— 01 · Agriculture

Precision agriculture

Agronomic decisions depend on observing the same area over time, in the right place. Convex turns orbital imagery, spectral indices and climate context into interpreted time series per monitored area, so within-field variability stops being invisible.

Agronomic decisions happen in short windows, over areas that are never uniform. Convex observes each monitored area on every eligible orbital pass, computes spectral indices, builds the time series and returns an operational reading with local climate context.

Talk to Convex Back to home

01 · Agriculture

— Operational context

The operational problem

01
Within-field variability
A field average hides zones that behave differently. Without a spatial cut, management stays uniform over ground that is not.
02
Discontinuous observation
Field walks are occasional and expensive. Between visits, meaningful change happens with no record.
03
Optical coverage limited by cloud
Not every satellite pass yields a usable scene. The cycle must handle the absence of a valid observation explicitly.
04
Raw data with no agronomic reading
Isolated indices do not say what to do. The missing layer is the one that interprets the trend and writes it down.
05
Climate decoupled from imagery
Rainfall, temperature and water deficit explain much of what an image shows, yet usually live in a separate system.

— Convex architecture

From observation to action

The same architecture runs across every sector. What changes is which layer is critical and which proprietary systems take part.

Observe→

Interpret→

Decide→

Coordinate→

Act

Observe

Nexa Vision

Clipping to the drawn area, acquisition of eligible Sentinel-2 scenes and per-pixel spectral index computation.

Interpret

Nexa M1

Reading of the time series, comparison against previous cycles and a written operational diagnosis of the area.

Decide

Atmos

Local weather context folded into the reading: rainfall, temperature and operational window conditions.

Coordinate

Synapse Control

Monitoring cycle scheduling, mission state and automatic report delivery per area.

Act

Axis

Physical execution layer in the field. In development — see technology stage.

— Data layers

What feeds the pipeline

Sources compatible with the systems deployed in this sector. No layer is assumed: every monitored area declares what is actually available.

Orbital optical imagery

Sentinel-2 multispectral scenes filtered by cloud cover and validity over the area.

Spectral indices

NDVI, NDRE and derived indices computed over the exact declared polygon.

Geospatial data

GeoJSON polygon of the monitored area, server-side area calculation and per-field clipping.

Time series

Cycle-by-cycle history, so the same area can be compared across months.

Climate context

Local weather series matched to the period of each observation.

Operational history

Previous reports for the area, preserved as a record of what has been observed.

— Applications

Technical applications

Each application states the problem addressed, the data used, the Convex system involved and the operational output produced.

Variability map per area

Uniform management over a heterogeneous field.

Data: Sentinel-2 scenes, spectral indices, area polygon.
System: Nexa Vision
Output: Per-index maps clipped to the area, with a relative severity scale.

Temporal tracking of the cycle

Observations that cannot be compared to each other.

Data: Per-cycle spectral index time series.
System: Nexa Vision + Nexa M1
Output: A curve per area showing how the cycle evolved, with change highlighted.

Signals consistent with anomalies

Losses noticed too late.

Data: Spectral indices, area history, climate context.
System: Nexa M1
Output: Zones behaving differently from the rest of the area, flagged for ground verification.

Climate reading of the period

Imagery without an explanation of what happened in the field.

Data: Local weather series for the observed interval.
System: Atmos
Output: Rainfall and temperature context attached to the area report.

Automatic delivery per cycle

Reports that depend on somebody remembering to generate them.

Data: Subscription state, calendar of eligible passes.
System: Synapse Control
Output: PDF report delivered automatically when each cycle closes.

— Operational flow

One cycle, from sensor to record

01Capture
The eligible orbital pass over the area is identified and the scene is validated for cloud cover.
02Processing
Spectral indices are computed over the exact polygon declared by the customer.
03Interpretation
Nexa M1 reads the series, compares it with the area history and writes the cycle diagnosis.
04Alert
Divergent zones and relevant change are highlighted for ground verification.
05Action
The agronomy team decides the intervention with the spatial cut in hand.
06Record
The cycle is archived and becomes part of that area's time series.

— Deliverables

What the system delivers

Maps

Per-index clips over the area polygon.

Time series

Cycle-by-cycle evolution of the same monitored area.

Diagnosis

Written interpretation of the cycle, produced by Nexa M1.

Reports

One PDF per cycle, delivered automatically.

Climate context

Weather conditions for the observed period.

Operational history

Preserved record of previous cycles.

— Technology stage

Maturity by capability

We separate what is in operation from what is still under development. Research and prototypes are never presented as deployed product.

In operation
Area monitoring and spectral indices
Full cycle of acquisition, computation and report delivery.
In operation
Nexa M1 operational interpretation
Written diagnosis derived from the area's time series.
In operation
Temporal history
Retroactive queries over previous cycles of the same area.
In development
Specific pest and disease classification
Models flag divergent behaviour, not a phytopathological diagnosis. Validation in progress.
In development
Field execution with Axis
Robotic actuation integrated into the monitoring cycle.