, bounded AOI/time, and tests

Fork this question

Change the AOI while preserving the decision logic.
Swap one source flavor and compare source assumptions.
Replace one noun only if the new noun changes the question coherently.