How can remote sensing make agriculture more sustainable?

Talk like a remote sensing researcher

Eutrophication — the process that occurs when too many nutrients enter the water (for example due to fertiliser being washed off fields), causing excessive algae growth that leads to a lack of oxygen in the water

Reflectance value — how much light is reflected from a surface

Remote sensing — the science of collecting information about Earth’s surface, oceans and atmosphere from a distance, typically using sensors on satellites, aircraft or drones

Spectral signature — the unique pattern of light that a material reflects or absorbs at different wavelengths

Spectral vegetation index — a mathematical value calculated from remotely sensed spectral data, which indicates vegetation properties such as health or density

When forests and wetlands are cleared for farming, wildlife habitats are lost, biodiversity declines and soils become more vulnerable to erosion. Agriculture damages water quality when chemicals wash into waterways, as pesticides kill organisms and fertilisers causes eutrophication which harms aquatic life. And poor soil management can lead to erosion and loss of stored carbon. Together, these pressures can seriously affect ecosystems and long-term food production.

“Agriculture is essential for food security and livelihoods,” says Dr Angela Kross, a remote sensing scientist at Concordia University who uses Earth observation technologies to study agricultural systems. “Agriculture is also one of the largest drivers of environmental change. It affects biodiversity, soils, water and climate, and is responsible for 10-12% of global anthropogenic greenhouse gas emissions. My research aims to make these impacts visible and measurable so they can be managed more effectively.”

How does Angela analyse remote sensing images?

Remote sensing involves taking measurements from a distance, typically using satellites, aircraft or drones. Angela uses a mix of satellite and field-based technologies to study vegetation and environmental properties at different scales. “Remote sensing technologies measure how the Earth’s surface reflects sunlight in different wavelengths of the electromagnetic spectrum, such as visible light and near-infrared wavelengths,” explains Angela.

At the field scale, Angela collects data using drones equipped with red-green-blue cameras and thermal cameras. Red-green-blue cameras produce true-colour images by recording red, green and blue light, just like human eyes do. Thermal cameras measure infrared radiation and record temperature differences, which allows Angela to identify crop stress and wildfires. At the wider scale, Angela uses images collected by satellites.

To analyse the images produced by remote sensing cameras, Angela interprets patterns called spectral signatures. “Spectral signatures show how different surfaces, such as vegetation, soil or water, interact with sunlight by absorbing, reflecting or emitting energy across various wavelengths,” she explains. “Each material has its own ‘spectral fingerprint’ in the electromagnetic spectrum, allowing me to identify what is on the ground and assess its condition. Interpreting these signatures allows me to monitor plant health and reveal information that would otherwise remain invisible to the human eye.”

Spectral signatures guide the creation of colour composite images and the calculation of spectral vegetation indices. For example, in some false-colour satellite images, near-infrared light (which humans cannot see) is displayed as red. Healthy vegetation reflects a large amount of near-infrared radiation, so it appears bright red, making differences in plant health easy to visualise.

Spectral vegetation indices build on the same idea by converting the complex data from spectral signatures into simple numerical values. “This is done by mathematically combining reflectance values from specific parts of a spectral signature, usually wavelengths where the differences between healthy vegetation, stressed vegetation, soil or water are most pronounced,” Angela explains. The most well-known index is the normalised difference vegetation index (NDVI), which gives a simple score that indicates vegetation health based on how plants reflect red and near-infrared light. Healthy, leafy vegetation absorbs red light for photosynthesis and reflects near-infrared light, so NDVI values are high. In contrast, stressed, sparse or non-vegetated areas have low or negative NDVI values. NDVI is widely used by scientists but it has limitations, so many other indices have been developed to study different crops and environmental properties.

These indices are used as model inputs so satellite images can estimate real-world measurements of variables such as crop water content (an indicator of crop stress). “It’s important that remote sensing models are validated with data from the ground,” says Angela. “For example, we can develop models based on the relationship between spectral indices and chlorophyll concentration (an indicator of eutrophication) measured from water samples, then apply the validated model to satellite images to map chlorophyll concentration in lakes and locate areas contributing to nutrient runoff.”