Achieving large-scale precision with multispectral drone data
What is multispectral imaging?
How does multispectral imaging work?
What do the five bands do?
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). So chlorophyll production in leaves results in a lot of infrared light reflectance, but we can’t see this. Healthy leaves reflect a lot of near-infrared light so tracking it with multispectral sensors is key to managing plant health on a large scale.
So you have blue, green and red bands of light recorded by a multispectral. The near-infrared (NIR) sensor captures the fourth band of light just beyond the visible spectrum, in the near-infrared band.
The fifth sensor captures short-wave infrared bands (SWIR), also known as the red edge band. This band of light sits in the transition zone between the red and NIR wavelengths. By combining the information from these bands, it’s possible to analyze and monitor vegetation, water bodies, soil, and other surface features in detail.
Plants and soil, for example, absorb and reflect wavelengths from sunlight depending on their contents. So reading the light that is reflected via multispectral imaging enables precise monitoring of the contents of plants and soil in order to track plant health over time.
Why use pansharpening in multispectral imaging?
These orchard outputs display the difference between multispectral data captured with and without a panchromatic sensor and pansharpening. Image courtesy MicaSense.
What multispectral sensors capture as a strength in each distinct band of light, they lose in spatial resolution. They can give you the colors reflected in a specific area, but not the fine detail that an RGB image would. That presents a drawback in terms of orientation and understanding the context when you are analyzing different colored splotches of light in a field.
To make up for this, a lot of sensors, like the MicaSense RedEdge-P for example, now pair their high-quality multispectral capture with a panchromatic lens.
A panchromatic lens captures visible light and features in black and white and at a resolution that allows for feature recognition. To improve the spatial resolution of multispectral imagery, a technique called pansharpening is used. It consists of merging the two different imagery types of the same area and blending them into a single, colored image with contextual details.
On the one hand, it contains the high-resolution spatial detail of the panchromatic band, and on the other hand the color information from the multispectral bands. This allows for both feature recognition and deep assessment of the mapped area for visible as well as invisible light band readings.
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.
What solutions does multispectral imaging offer today?
- More frequent and accurate contextual analytics for precision management: When you combine the high spatial resolution of the panchromatic sensor with the spectral information from the multispectral bands, you get deep analytics via a range of reflected light and the context to identify features precisely.
- Vegetation indices for plant health assessment and management: Vegetation indices such as the normalized difference vegetation index (NDVI) and the soil-adjusted vegetation index (SAVI) can be drawn from multispectral data. These are critical to assessing plant health, biomass, drought and canopy density by measuring the difference in reflectance between the red and near-infrared bands, indicating the presence and condition of vegetation.
- Soil and water analysis for quality control, erosion monitoring and optimized irrigation: Multispectral sensors can analyze soil composition and moisture levels by measuring reflectance in specific bands that change based on soil and water properties. This capability aids in understanding irrigation levels, soil types, states of erosion, water quality and even turbidity—i.e., particle concentrations—in aquatic environments.
- Land cover classification monitoring and management: Multispectral imagery helps you see what is happening on the area you are mapping precisely so that you can classify sections of it according to, for example, forests, urban areas, water bodies, and agricultural fields. Since each type of land has a distinct spectral signature, you get an accurate reading to classify precisely across a mapped area. Large, area multispectral maps are a time-saving boon environmental monitoring, urban planning, and resource management.
- Change detection: By comparing multispectral images acquired at different times, changes in the environment can be detected and analyzed. This includes monitoring deforestation, urban expansion, crop growth cycles, and the effects of natural disasters, providing valuable insights into temporal dynamics.
- Estimating surface temperature: Multispectral sensors can capture thermal infrared information that helps assess the heat distribution across landscapes. This is useful for monitoring urban heat islands, and studying environmental processes such as evapotranspiration—i.e., how water cycles back up into the air as evaporation or via plants absorbing and releasing it.
- Mineral and material identification: Different minerals and building materials have unique reflective properties. Multispectral sensors provide analytics of specific reflected bands so you can identify and classify the materials in any area under study. This is great for geological study, mining exploration and construction material analysis.
- Urban mapping: Multispectral data avails detailed building and infrastructure maps. This helps in identifying buildings, roads, vegetation, and other urban features, which is crucial for urban planning, development, and management.
- Agricultural monitoring: Multispectral sensors are extensively used in agriculture to monitor crop health, predict yields, and manage fields. By analyzing specific bands that signal key plant properties, farmers can detect stress, disease, and nutrient deficiencies. Multispectral drones and the speed at which they capture this data are proving increasingly vital to crop management best practices.
- Disaster assessment: Multispectral imagery is invaluable for assessing areas affected by natural disasters like floods, fires, and earthquakes. It provides detailed information on the extent of damage, helps in planning relief efforts, and supports recovery and mitigation strategies.
Why use a multispectral drone?
Much better accuracy
Satellite multispectral resolution will be measured in meters. Whereas with the more advanced multispectral drone sensors, like the MicaSense RedEdge-P, you can get down to 2 cm resolution and 3 cm absolute horizontal accuracy.
This makes a massive difference when it comes to precision analysis of details and the resultant actions you take. For example, if you have cm-level insights about disease in plants, you can apply solutions to these areas precisely without damaging neighboring growth. In any situation, better resolution will avail a more accurate application of solutions.
Watch the introductory webinar on the MicaSense RedEdge-P and its use cases.
More flexibility
Satellite data costs money for every time it is collected from a service. And the amount of time it can be collected from above your area is obviously limited by the Earth’s motion relative to the tech.
The other obvious factor here is cloud cover. Satellites cannot penetrate it and this can be a huge block to obtaining data when and where you need it. Multispectral drone data capture is possible so long as it’s daylight and the weather conditions do not trespass over the specs of the platform.
High-efficiency drones make multispectral capture even more feasible because they capture so fast that you can get what you need even in tight weather windows.
More cost effective when large areas or repeat insights are needed
Multispectral satellite data can cost an average of 40 USD per km2, with a minimum order of 25km2 or even 100km2. If you need weekly insights on an area, this adds up.
You can even get a full drone platform in a growing season on a farm, plantation or orchard for this. Environmental monitoring projects can span up to 100 km2, so this is already the price of a platform for just one survey’s worth of data.
How a multispectral drone camera transforms your workflows
The biggest differences you will see with a multispectral drone are the following:
- Data on-demand. On small-area or inspection projects, a multirotor can be deployed in a pinch to capture data. For larger areas, a good VTOL system will provide you with quick coverage and analytics.
- Regular insights to optimize and pivot when changes in vegetation or the environment occur. Imagine capturing data every week instead of every month. You can then take preventative measures based on precise areas, thanks to better resolution.
- Faster response times. When you need records of damage or changes due to natural disasters or accidents, time is of the essence. Drones like WingtraOne GEN II make it possible to fly with them and their batteries to a destination without special clearance.
- Lower overhead over the long run. With a reliable, durable system, you can map day after day, week after week and face minimal costs for a wealth of data that ensures precise project management. What you save in potential errors and outsourcing of data, you also save as the system pays for itself.
What is the best multispectral drone platform?
The answer to this question depends on your needs. If you require multispectral data for vertical assets or small areas, you can use a multirotor setup. But as your area gets bigger, you will see that this is not efficient or cost effective. A fixed-wing with VTOL capabilities is your best bet for larger areas.
Since multispectral is a complex form of data capture, we recommend carefully choosing your payload. If you compare all the drone multispectral payloads on the market today, you will be hard-pressed to find one better than MicaSense. The RedEdge-P is a great all-around choice for getting pan-sharpened multispectral data.
FAQ
Is multispectral imaging always done by a drone?
How much land can a multispectral drone capture?
The amount of land a multispectral drone can capture depends on the drone model, as well as other factors, including flight altitude and battery life. For most purposes, you will need to cover a lot of land fast due to weather windows and the types of projects that multispectral imagery is used for.
The Wingtra platform enables you to get data in tight weather windows efficiently and reliably. For example, a WingtraOne GEN II with a MicaSense RedEdge Panchromatic (RE-P) multispectral payload camera has a coverage area of 90 ha at 60 m altitude, or 160 ha at 4 cm/px GSD flying at a height of 120m.
The MicaSense RedEdge-P combines a panchromatic sensor with five narrow bands to produce high-resolution multispectral and RGB from one flight.