Sources & caveats
Everything on this site traces back to one peer-reviewed study. This page is the ledger: where the numbers come from, what the words mean, what the model does not cover, and the record of every question raised and resolved.
The study
The paper is the authority; FLAGRI is a working interpretation of one bounded slice. Every constant carries its citation to the paper's Table 2; land answers reproduce Table 3; canola answers reproduce Tables 4–6. FLAGRI is not a forecasting or planning tool — it makes a published analysis explorable.
The name
Daily Intake
Daily Intake is what one person consumes per day, on average: 8,686 kJ of food energy and 81 g of protein — the study's figures, "based on what is currently consumed as per a national nutrition survey" (§2.1, Table 2). Two honesty notes follow from that definition. First, these are current consumption levels held steady, not survival figures — the land answers keep everyone eating as they do today, which makes them deliberately comfortable minimums of land, not a diet plan. Second, Daily Intake does not yet cover everything a body needs: fat, micronutrients and drinking water are outside the model so far. Land is worked out for each intake separately and whichever needs more land decides the answer — protein is what makes potatoes need extra land.
Engine & fidelity
The calculation engine (calcCrop / calcCanola / calcBiodieselEquiv) is ported from the published FLAgri tool. At the paper's settings it reproduces: Table 3 land (116,998 / 84,011 / 642,062 ha), Table 4 canola (wheat 4,398 + 227 = 4,625 ha; dairy 31,522 ha), Table 6 percentages (0.9 / 2.2 / 6.8% of grain land), the annual requirement (16.2 trillion kJ; 151.2 billion g), and biodiesel yield ≈1,227.7 L-equivalent/ha.
Resolved question (22 Jul 2026): the tool's dairy canola base originally sat ≈0.15% above the paper's Tables 4–5. Traced to display rounding: the paper's tables were computed with the unrounded dairy diesel rate (54.913 L/ha = 95 M L ÷ 1,730,000 ha) while Table 2 displays the rounded 55. Agreed with the study authors, the engine now uses the unrounded value and matches Table 4 exactly; the dairy loop figure sits within ±1 ha of Table 5 (trip-count rounding).
The recursive fuel loop ("fuel to grow the fuel") is counted once — ≈+5% at baseline, rising with severity — a first approximation, as in the paper (p.7).
Framing rules & what is not modelled
- One food at a time: wheat OR potatoes OR dairy — pathways are alternatives, never added together. Real diets want variety, and variety costs more land and fuel.
- Minimums of land, holding today's Daily Intake — all figures are floors, not plans.
- One slice of readiness: fuel is one choke point of several. Spare parts, fertiliser and agrichemicals, competing claims on the same diesel, processing energy beyond the factory gate, and growing seasons / time-to-first-harvest are all outside the model — and the paper says so.
- "Consumption", not "supply" — the paper's own wording for national diesel figures.
- Scenario severities (baseline / 5 Tg −7.9% / 150 Tg −61%) follow Xia et al. (2022) as used in the study; yield reductions are applied uniformly.
Status & contact
FLAGRI is a working draft — figures are under review with the study authors, and the draft ribbons stay up until the authors call it validated. We have checked it carefully but cannot guarantee complete accuracy. Questions, corrections and ideas are very welcome: sam@rongo.co.nz. Hosted by Rongo.