Select Page

By Sam Lavender, Pixalytics

Pixalytics Ltd is leading the generation of products from satellite imagery for EcoProMIS. Although satellites can cost a lot to build and launch, once in space, they have the advantage of continuously collecting data.

Satellite Lifetimes

Many satellite missions last much longer than their designed lifetime, which is typically five years, although they can go wrong, and then it is not easy to fix them. A successful mission includes Landsat-8 that was launched in February 2013 and continues to operate, with Landsat-9 planned for launch in 2023.

One of the longest-lived optical satellite missions is CHRIS/Proba-1 that has been collecting data for nearly 20 years.

In contrast, WorldView-4 was launched in November 2016 and the gyros failed in January 2019 and prevented the spacecraft from pointing accurately. The manufacturer said that while efforts are continuing to restore the spacecraft, “Maxar believes that WorldView-4 will likely not be recoverable and will no longer produce usable imagery.”

Unlike the Hubble Space Telescope, which was serviced by astronauts from the International Space Stations (ISS), polar-orbiting Earth Observation (EO) satellites tend to be orbiting at an altitude almost twice that of the ISS; at around 700-800 km. The exception is the smaller satellites, called CubeSats, that are orbiting at lower altitudes and have been deployed from the ISS. The disadvantage of this lower orbit is that these CubeSats do not have the power to maintain their orbit’s altitude, and so they burn up within the Earth’s atmosphere within a few years.

Data Collection

For EcoProMIS, we are using free-to-access datasets for the baseline data collection: the European Union’s Copernicus Sentinel missions alongside the U.S. Geological Survey/NASA Landsat-8 mission.

These missions are termed free-to-access as anyone can download and process this data. Still, some practicalities include understanding what has been collected, where to get the data and how to handle it. Therefore, Pixalytics continually processes the satellite data to generate near-real-time products that are made available to EcoProMIS.

An example shown below is the classification of land cover using Sentinel-1 and -2.

EcoProMIS Land Cover Classification, data courtesy of Copernicus/ESA.


The satellite-derived products are focused on understanding the health, growth, and potential yield of the crops alongside greenhouse gases. Sentinel-1 provides microwave data that can see through clouds and detect the roughness of a surface, with Sentinel-2 offering high resolution (circa 10 m) optical data and Sentinel-5P calculating the concentration of atmospheric gases.

Landsat-8 compliments Sentinel-2 by providing high resolution optical and thermal data. The satellite products are combined with the UAV (Unmanned Aerial Vehicle, or drone) products and ground collected phenological/gas measurements to give the best estimation of what is happening.

For a rice field, below is a comparison of what is seen by the UAV (left) and satellite (right) data. The UAV data is of a higher spatial resolution (smaller pixel size) compared to the satellite imagery and is taken closer to the ground so is affected less by the effects of the atmosphere. For satellite data, 50% of the signal can come from the atmosphere, and so it is vital to remove this accurately.

Once the atmospheric effects are removed, both the UAV and satellite imagery have the same algorithms applied. They can then be compared to understand the accuracy of the atmospheric correction and increased error caused by the satellite instrument’s larger pixels.

EcoProMIS UAV and Sentinel-2 RGB pseudo true colour product comparison, data courtesy of Copernicus/ESA for Sentinel-2.


Why Both Satellite and UAV

The advantage of the satellite over the UAV data is that we can assemble a time-series plot. An example of this is shown below for the Leaf Area Index (LAI) product, with the higher values being for when the crop is fully grown.

Time-Series plot of LAI, input data courtesy of Copernicus/ESA and USGS/NASA.



By using the satellite data and tracking over time the LAI, we can create models to predict the yield when the crop is harvested, and by comparing it to the crop modelling outputs, we can understand whether interventions can be undertaken to improve the future crop yield.

It is this combination of satellite data and UAV data, together with ground-based agricultural data and modelling that provides the full picture for EcoProMIS. By combining these multiple data sources in our cutting-edge platform, EcoProMIS will be providing knowledge services on mobile apps to growers, allowing growers to make more informed management choices about their crops and land. It is quite something to be using space technology to contribute to the project’s goal of supporting sustainable agriculture.