Crop recipes

Water

Import lab reports

Water library by greenhouse

Every saved source, under the greenhouses it serves, with its latest analysis. Change a kind right here; open a source to use it in the mix.

Custom mix (advanced)

Three or more sources, an older analysis, or values typed in by hand. Most weeks the water is picked on the Calculation view instead.

Drain helpers: how much drain can go in?

With the other sources kept in their ratio: the largest drain share that keeps sodium or chloride under the cap, and the share that brings the start water to the EC. Needs a source of kind "drain".

Fertilisers on the shelf

Solid fertilisers

Salts

Liquid fertilisers

Common liquids
Acids

Micronutrients

Element Target (µmol/L) Product
Fe
Mn
Zn
B
Cu
Mo

Stock tanks

Water for this recipe

Three or more sources, an older analysis, or values typed in by hand:

Target recipe

Booklet: EC – dS/m · pH –

Macro elements — mmol/L Micronutrients — µmol/L
NO₃⁻NH₄⁺PK⁺Ca²⁺Mg²⁺SO₄²⁻Cl⁻SiFeMnZnBCuMo
Booklet
Your adjustment
About this recipe

Saved recipes

How the water has moved

The Bal: nitrate, ammonium and sodium

Every four weeks, per greenhouse

Onderbemaling: nitrate every quarter

Method

Where the numbers come from

Nothing here is read from a fertiliser table. A salt is its molecular formula; its nutrient content follows from IUPAC atomic weights. Recipes are the standard solutions of "Nutrient Solutions for Greenhouse Crops" (Van der Lugt et al. 2020, on the Dutch Bemestingsadviesbasis), or you type your own.

EC is computed from the ions themselves (Kohlrausch's law with Davies activity coefficients, the model PHREEQC uses) — within 0.3 % of measured values for a single salt and about ±10 % on a mixed nutrient solution. The glasshouse rule of thumb, 0.095 × Σ cations + 0.19, is shown alongside as a cross-check.

Acid demand uses carbonate speciation at the target pH; phosphoric acid is credited with the 1.02–1.17 protons it actually donates at fertigation pH rather than a whole number. Doses are found for all products at once by non-negative least squares, so the compromise a limited set of salts forces is spread across the recipe instead of landing entirely on whatever is dosed last.

For the diluted drip solution, the saturation index of calcite, gypsum and brushite (the calcium phosphate that forms first in a dosing line) is computed the way PHREEQC does — Davies activities against published solubility products. A 1:100 stock tank sits at 3 mol/L, six times past where any activity model holds, so the tanks are handled by rule — calcium never with sulphate or phosphate, potassium nitrate split to level the tanks — and each salt is read against its CRC solubility at the tank temperature.

A lab report is read by position, not by its text: pdfplumber gives every word with its coordinates, the header row of ion symbols defines the columns, and each value belongs to the column it sits under — the only reading that survives a PDF whose text layer comes out in drawing order. Values below the detection limit are filled in as 0 and marked; Si, Al and K/Ca are read and shown but never filled in; a lab's EC-corrected history rows are kept but flagged, never mistaken for a measurement. Water sources mix conservatively — each ion by share — except the carbonate system, where total inorganic carbon and alkalinity are what a closed mix conserves, so the pH of a blend is solved from those, not averaged: a pH 8.2 bore diluted with unbuffered pH 6 rain stays at 8.2.

Treat the output as a starting recipe: confirm EC and pH of the drip solution with a calibrated meter, and dissolve a new stock recipe once in a bucket before filling a tank. Solubility limits are per litre of water and ignore common ions; the warning from 70 % is the margin for that.