Fertiliser estimate

Nitrogen requirement per bed area

N [g/m²] = N [kg/ha] / 10

1 kg/ha = 0.1 g/m². The value is published per crop at a stated average yield: tomato 225 kg N/ha at 600 dt/ha, i.e. 22.5 g N/m².

Phosphorus and potassium requirement

A = Y × c;  requirement = A + Z, not below zero

The adjustment Z is set by the soil supply class: classes A and B add 60 kg P₂O₅/ha and 75 kg K₂O/ha, class C adds nothing, class D subtracts half the requirement and class E all of it. A negative result is set to zero, as the source states.

Liming dose per bed area

kg CaCO₃/m² = D [dt CaO/ha] × 0.01 × 1.785

1 dt/ha = 0.01 kg/m². The factor 1.785 is the ratio of the molar masses of CaCO₃ and CaO (100.09 / 56.08).

Phosphorus and potassium dose by index

dose [kg/ha] = cell(crop, J);  above 20 % organic matter, no higher than the index 3 cell

Nothing is computed between the cells: four values are printed, not a function. The cap belongs to an Irish environmental regulation (SI 378 of 2006) printed in the table's own footnote rather than to agronomy, and it is named on screen together with the figure the source prints without it.

Nitrogen requirement after the deductions

N_net = N − Σ Ab,  not below zero

There are two deductions: 20 kg N/ha where the humus content is above 4.0%, and 0, 10 or 20 kg N/ha for last year's main crop. The source calls both of them minimum deductions — a lower bound, not a value — and the soil's mineral nitrogen (Nmin) is still not subtracted from the requirement.

Converting a nutrient mass into fertiliser mass

fertiliser mass = nutrient mass × 100 / share [%]

The share of active ingredient is published per product: urea 46% nitrogen, calcium ammonium nitrate 27%, ammonium sulphate 21%, superphosphate 18% P₂O₅, triple superphosphate 45%, potassium sulphate 50% K₂O. It is applied to N_net: the deductions are already taken off, and what is worth converting into fertiliser is what remains to be applied.

Sources

The nitrogen requirement, the liming dose and the share of active ingredient come from Basisdaten für die Düngung und Nährstoffflüsse of the Bavarian State Research Center for Agriculture, tables 9a, 9c and 9d, 8a and 3 respectively. The soil supply classes for phosphorus and potassium and the adjustment to removal by class come from Leitfaden für die Düngung von Acker- und Grünland of the same institute, tables 11, 12 and 32. Both documents serve as the calculation basis for determining fertiliser requirements under the German fertiliser ordinance.

The requirement is a property of the crop

The values are published per hectare of a single crop. Where a bed carries several crops the app shows each figure separately and does not add them: adding would count the bed's area twice.

The supply class is set by a figure, not by a word

Class boundaries are given in milligrams of nutrient per 100 g of soil: for phosphorus by calcium-acetate-lactate (CAL) extraction, for potassium with soil texture taken into account, since the same content falls into different classes on light and heavy soil. A home kit's low/medium/high readings are not those quantities and do not enter the calculation.

Humus content in liming

The dose table is printed in three parts by humus content: mineral soils (up to 4%), humose soils (4.1–15%) and boggy soils (15.1–30%). Their doses differ fivefold, so without a measured humus content no dose is given. A laboratory computes this quantity as Corg × 1.72.

Requirement and single application are different quantities

The source states separately the largest amount of lime for a single application: on mineral soils from 15 to 100 dt CaO/ha depending on texture. On heavy acid soil the requirement exceeds that amount twofold or more, and the remainder is applied in later years.

Deductions from the nitrogen requirement

The German fertiliser ordinance (DüV, Anlage 4) subtracts several quantities from the requirement. Two of them are calculated from data already on the device: the soil reserve from the humus content (table 6, a deduction of 20 kg N/ha where the content is above 4.0%) and the previous crop from last year's main crop (table 7: 20 kg N/ha after grassland, lucerne and clover, 10 after oilseed rape, grain legumes and cole vegetables, 0 after cereals, maize, potatoes and other vegetables). The previous crop is determined from the bed's ground rather than from its record: a bed dug over and marked out afresh gets a new identifier while the soil stays the same.

The catch crop is not calculated

The second half of table 7 sets a deduction for a green manure, and its rows distinguish frost-kill from the time of incorporation: for legumes 10 kg N/ha if frost-killed and 40 if incorporated in spring. The app knows neither — it has no incorporation event, and frost-kill is recorded nowhere — and substituting one of the rows would mean choosing on the source's behalf.

A second method: the dose by RB209 index

The British Nutrient Management Guide (RB209), section 6, answers with a dose rather than a crop requirement, and arrives at it without a soil analysis: the Soil Nitrogen Supply Index comes from soil category, previous crop and annual rainfall (under 600 mm, 600–700 mm, over 700 mm), and the index then gives the dose from a per-crop table. For carrots that is 100 kg N/ha at index 0 and nothing at index 3. Figures from the two methods neither add nor average: they are two different answers to one question, and each has its own heading on screen.

The method is chosen by the laboratory report, not by country

Class and index boundaries belong to the extraction method rather than to the nutrient: calcium acetate lactate for the German method, Morgan for the Irish one, Olsen and ammonium nitrate for the British one, Kirsanov (GOST R 54650) at a Russian laboratory. The method is recorded on the soil test and decides which method may speak from it. The app converts units — mg/kg and mg/100 g differ by a factor of ten, and dividing by ten is the definition of the unit — but never converts between extractions. Until a method is named the figures are read against the CAL boundaries, as they were before the field existed, and this is said alongside them.

A third method: the index by Morgan

Teagasc, The Green Book, chapter 16: four soil supply indices by the Morgan extract and a per-crop dose in kg/ha. Of the twenty-five vegetables the source covers, eight match the app's catalogue. The source prints nitrogen too, and the app does not take it: the nitrogen index is derived from the previous crop, and for a vegetable predecessor the row requires knowing how much nitrogen it received.

A fourth method: the standard doses of Emilia-Romagna

Regione Emilia-Romagna, Disciplinari di produzione integrata 2026: each crop has a sheet with a standard dose of N, P₂O₅ and K₂O in kg/ha. Nitrogen is printed at a stated yield with a list of increases and decreases and a cap on the increases; the app applies those it has data for — the organic matter class from humus and texture group, an annual legume predecessor, and rainfall above 300 mm from 1 October to 28 February — and shows the rest verbatim. N = max(0, standard + min(Σ increases, cap) − Σ decreases). Phosphorus and potassium are printed by soil supply class; the general norms (ch. 11.1) treat a vegetable plot under 1000 m² and a fruit plot under 5000 m² as well supplied without an analysis, and the app takes the “elevata” column only when the garden area is below the limit. Twenty-five catalogue crops, twenty-eight rows: tomato and eggplant have greenhouse and open field, asparagus establishment and production years.

What was examined and not taken

The Dutch Adviesbasis is the only model found in which the dose is a continuous function of the analysis rather than a step of a table. It is not taken for two structural reasons: its phosphorus formulas branch on Dutch soil types that cannot be derived from USDA triangle classes, and its potassium route goes through an intermediate kaligetal that takes the clay content — a quantity the app does not store.

Phosphorus and potassium are not taken from RB209

The per-crop RB209 tables print phosphorus and potassium too, but their indices are defined by the Olsen extract and the ammonium-nitrate extract, while the app stores milligrams from the calcium-acetate-lactate extract. A published review of conversion equations between extracts (Steinfurth et al., 2021) concludes that coefficients differ more than twofold between authors and that conversions should only be carried out if absolutely necessary.

What remains the app's own decision

Assigning USDA texture triangle classes to the four texture groups of the German classification is done by clay content. Assigning the app's crops to the rows of table 7 is done the same way: the source names groups (“Kohlgemüse”, “Gemüse ohne Kohlarten”), not species. RB209 soil categories are determined by describing the profile down to the subsoil, whereas the app knows the topsoil, and the assignment is made from it.