Multispectral drone data is quietly revolutionizing agriculture and forestry—are you missing out?

Multispectral data and its use in vegetation analysis is not a new thing. Agronomists, farmers and forestry professionals as well as environmental researchers have depended on this information for decades, thanks to satellite imagery. However, satellites introduce their limits. Namely: lower resolutions, gated access and interruptions with cloud cover. Drones with multispectral cameras, however, address all of these limits. Suddenly, farmers, foresters and researchers have a powerful, on-demand solution in their hands.

Indeed, the tech revolution is taking the agricultural industry to an entirely new level, since rapidly-collected data can now be fed into automated machines to lower the overhead on irrigation and the application of fertilizers and pesticides. Likewise, forestry and environmental fields are impacted with this highly-detailed, bulk data on large areas of land or sea. As climate change introduces fast shifts in planting zones, water availability and pests, drones emerge as dynamic tools to give everyone a comprehensive view of what’s happening and the means to adjust tactics.

Before running out and getting a UAV multispectral camera, however, it behooves us to know how all of this works so we can get the right equipment for the best multispectral drone ROI over the long run. 


How does multispectral drone data work, technically?

Only a fraction of wavelengths along the electromagnetic spectrum are visible to the human eye.

The electromagnetic spectrum is a wide range of wavelengths, each carrying information. Only a fraction of these are visible to the human eye. When we look at anything, what we can see is a reflected color spectrum of reds, greens and blues (RGB) that we interpret as any combination of the colors of the rainbow based on their wavelengths. Normal RGB cameras filter wavelengths for the information that we can see. Multispectral cameras are equipped with lenses and filters that pick up wavelengths beyond the visible spectrum—in the direction of infrared wavelengths.

Why is this important for analyzing crops and soil? Plants and soil absorb and reflect wavelengths from sunlight depending on their contents. For example, when a plant is healthy and engaged in photosynthesis, it will absorb a lot of red and blue light and reflect green and a lot more infrared light.  

When you look at a chlorophyll-rich (green) plant, you are seeing the wavelength of light that the plant reflects, not the ones it absorbs (blue and red). Chlorophyll production in leaves results in a lot of infrared light reflectance, but we can’t see this. Multispectral sensors can record this so we can. This is useful, since reduced infrared reflectance signals lower chlorophyll production before we see it with the naked eye, so we can proactively track plant health via multispectral UAV sensors.

light wavelengths reflecting off leaves

Healthy leaves reflect a lot of near-infrared light, which we can’t detect but multispectral sensors can. So this (red to NIR) part of the spectrum is key to tracking plant health.

Researchers and agronomists have been working with multispectral data for decades, and they’ve organized the reflectance ratios into indices that are used for analyzing crops and detecting anything from irrigation stress to pest infestations to weeds. A couple of common indices are the Ratio Vegetation Index (RVI) and the Normalized Difference Vegetation Index (NDVI), which is the most common index in remote sensing.

Since red and infrared light are right next to each other on the spectrum, and they are treated the opposite way in healthy plants, the zone right between them is called red edge. The Normalized Difference Red Edge (NDRE) index is a powerful analytical tool for reading this band and detecting changes in plants at their earliest stages. 


The three kinds of resolution that matter.

Spatial resolution is where UAV multispectral information shines, because it refers to the pixel density, resolution and accuracy of the data. This is the key differentiator when comparing spectral data across the two methods of capture, because the resolution of satellite data lingers around a meter, while UAV can pick up details down to a centimeter.

Temporal resolution implies how frequently multispectral information is recorded on the same area. With drones, this resolution can be increased to whatever degree necessary to achieve management aims.

Spectral resolution highlights how much of the spectrum is being captured with a given sensor to achieve different outputs for analysis.


Exploring the agricultural benefits of multispectral drone data

Farming is inherently close to the ground. Yet even with the industrial revolution, a huge leap forward happened, from walking entire fields on foot with livestock, to driving them in large part with tractors. The latest leap forward is happening now. While you still need to get out in the field, flying over crops with a drone and a good multispectral UAV sensor produces maps rich with data in the form of spectral information that saves immeasurable amounts of time (to scout every plant on foot or with a tractor) and resources (water leaks, or misapplied pesticides and fertilizer).

Here we’ll go specifically through some of the main ways that drone data saves time and money well beyond the investment in equipment and training.


Vegetation health mapping/crop monitoring

Fact: You can look at a plant, see it is green and not know it is under stress that could impact its yield. Beyond what you can see with your eyes, multispectral sensors can pick up invisible spectral information that helps you evaluate the real status of vegetation. With the red edge band, in particular, you can interpret chlorophyll levels in individual plants and compare them over time to respond and manage the environment for best outcomes. You can take preventative action during the early stages of a pest infestation, disease onset, or nutrient deficiency. And you can track the progress of your approach with regularly-updated multispectral UAV maps. 


Insurance claims

With climate change a looming consideration, drought, disasters and invasive species and pests are an issue worth investing in to handle. Even storms can damage crops and impact yields, so farmers invest in insurance to protect their income stream if the harvest is hit by a natural disaster. With multispectral data, you can speed up the claims process by offering high-accuracy, clear before-and-after maps of impacted lands. 


Disease detection

As discussed above, the first signs of disease and related plant stress are apparent through multispectral data analysis. The red edge band, in particular, avails the first inklings of change in chlorophyll production. You can analyze data for changes to this band and assess the part of the crop where they might show up. If disease is a culprit, you can take action fast and prevent its spread, saving the money to replant a larger area and on lost yields.


Irrigation/water management

Water is precious, especially in agriculture. While in dry climates, irrigation is critically important, in some crops it’s actually important to stress the plant to a controlled degree. With multispectral drone data, you can see how plants are responding to the amount of water available to them all across a given field. Return on investment results due to healthy crops, less water expense when not needed and identification of costly leaks.


Pest detection

As in the case with disease detection, pests often leave hallmark traces of activity. When these indications are spotted early via multispectral data, you stand to save a lot of money and time through targeted pesticide application, which prevents their spread efficiently. As a common feature of climate change, multispectral UAV data can help precisely track damage from invasive species as well.


Weed detection

Weeds are plants, too. But they have different light reflection patterns than the crops they visit. So multispectral data can isolate them easily. It’s important to remove weeds as soon as they start growing so that they don’t suck up nutrients that the crops need to thrive. Additionally, weeds provide a home to pests and disease, so its best to get them out quickly. Saying this, herbicide application is expensive and best kept to a minimum. So accurate data helps to target the applications where they belong and protect the environment.


Smarter management of forests

As homes to wildlife and natural carbon sinks, forests are an invaluable resource worldwide. They are also under threat for a variety of reasons, including invasive species and wildfires. Monitoring forests is therefore an important task, and the better the data, the more effective this will be. Let’s look at a few key ways multispectral drone data is proving to be a key asset in the field of forestry, as well as some illustrative outputs to demonstrate.


Tree density

This is a big concern in forestry, since higher density tree stands may render some individuals less able to pick up nutrients. They can then die back/dry out. And this can make a forest prone to wildfires. Multispectral data can be used as the foundation of NDVI analysis of areas to determine biomass and forest density. This is a huge improvement over terrestrial methods which assumed a studied area would resemble the rest of the forest.


Stand count

Stand count is as it sounds–how many trees are standing in a given area. This is important to determine the management approach to the forest. It’s tricky to do with RGB, because competing vegetation can confuse the analysis. Multispectral cameras availing high-resolution red edge and near-infrared information can help foresters parse out what kinds of trees or plants are in a given area. Machine learning can be applied here so that stand count accuracy improves and money is saved on terrestrial counting and overall management.


Tree health

Diseases, fungus, pests, nutrient deficiencies—all of these are realities in a forest. High-accuracy multispectral data, especially including the red edge band for early identification of chlorophyll fluctuations, allows for frequent data collection on individual trees so that those managing the forest can respond to any threats and adjust approaches to prevent wide-spread problems and loss. 


In summary