Why narrow bands matter

The goal of multispectral imagery

The primary goal of multispectral imaging in agriculture is to detect subtle variation in plant health before visible symptoms appear. For instance, a grower could spot a small reduction in a plant’s chlorophyll content before the leaves start to turn yellow.

Such early detection is possible because the amount of sunlight plants reflect in different wavelengths vary as their health changes. Multispectral sensors then capture and record this variation. However, some sensors capture more precise data than others. If a sensor measures too broad a region of the light spectrum, any subtle variation will be lost.

The benefits of a narrow band sensor

The band center and its bandwidth are key in determining how precise a multispectral sensor is. Broadband sensors, such as converted cameras, measure large “chunks” of the light spectrum, taking the average reflectance in that region to produce one single value. On the contrary, narrow band sensors capture the reflectance of a very specific area of the spectrum. This yields information that is more precise and insightful about subtle changes that may be happening at key points.

To demonstrate, let’s examine the red edge waveband. The red edge band is located between the red and NIR bands and plays a key role in deriving plant chlorophyll content. If a plant is experiencing stress, it’s likely that variation will be the most obvious in the red edge band, as shown in the graph above. When comparing the red, red edge and NIR bands, the largest magnitude of change shows up in the red edge band, therefore it is the most critical to measure. A wide red edge waveband would likely overlap with neighboring bands and suffer from data contamination, resulting in a less precise measurement.

The bandwidths of a multispectral sensor will also impact the vegetation indices resulting from the collected data. Vegetation indices are created by combining different bands to highlight certain features in the vegetation. Since broadband cameras have a wider range of values, the resulting indices won’t highlight subtle variation. Instead, a narrow band camera will collect more precise measurements, resulting in indices capable of detecting changes in the health of the plant at early stages.

Conclusion

In agriculture, broadband sensors, such as converted cameras, work well for the big picture, allowing for the generation of RGB maps and generic indices like NDVI. However, the information provided by these cameras is limited and does not allow further analysis of the field over time.

A narrow band camera, on the other hand, provides more detailed and accurate information by capturing precise measurements of specific wavelengths, which allows for early detection of health issues. This gives farmers more tools and information to enable them to make better management decisions.