Satellite field monitoring in Uzbekistan: how it works

What is satellite field monitoring

Earth observation satellites image the same area again and again. Overlaying a field boundary on these images produces a series of observations for the whole season and for past years: when the field greened up, when it built biomass, when it began to turn yellow.

Raw images are just values of reflected light or radar signal in pixels. What makes them useful to a farm is processing: masking clouds, calculating indices, summarising them per field and comparing them with what counts as normal for that field at that crop stage.

• The crop's growth stage and how the field is doing compared with its own past seasons
• Patches within a field where the crop is thinner, shorter or yellower than the rest
• Fields that are behind neighbouring fields of the same crop
• Dated traces of events – flooding, harvest, a sharp drop in greenness
• The field's history over past years: how it usually behaves and where problem spots recur

How monitoring works: from image to conclusion

Processing follows one chain, and some images drop out at every step. An image that was found is not necessarily usable: a cloud over the field makes it useless even if the scene as a whole is clear.

A map of indices and a risk warning are different things. An NDVI map shows where a field has less green, but not whether that is bad: low NDVI is normal early in the season and alarming at cotton flowering. A risk warning takes into account the crop stage, the weather and the field's history, names a likely cause and suggests what to check.

• Field boundary: drawn on a map or uploaded as a file
• Image collection: every satellite scene over the field for the season and past years is found
• Quality screening: clouds, their shadows and snow are masked pixel by pixel; if too little clear area is left, the image is not used for that field
• Index calculation: NDVI, NDMI, EVI and other indices are computed for pixels inside the boundary, and for the field as a whole – the mean, the spread and the share of the area seen without clouds
• Comparison: the value is set against the field's past seasons and against neighbouring fields of the same crop
• Field check: an agronomist goes where the series has departed from the norm

Which satellites are used and what each provides

Sentinel-2 are optical satellites of the European Copernicus programme. They record reflected sunlight in the visible, near-infrared and shortwave infrared ranges; vegetation indices are calculated at 10 m detail. Nominally the two satellites revisit about every 5 days, but under clouds a usable image comes once a week or less often.

Sentinel-1 are Copernicus radar satellites. Radar emits its own microwave signal and records the echo, so clouds and darkness do not stop it. It shows not greenness but crop structure, soil moisture, standing water and irrigation; NDVI cannot be calculated from radar. The revisit is about 6–12 days per track, imaging from two orbit directions gives more frequent observations, and only images from the same track can be compared because the signal depends on the viewing angle.

Landsat 8 and 9 provide a thermal band – land surface temperature. A crop short of water closes its stomata and heats up more than a healthy one, so the thermal band helps detect water and heat stress. But the thermal pixel is 100 m, before the rows close it includes bare soil, and each satellite revisits about every 8 days, both together about every 4.

Images are complemented by weather data, for example the ERA5 reanalysis with a forecast: weather warns of frost, hail or heat before the event, and the satellite shows the consequences after it.

• NDVI – the Normalized Difference Vegetation Index from the near-infrared and red bands: the more green biomass, the higher the value; it saturates over a dense closed canopy
• NDMI – a moisture index from the near-infrared and shortwave infrared bands: it shows how much water is in the leaves and is meaningless over bare soil
• EVI – the Enhanced Vegetation Index: it corrects for soil and atmospheric effects and saturates less over a dense canopy
• NDRE – the red-edge index, available only from Sentinel-2: sensitive to chlorophyll in a closed canopy where NDVI no longer separates fields

Satellite field monitoring in Uzbekistan

Almost all farming in Uzbekistan is irrigated, and the main risk to the harvest is water. The country has about 4.3 million ha of irrigated land, about 175,000 ha of fields at the tail ends of canals suffer from water shortage, and the water deficit may reach 15 billion m³ a year by 2030 (Kazinform, 30.11.2023).

The state already uses satellites. A decree of 13 March 2026 provides for an Agricultural Monitoring Centre that will analyse satellite data, weather and crop vegetation (Gazeta.uz, 16.03.2026). Over two years the Uzbekcosmos agency checked more than 4 million ha from space and found discrepancies with official reports, including unsown cotton (Kun.uz, 01.08.2025).

Such systems serve top-down control. For the farm itself, satellites are useful in a different way: to notice a problem on its own field while it can still be fixed, and to keep an objective record of an event – date, boundary, images before and after. This also matters for agricultural insurance, which by 2027 is meant to cover 45,000 agricultural enterprises and 60% of farm output (Spot.uz, 25.09.2026).

What satellites cannot see and what to check in the field

A map is not a diagnosis. A satellite signal tells you where to go; what actually happened is determined by an agronomist on site – inspecting leaves, checking irrigation and soil moisture, counting plants, testing soil and leaves.

• Clouds: optical sensors cannot see through clouds, and in spring a clear image of a field comes once every 10–20 days; a missing image is not a sign of a problem in the field
• Pixel size: at 10 m individual plants in a row are not visible, so the satellite does not count plant stand but shows that a field is below normal; thermal data are less reliable on fields under 5 ha
• Saturation: over a dense canopy NDVI stops rising and differences between fields flatten out
• Cause: an image shows that a field has little green or is lagging, but not why – water stress, salinity, nitrogen shortage and hot dry winds look alike
• Sparse data: radar and thermal scenes are few, and an event lasting a couple of days can be missed

What it looks like in practice: the ClusterOS example

ClusterOS is an early-warning platform for agricultural risks built for farms and agro-clusters in Uzbekistan. It collects Sentinel-2, Sentinel-1 and Landsat imagery and ERA5 weather for every field, with field history available since 2019: the map shows where the crop is dense and where it lags, and the image feed can be opened on any date of the season.

Index values for a field – the mean, the spread and the share of the field seen without clouds – are compared with that same field's norm. On top of the map the system tracks 24 agricultural risks with the crop and its stage in mind and sends a warning to the web dashboard and Telegram: what is threatening, why and what to check. If there is no observation, for example because of clouds, the system says so, and the decision stays with the agronomist.

How to start satellite field monitoring

No sensors or equipment are needed: Sentinel and Landsat imagery is open and available for any field. What you need is field boundaries, crops and someone on the team who goes out to check.

• Prepare field boundaries: a GIS file or marks on a map
• Enter the crop and, if known, the sowing date for each field – this decides what the images are compared with
• Review the fields' history over past seasons to find patches that lag year after year
• Agree who receives signals and goes to the field, and record irrigations and inspection results – this makes the assessments more accurate

Related risks

• Water stressThe crop is short of water. Imagery and the forecast show it before the leaves start to wilt.
• Poor emergenceSeedlings are sparse or uneven – it shows in imagery in the first weeks.
• Development lagThe field develops more slowly than in its own past years and than neighbouring fields at the same stage.
• Nutrient deficitImagery shows signs of nutrient shortage – it is clear where top dressing is needed.
• SalinitySalt rises to the surface on irrigated land – we find the problem parts of the field.
• Anomalous dynamicsThe crop changes unlike what is usual for this crop and stage – a reason to check the field.

Questions