askcust develops NDVI crop monitoring on Sentinel-2 imagery: index maps for every field, seasonal trend curves, field comparison and zones for scouting. For agronomists who need an answer to where to go and what to check rather than a colourful image.
NDVI crop monitoring is the regular calculation of the normalised difference vegetation index, NDVI = (NIR − Red) / (NIR + Red), from satellite images of each field, with its changes tracked through the season. The index ranges from −1 to 1 and rises with the amount of green biomass. Unlike satellite monitoring in general, it focuses on one index: what it shows, where its limits are and how to read its dynamics.
The service does more than colour the field: it calculates the mean, spread and share of areas below a threshold and compares them with the previous image. The agronomist sees which field got worse and by how much instead of guessing from shades of green.
NDVI values depend on the crop and growth stage: 0.4 is normal for emerging maize but alarming for wheat at heading. That is why the service compares a field with itself in past seasons and with fields of the same crop and stage this year.
Once the canopy closes, NDVI saturates and no longer separates strong and weaker areas. For these stages the service adds NDRE on the red-edge band, which stays sensitive to chlorophyll content for longer.
The field is split into zones by index value, and inspection points with coordinates are formed from the weakest zones. The scout goes where the image showed a lag, and the inspection result returns to the map together with the cause.
Indices are calculated on cloud-masked Sentinel-2 imagery; for more precise tasks the farm's commercial imagery can be connected.
An index map for every clear image within the field boundary, with the same colour scale so that fields and dates can be compared with each other, not just viewed one by one.
Curves of the field's mean NDVI over the season next to curves of previous years and of fields with the same crop; you can see when a field started to lag and whether it caught up after a treatment.
MSAVI for early stages with exposed soil, NDRE for dense crops and NDMI for leaf water content, each as a separate layer on the same field map.
The field is split into several zones by index value or its multi-year mean; zones export as boundaries for scouting and for variable rate prescription maps.
If the mean NDVI of a field or part of it drops between two clear images by more than the agronomist has set, the service sends an alert with a change map and a link to the field.
A weekly summary: which fields improved, which got worse, which are covered by clouds, and a list of inspection points for the coming days, without opening every field separately.
Formulas are given for Sentinel-2 bands; an index's resolution equals the coarsest of its bands.
| index | formula | what it shows | limitations |
|---|---|---|---|
| NDVI | (NIR − Red) / (NIR + Red); bands B8 and B4, 10 m | amount of green biomass and overall crop condition | saturates on dense closed canopies; affected by soil at early stages |
| NDRE | (NIR − RedEdge) / (NIR + RedEdge); red-edge band B5, 20 m | chlorophyll content, including in dense crops | 20 m resolution, so small fields show less detail |
| MSAVI | NDVI adjusted for exposed soil; bands B8 and B4, 10 m | crop condition at early stages, when soil shows between rows | no advantage over NDVI once the rows close |
| NDMI | (NIR − SWIR) / (NIR + SWIR); SWIR band B11, 20 m | leaf water content and early signs of water stress | does not show soil moisture directly; 20 m resolution |
askcust collects the list of crops, varieties and typical stages per field so that the index is compared against the right baseline rather than a neighbouring field with another crop.
NDVI curves for previous years are built for every field from the Sentinel-2 archive; the agronomist marks seasons with known problems, and alert thresholds are tested on them.
Splitting fields into zones and forming inspection points are set up; the first scout routes check whether the zones really match what can be seen in the field.
A weekly summary and alerts about index drops go to the agronomist and the manager; zones are passed to the field map and precision farming systems.
examples of askcust's work →not agriculture, but the same kind of task: records, CRM, automation
Roughly, 0.3 means sparse vegetation (emergence, late ripening or a crop with problems), while 0.7 means dense green cover. According to USGS, sparse vegetation usually gives 0.2–0.5 and dense vegetation 0.6–0.9. The conclusion, however, depends on the crop and growth stage, so a field's values are compared with the same field in past seasons and with fields of the same crop.
Because green leaves absorb almost all red light already at moderate canopy density, after which the red band barely changes and NDVI stops growing. As a result, strong and weaker areas of a closed canopy get similar values. For these stages NDRE, which uses the red-edge band, is more useful.
NDRE stays sensitive to chlorophyll content longer, when the crop is dense and NDVI has already saturated, so it is used in the second half of the season and for assessing nitrogen status. The drawback is that Sentinel-2's red-edge band has a 20 m resolution versus 10 m for NDVI, so small fields show less detail.
With every new clear Sentinel-2 image: at Ukraine's latitudes the satellites image a field every 2–3 days, but clouds filter out some of the images. In clear weather new maps appear several times a week, in cloudy periods less often. The service shows the date and cloud cover of every image so the agronomist can see how fresh the data is.
Yes, as a basis: field zones by NDVI or its multi-year mean are exported to Shapefile or ISOXML and become a prescription map once the agronomist sets a rate for each zone. Before that the zones should be checked in the field and with soil analyses, because a low index can have different causes.
The cost depends on the number of indices, the archive depth, the alert rules and integrations with the field map and precision farming systems, not on area. askcust estimates the first stage (NDVI maps and curves for the farm's fields) with a fixed scope; zones, other indices and alerts come in later releases.
A reply within one business day; a call only if we agree on one.