Pasture.io

From plate meters to pixels: why satellite pasture estimates keep improving

Practical pasture decisions for farms that need clear, reliable information.

You’ve got stock to feed, a grazing round to keep moving and a list of jobs that hasn’t got any shorter. Getting around the farm with a plate meter is useful work. Finding the time for it can be another matter.

So the idea of checking pasture from space has obvious appeal. But how does a picture taken hundreds of kilometres away become a useful estimate of the grass in your paddocks? And why should those estimates keep getting better?

The answer involves more than a sharper photo. Satellite sensors, weather information and models that learn from farm data all contribute. Together, they can help you compare paddocks and follow how your feed supply is changing across the farm. Here’s what sits behind the numbers.

From measuring height to reading light

A rising plate meter starts with compressed pasture height. A calibration equation turns that into an estimate of pasture mass. The equation needs to suit the pasture and season.

Satellite measurement starts with a different clue: the light reflected from the paddock. A model interprets that information to estimate pasture mass, often called pasture cover. This is commonly expressed in kilograms of dry matter per hectare (kg DM/ha), meaning the amount of pasture after its water content is excluded. It isn’t counting blades of grass from orbit, impressive as that would be.

That difference opens up another way to follow pasture across a farm. It also explains why the information behind the estimate matters just as much as the image itself.

Why a green paddock doesn’t tell the whole story

NDVI is a measure of vegetation greenness derived from satellite observations. It’s useful, but in thick pasture the score can level off while more grass keeps growing. Think of bathroom scales that have reached their limit. Adding more weight doesn’t move the reading much.

Satellites can capture more information than one greenness score, including light our eyes can’t see. Giving a model more of that information offers more clues about the pasture below.

A 2025 remote sensing study tested this using Sentinel-2 imagery and measurements from 16 NSW dairy farms. Models combining the satellite’s wider range of observations with weather data estimated pasture mass better than approaches based on vegetation indices alone.

Weather adds the story between the pictures

You wouldn’t expect two paddocks to grow the same way after a warm, wet week and a cold, dry one. A useful pasture model needs that context too.

An image captures one moment. Weather information helps a model interpret growing conditions around it. The same study also found that filling gaps between field measurements and helping models learn from examples relevant to the season improved their estimates.

Put simply, progress comes from giving models better clues and better examples to learn from. That’s encouraging research, rather than a promise that every satellite service will give the same result.

And then the clouds roll in

Of course, the satellites still have to see the ground. Cloud, haze, smoke and shadow can get in the way of a usable image. Clouds don’t tend to cooperate with the grazing plan.

Cloud covering pasture within a mapped farm boundary in a satellite image.
Cloud obscures the pasture in the satellite image above, leaving no clear view of the paddocks below.

This is why a flyover and a fresh satellite-backed reading are different things. More frequent flyovers give more opportunities to catch a clear view, but they don’t guarantee usable imagery every day.

Models can help bridge those gaps using previous observations, weather and farm records. It helps to understand the difference between an estimate informed by a new image and an update made between usable images. Pasture.io’s guide to satellite-backed and modelled readings explains how this works in our service.

What could this mean for your grazing decisions?

The useful part isn’t knowing which wavelengths a satellite records. It’s having information you can use when the herd needs feeding.

See how paddocks compare

Looking across the farm can help you spot paddocks getting ahead of the round or recovering more slowly than the others. Better information in heavier covers matters here: two paddocks can look equally green while carrying different amounts of pasture.

Bring paddock estimates together in a feed wedge and you have a starting point for choosing the next grazing or considering an area for silage. Access, pasture quality and stock needs still shape the final call.

Follow the direction of growth

One reading tells you about a point in time. A run of readings helps you see whether feed is building or coming under pressure. Is growth slowing across the farm, or is one paddock simply taking longer to recover?

Those patterns can prompt a closer look at round length, grazing area or supplements. Our guide to turning pasture data into grazing decisions explores how to put the numbers alongside demand and your farm’s targets.

Where Pasture.io fits into the picture

Pasture.io draws on imagery from multiple satellite services with different sensors. Our own machine learning models have been developed using years of data from farms around the world. The Sentinel-2 research above illustrates one approach, but doesn’t describe the whole Pasture.io system.

Pio, our pasture model, combines satellite imagery with local weather, farm records and modelling to estimate pasture cover and growth. NDVI is just one of many inputs. Grazing records add context about when pasture has been removed, rather than leaving the model with an unexplained change.

Pio’s Cloudy Days model estimates pasture cover and growth between usable satellite images. Our Ultimate satellite add-on provides more frequent flyovers, giving you more chances to capture usable imagery between cloudy spells.

In the example below, the table brings cover and growth estimates for different paddocks into one view. You can compare how they’re tracking, while the coloured map shows variation in vegetation greenness. The colours provide another clue, rather than a direct measurement of pasture mass.

Pasture.io reading showing paddock cover and growth figures alongside the selected NDVI map overlay.
Above: A Pasture.io reading, informed by a satellite image. The coloured NDVI overlay shows variation in vegetation greenness; the table shows the pasture cover and growth estimates.

Keep your farm knowledge in the mix

Better models don’t make every paddock straightforward. Dense or rank pasture, dead material, trees and recent grazing can complicate what an image tells you. A green signal and the total feed on offer aren’t the same thing.

Whether you use a plate meter, visual assessment, satellite estimates or a mix, keep a routine you can stick to. Review the farm often enough to follow changing conditions, keep grazing records current and have a closer look when a paddock surprises you.

If you compare methods, compare the same date, area and grazing stage, and understand what each figure includes. You’ll get more from a consistent picture over time than from chasing agreement between two isolated numbers.

The promise of improving satellite measurement is a clearer view of pasture across your farm, with more context behind the numbers. You still bring the knowledge of your paddocks, your stock and what needs to happen tomorrow.