Satellite Crop Advisory Central Asia: 10 Crops

The portfolio in one frame. The percentages shown are the client’s own reported estimates from our in-app feedback form. The sections that follow set out, field by field, what the advisory actually told them.

A single farm growing a single crop is the easy case for any advisory system. The hard case is the one on this page. AGROEX Consultancy manages 15 separately mapped fields totalling 631.7 acres (255.6 hectares) for growers in three different countries, carrying ten different crops that were, in the week we examined, sitting at seven different points in their season. One field was at bud burst. Another was in full fruit development. A third was a cotton block at squaring. There is no version of a generic recommendation that is useful to all of them at once.

In June 2026 the consultancy completed the feedback form attached to its satellite crop advisory and reported that the reports had helped it optimise irrigation by 30–50% and fertiliser use by more than 70%, and had helped it act on pest, disease and weed pressure. This article shows exactly what the advisory told the consultancy, field by field — and traces the specific mechanisms in it that a ten-crop, three-country portfolio depends on.

Illustrative. The production landscape this case study covers: irrigated orchard on a valley floor, poplar windbreaks, earthen distribution channels, arid foothills behind. Representative imagery, not a photograph of the client’s own fields.

One account, three countries, ten crops

The portfolio is unusual enough to be worth laying out plainly. By field count it is mostly Uzbek; by area it is mostly Tajik, because the two largest blocks — a 224.8-acre melon field and a 96.9-acre peach orchard — are both in Tajikistan.

Each field carries its own satellite crop advisory, refreshed on every pass. The consultancy onboarded on 26 March 2026. By the date of the feedback, 9 June, the account had received 181 individual field advisories drawn from 46 separate satellite acquisition dates in 75 days. Because the fields sit at different longitudes across three countries, they are not all imaged on the same pass — so on most days of that period, a new advisory landed for some part of the portfolio.

Ten crops. Seven growth stages. Three countries. One week of advisories, and no two of them alike.

What the satellite crop advisory delivers per field

Every field on the account receives its own report per satellite pass, carrying five working sections: irrigation, fertiliser, pest and disease, weed, and soil management, plus a growth and yield estimate. The three sections below are the ones the consultancy credited — and the ones where a ten-crop portfolio makes differentiation impossible to fake.

Pest and disease: the model names the species, not the symptom

This is where a multi-crop portfolio either exposes a satellite crop advisory or validates it. A shallow model returns a plausible-sounding disease name. A useful one returns the pathogen that actually attacks that host at that stage — because the control product, the timing and the threshold all follow from the species, not from the symptom.

The clearest test in this dataset is scab. The advisory raised Apple Scab (Venturia inaequalis) on the apple blocks and Pear Scab (Venturia pyrina) on a pear block. Same genus, different species, each on its correct host. Powdery mildew behaved the same way: Podosphaera leucotricha on apple, Uncinula necator on grape, Sphaerotheca pannosa on peach — three different organisms that a generic “powdery mildew” label would have collapsed into one.

Put two cards from the same account side by side and the differentiation stops being an argument and becomes an observation. Both fields belong to the same consultancy, in the same country, in the same week.

The pear orchard card: Pear Psylla (Cacopsylla pyri) at 85%, Codling Moth at 80%, Field Bindweed at 85% — each with an organic and a chemical option.

The cotton card from the same account, same week: Cotton Bollworm (Helicoverpa armigera), Verticillium Wilt, Pigweed, Cotton Aphid. Nothing carried over from the orchard.

Weed pressure separated the same way. Field Bindweed (Convolvulus arvensis) — the classic perennial problem of Central Asian orchards — was flagged on nearly every tree-crop field, reaching 96%. The cotton block instead led with Pigweed (Amaranthus spp.) at 88%, and the melon field with Common Purslane (Portulaca oleracea) at 85%. Every alert carried both an organic and a chemical control option, from pheromone mating disruption for codling moth to Bacillus thuringiensis for bollworm.

Irrigation: the method changes with the crop

Here the mechanism is not the one people expect. Only 22 of 105 field-days in the week carried a zero-irrigation instruction, so skipping days is not the story. The story is method.

Across the portfolio the advisory recommended drip irrigation on 12 of the 15 fields, against the furrow and basin flood irrigation traditional to the region. On the fields where drip was recommended, the recommended daily quantity was 27.8% lower, area-weighted, than the flood alternative the same report listed — with per-field differences running from 18.8% to 40.0%.

Illustrative. Furrow irrigation in an orchard: water sheeting across bare soil between the rows, much of it evaporating or draining below the root zone rather than reaching the trees. This is the baseline the drip recommendation is measured against. Representative imagery.

What makes this credible is the three fields where the model did not recommend drip. On the melon block, the cotton block and one apricot orchard, furrow irrigation was retained. A system that simply preferred the modern technology would have said “drip” fifteen times out of fifteen. This one recognises that a sprawling melon crop on raised beds and a furrow-irrigated cotton field are not orchard rows, and left them alone.

The pear orchard’s irrigation card: Drip selected, four irrigation days, 18.0 mm for the period, window 05:00–08:00.

The cotton field’s card, same account: Furrow irrigation retained, three irrigation days, 12.3 mm, window 05:00–08:30. The method selector is not set once for the farm — it is resolved per field.

Irrigation windows across the whole portfolio fell between 05:00 and 08:30 — before the continental daytime heat drives evaporative loss. The underlying calculation is named in the report’s own references: FAO-56 Penman–Monteith reference evapotranspiration, crop coefficients for stone fruits, Sentinel-1 radar soil-moisture and vegetation indices, and Sentinel-2 NDVI, adjusted to regional irrigation norms.

Illustrative. The recommended alternative: a drip emitter wetting a small circle at the tree base while the soil between rows stays dry. On loess soils that is the difference between water in the root zone and water in the subsoil. Representative imagery.

Fertiliser: no two fields get the same mix

The fertiliser section reports five nutrients per field — nitrogen, phosphorus, potassium, sulphur and zinc — each with the modelled current level, the ideal range for that crop, a rate, and both an organic and a chemical source. The report even names the grades its arithmetic assumes: Ammonium Nitrate at 34% N, Superphosphate at 20% P, Potassium Sulfate at 50% K.

Nitrogen varied a relatively modest amount across the portfolio — 9.6 to 40.4 kg/acre, a 4.2-fold spread. The interesting variation is in the other two. Phosphorus ranged from 4.0 to 57.0 kg/acre, a 14.4-fold spread. Potassium ran from zero to 96.0 kg/acre — one apple block needed no potassium at all, while a plum block needed 96 kg/acre of it.

Summed across the portfolio the advisory prescribed roughly 76,700 kg of nutrient across all five elements. If the consultancy had instead applied one uniform, area-weighted average rate to every acre — a defensible thing to do without field data — then 34.0% of that programme would have landed in the wrong place: over-applied on the fields that needed less, under-applied on the fields that needed more. For potassium alone the misallocation would have been 48.8%.

The fertiliser card triages the five nutrients into Need Attention / Monitor Closely / Optimal, then gives current-versus-target per nutrient with both an organic and a chemical source. A nutrient already inside its target band generates no application.

Potassium ran from zero on one apple block to 96 kg per acre on a plum block — same client, same country, same week.

What the numbers add up to

Pulling the three mechanisms together gives a clear picture of where a portfolio like this one gains, and each figure below is computed from the advisory itself rather than estimated.

Grower-reported outcomes, as submitted

The cards below reproduce AGROEX Consultancy’s answers to our in-app feedback questionnaire for the advisory dated 7 June 2026, exactly as recorded. They are structured survey responses, not paraphrases, and not verbatim quotations.

Building soil, not just applying fertiliser

One reading runs across all 15 fields in all three countries and shapes the entire nutrient strategy: soil organic carbon sits between 0.12% and 0.25%. Organic matter is what holds both water and nutrients in the root zone, so that reading changes what the right recommendation is. Soil pH came back between 6.8 and 7.8 — neutral to mildly alkaline, which for fruit trees is where iron and zinc availability becomes the thing to manage rather than the total applied.

On six of the fifteen fields salinity reads moderate rather than low — a well-known characteristic of long-irrigated Central Asian farmland, where evaporation concentrates salts in the upper profile faster than leaching clears them. The advisory tracks it per field and cites regional soil-profile studies and World Bank agriculture sector reporting, so the grower sees exactly which blocks to watch rather than a portfolio-wide guess.

Illustrative. The 224.8-acre melon block is the single largest field in the portfolio and one of the three where the advisory kept furrow irrigation. It also returned moderate salinity and a pH of 7.8. Representative imagery.

This is why the fertiliser guidance for these fields leans on organic-matter building alongside the mineral rates — composted pine bark is the organic source the advisory names for sulphur, and it does double duty on soil structure. Reading carbon and salinity per field is what turns a nutrient plan into a soil plan, and it arrives on every pass at no extra cost.

Why per-crop beats per-farm at portfolio scale

The yield estimates make the final argument on their own. Across the portfolio, expected yields ran from 1,020 kg/acre on the cotton block to 9,100 kg/acre on an apple orchard — an 8.9-fold spread. Those two fields belong to the same client, sit in the same country, and were advised in the same week. Any metric averaged across them describes nothing that exists.

Zone-level crop health for a single orchard block, read per pixel from satellite. Within-field variation like this is invisible to a farm-level average and is what makes a targeted intervention cheaper than a blanket one.

This is the practical case for a per-field satellite crop advisory in a consultancy rather than an owner-operator setting. The consultant’s problem is not “how is my farm doing” — it is “which of my fifteen fields, in three countries, needs me this week, and what specifically do I tell that grower”. Answering it requires the guidance to be resolved to the field, the crop and the growth stage. On the evidence here, it is.

Illustrative. The cotton block at squaring — the one field in a fruit-dominated portfolio, and the one that got bollworm, Verticillium wilt and pigweed while every orchard around it got codling moth and bindweed. Representative imagery.

Running this on your own portfolio

The workflow that produces a satellite crop advisory needs no hardware in the field.

1. Map every field separately, even across borders. This account’s value comes from 15 boundaries in 3 countries, not one region. Fields in different countries are simply different fields.
2. Set the crop and sowing date per field. Crop coefficient, phenological stage and pest model all key off this — it is what produced Venturia pyrina on the pear and Venturia inaequalis on the apple.
3. Read the irrigation method selector, not just the millimetres. On this portfolio the method carried more saving than the schedule did.
4. Compare nutrient mixes field by field. The totals hide the misallocation; the ratios expose it.
5. Act on the soil warnings. Low organic carbon and creeping salinity are slow problems that only get more expensive.

Explore the platform at sat.farmonaut.com, read more about our satellite and AI-based farm advisory in the JEEVN AI overview, or contact our team about multi-country portfolio onboarding. JEEVN AI is a paid product; plan detail is on the platform.

Frequently asked questions

Glossary

Phenological stage

Where a crop sits in its season — bud burst, pink bud, flowering, fruit set, fruit development, squaring. It determines water demand, nutrient demand and which pests are relevant. Seven stages were live across this portfolio at once.

Crop coefficient (Kc)

The multiplier that converts reference evapotranspiration into a specific crop’s water demand at a specific stage. Different for apple, melon and cotton, which is why one irrigation figure cannot serve all three.

Furrow and basin irrigation

Traditional gravity flood methods that deliver water along channels or into bunded areas. Simple and cheap, but a large share of the applied water never reaches the root zone.

Soil organic carbon (SOC)

A proxy for soil organic matter, governing structure, water-holding capacity and nutrient buffering. Flagged low on all 15 fields here, at 0.12–0.25%.

Salinity (soil)

Accumulated soluble salts in the profile. In long-irrigated arid regions, evaporation concentrates salts faster than irrigation leaches them. Six of these 15 fields returned moderate rather than low salinity.

Case-study basis: JEEVN AI field advisories for a single Farmonaut account, taking each field’s most recent advisory as of the client’s feedback submission (portfolio state at 7–9 June 2026, 15 fields), plus the account’s in-app feedback submission for the advisory dated 7 June 2026. Client organisation name, crops, countries, field count and area are reported from the account’s own records on the platform; individual grower names and field locations are withheld. The 30–50% irrigation and above-70% fertiliser figures are the client’s own estimates from a structured feedback form, not metered or invoiced measurements; the 27.8% water and 34.0% nutrient-misallocation figures are our own calculations from the advisory data against stated baselines. Pest, disease and weed percentages are modelled probabilities, not confirmed infestations. Yield figures are estimates, not guarantees. External agronomic references are attributed as cited within the advisory itself, including FAO crop guidelines, UC IPM pest management guidelines and regional soil-institute standards. Imagery: the five platform screenshots and the crop-health zone map are genuine captures of this account’s own advisory and satellite data; the five wider field photographs are AI-generated illustrative imagery, each labelled as such in its caption, and do not depict the client, its growers or their fields. JEEVN AI is a paid product.