AI farm advisory for each of your fields

How an AI farm advisory works out irrigation scheduling

Most irrigation advice, from extension services and from software, is built on one idea from the FAO's guide to crop evapotranspiration (Irrigation and Drainage Paper 56). Evapotranspiration is the water a field loses to evaporation from the soil and transpiration through the leaves. You start from a reference figure, ET₀, calculated from the weather: temperature, humidity, wind and sunshine. Hot, dry, windy days push it up.

Crop coefficients, simply

A crop coefficient (Kc) scales that reference to your crop at its current stage. A young maize crop with bare soil between the rows uses far less than the reference, with a Kc around 0.3. At silking, with the canopy closed, it's around 1.2. So on a day with an ET₀ of 6 mm, mature maize uses roughly 7 mm. Subtract the rain that will actually soak in, allow for what the soil is still holding, and you have the net amount to put back.

Our irrigation plan uses the crop's stage and water use, recent satellite readings of plant and soil moisture, and forecast rain with its probability. That's why the four pivots in the Algerian table above got such different numbers in the same week. The satellite part matters because it shows how the crop is actually doing, which can differ from what the calendar says it should be doing.

Where the fertilizer recommendations come from

A fertilizer recommendation is a balance: what the crop needs to reach a realistic yield, minus what the soil can already supply. The advisory reads each nutrient against a target range for that crop, then gives the shortfall as a dose in kilograms, with an organic and a chemical way to meet it. When a nutrient is already in range, the dose is zero.

Some nutrients are easier to see from orbit than others. Nitrogen shows up fairly directly in leaf chlorophyll, which red-edge indices such as NDRE pick up (our post on NDVI vs NDRE explains why). Phosphorus, potassium and micronutrients leave a much weaker signal in the imagery, so any satellite-based estimate of them is less direct. A lab soil test every season or two is still the best check on them, and it's worth comparing its results with the readings in the report.

How pest and disease risk is estimated

No satellite can see an aphid. What it can see is the crop's condition and stage, and combined with the weather that says a lot about which problems are likely. Fungal diseases need moisture on the leaf, so long humid spells and a wet canopy raise the risk of leaf spots, rusts and blights. Insects develop faster in warm weather and attack particular growth stages: fall armyworm in maize at the whorl and ear stages, for example.

The advisory puts these together into a named risk with a probability and shows the readings behind it. Treat a high-risk warning as a reason to scout that field this week. Before you spray, confirm the pest is there and check it against the action threshold your agronomist or extension service uses for that crop. That's standard integrated pest management, and the warning is meant to make the scouting trip more targeted.

How to judge an AI agronomist or crop advisory system

What to check

• Is the advice specific to the field, or does every field in the district get the same message?
• Are amounts given in units you can act on, such as millimeters and kilograms, with a date or window?
• Does it ever say "apply nothing" or "do not irrigate"? A system that always recommends something deserves a closer look.
• Can you see the readings and references behind a warning?
• Does it offer organic options, and does it reach your growers in their own language?

The fairest test is a season on one field with a strip managed your usual way next to it. Compare water used, inputs bought and yield at the end. Our article on AI crop advisory platforms and our comparison of AI farming assistants and apps go further.

What it can't know without you

An AI farm advisory sees the field from above and reads the forecast. From above it can't tell the variety you planted, when you last ran the pivot, what went on with the last spray, whether a tile drain is blocked or how much water your pump can actually deliver. It also can't know the label restrictions and pre-harvest intervals that apply to you, or what your buyer's contract allows. Keep your own records next to the report, and adjust the advice where you know something it can't.

Working with agronomists and consultants

The advisory works best as the first pass before an expert looks at the field. A typical routine for a consultant: on Monday, open each client's reports and sort the fields by risk and stress. Visit the ones that changed most, take the advisory's numbers as a starting point, and adjust them with what you know about the farm. Then send the grower the PDF, or a share link, in the language they read.

This changes where a consultant's time goes more than how much of it there is. The AGROEX consultancy in our case studies received 181 reports across 15 client fields in 75 days, a fresh report for each field about every six days. If you're curious how the role is shifting, we've written about agronomist careers and the AI agronomist. If you only need the crop health maps and indices, see satellite crop monitoring.

A season with the advisory, stage by stage

Here is how a grower would typically use the report across a maize season. The pattern is similar for most annual crops.

Toward the end of the season the irrigation plan winds down, as it did for the durum wheat at maturity in the Algeria example, which got only small amounts on three days that week. The yield estimate and harvest date become the numbers you plan around.