MAPIR Environmental Monitoring
MAPIR offers a complete environmental monitoring solution to measure changes in how light reflects off of objects such as plants and soil. The spectrum of light that is reflected is known as the material's reflective spectral signature, and can tell you information about the material from afar (remote sensing).
CONTACT US today to talk to a multispectral remote sensing expert. We are happy to assist with explaining our technology and helping you select the best solution for your project.
To accurately measure the reflected light you need three types of hardware:
Spectroradiometer Light Sensor: measures the ambient light spectrum (340 - 1010 nm) illuminating the objects of interest. The light source can be the sun or artificial (LED, fluorescent, halogen, incandescent, etc). Light measurements are recorded against GNSS (GPS) time, and as the light changes our post processing software will adjust the image reflectance calibration.
Reference Reflectance Targets: provide a lab-measured reference standard for reflectance and are used to correlate the image's pixel values to percent reflectance. Percent reflectance is a measurement of how much of the ambient light is bouncing off the object, and the differences in intensity is an object's spectral signature. Using a reference standard with the light sensor allows the reflectance data to be compared across time and different lighting conditions.
Multi-spectral Imaging Cameras: capture photos of the reference targets to establish the relationship between image pixel (DN) values and percent reflectance. After post processing each pixel in the processed image channel represents the percent of light reflected by the object in the spectrum(s) of light that the camera is sensitive to.
The light sensor measures the ambient light spectrum and the reference targets tell us how that light is reflected, so that the multi-spectral images can be calibrated and compared over time and location.
A single light sensor (spectroradiometer) and reflectance reference target is required to process images from any number of multi-spectral cameras. The light sensor records the ambient light spectrum (340 - 1010 nm) which the multispectral cameras are capable of recording.
The light sensor can be used as a standalone device (left) or connected directly to the multispectral cameras (right, below):
The light sensor will measure the ambient light spectrum every second using a high resolution spectroradiometer.
Below is a typical measurement from the sun on a clear day. Each 1Hz measurement contains 134 values, represented by the dots on the graph below. The line is the spectral signature.
While the light sensor is recording the ambient light spectrum, the multispectral cameras are used to capture images of the objects you want to measure the reflectance of. Images of the reference targets are also captured for post processing.
The reflectance reference target regions are measured by our software and the necessary reflectance corrections are calculated. The changes in the ambient light recorded by the light sensor are adjusted for during the reflectance calibration.
Our Survey3 multispectral cameras come in two lens options (W = wide, N = Narrow) and the below five filter model options. Depending on the filter model, up to three separate spectrums will be captured in each image.
| Camera Filter Model | Image Channels 1,2,3 | Spectrum Peaks |
| RGN | Red, Green, Near Infrared (NIR) | 660nm, 550nm, 850nm |
| OCN | Orange, Cyan, Near Infrared (NIR) | 615nm, 490nm, 808nm |
| NGB | Near Infrared (NIR), Green, Blue | 850nm, 550nm, 475nm |
| RE | Red Edge | 725nm |
| NIR | Near Infrared (NIR) | 850nm |
| Spectrum Peak | Spectrum Color | Spectrum Width | Filter Model | Image Channel |
| 475nm | Blue | 15nm | NGB | 3 |
| 490nm | Cyan | 36nm | OCN | 2 |
| 550nm | Green | 15nm | NGB, RGN | 2 |
| 615nm | Orange | 42nm | OCN | 1 |
| 660nm | Red | 15nm | RGN | 1 |
| 725nm | Red-Edge | 23nm | RE | 1 |
| 808nm | Near Infrared | 50mm | OCN | 3 |
| 850nm | Near Infrared | 30mm | NGB, RGN, NIR | 1,3,1 |
Using the OCN (Orange, Cyan, NIR) filter model as an example, you can see the filter transmission graph below. An example of a processed 3-channel image from the OCN camera, with channel 1 (615nm), channel 2 (490nm) and channel 3 (808nm) is below.
The image shows mostly healthy grass with some dead leaves. Splitting apart (decomposing) the processed image we can see the three image channels (left to right): orange, cyan, and near infrared (NIR):
The above three single spectrum/channel images are each originally 12MP (4000 x 3000px), and are similar to the results you would obtain from a multi-sensor multispectral camera (MAPIR Kernel2, Micasense RedEdge, etc). Each channel is completely separate from one another, and there is no overlap or cross-talk after processing. Since the three images were produced from a single image sensor they can be captured at any distance, and are already perfectly aligned to one another.
You can then process the image channels using any multispectral index you prefer. A common vegetation index for plant health/vigor is NDVI, which compares the contrast differences between red/orange and near infrared (NIR) light. The green-yellow-red color gradient (lut) applied below shows healthy grass as green and unhealthy grass (or non-plants) as yellow to red.
Using the spectral data from the light sensor keeps the relationship (contrast) between the image channels consistent when the ambient light changes. The effect from shadows is mostly removed unless the shadow is very dark.
Our light sensor (spectroradiometer) maintains the reflectance calibration when the ambient light changes.
Below you can see a 3.5 hour time lapse animation with an image captured each minute during partly cloudy weather. The scene is composed of healthy grass with some dead leaves. The images have been processed with a color gradient to show a range from healthy plants (green) to unhealthy/dead plants (red) using the NDVI formula. The 4 sections show the original images, the NDVI when not calibrated, the NDVI when calibrated only with our reference targets and the NDVI when calibrated with both our target and light sensor.
Below: Survey3W OCN, 3.5 hour time-lapse (1 photo/min, 206 frames), Locked Exposure, Cloudy Weather, NDVI LUT 0.2-0.8 [Resolution Reduced for Web Viewing]
Notice how the original images change in brightness constantly as the clouds move in front of the sun. Our light sensor tracks the changes in the ambient light and keeps the index (NDVI) contrast measurements consistent.
Below: Survey3W RGN, 3.5 hour time-lapse (1 photo/min, 206 frames), Locked Exposure, Cloudy Weather, NDVI LUT 0.2-1.0 [Resolution Reduced for Web Viewing]
Survey3W RGN (Left to Right) Original JPG, Calibrated RAW, NDVI LUT 0.2-1.0
Below: Survey3W RGN, Alfalfa Field 400ft, Cloudy vs Sunny Weather, NDVI LUT 0.2-1.0
Using the MAPIR light sensor on an aerial vehicle allows you to capture images even during cloudy weather. The darkening effect from clouds is greatly reduced, allowing consistent data capture in most lighting conditions. The above NDVI images show rows of alfalfa agricultural crops from 400ft (120m), and consist of many individual images that have been stitched together to make a map. Even though there was cloud cover, with the sun going behind the clouds often during the mapping mission, the resulting NDVI contrast map is similar to the next day when the sky was clear.
