Spectral Resolution - The key difference between HSI and MSI

Shubhavi
Shubhavi
  • Updated

Spectral resolution refers to the ability of a sensor to detect fine-wavelength intervals. It signifies the sampling rate and bandwidth at which the sensor collects information about the target scene or object. A sensor with high spectral resolution is sensitive to narrow wavelength ranges, providing detailed information about the target. On the other hand, low-spectral-resolution sensors are less sensitive, detecting broader bands across the spectrum and capturing less detailed information.

For example, multispectral imaging sensors split light into several bands, each representing a specific range of wavelengths. Each band may cover a distinct portion of the EMS, such as red, green, blue, and near-infrared. On the other hand, hyperspectral imaging sensors split light into hundreds of narrow spectral bands, allowing for very high spectral resolution and precise differentiation between different targets based on their unique spectral signatures.

 

In essence, hyperspectral sensors provide finer spectral resolution, while multispectral sensors offer broader coverage at coarser spectral resolution.

 

When comparing the Copernicus Sentinel-2 and Pixxel Firefly in capturing the red portion of the electromagnetic spectrum, notable differences arise in their spectral band configurations and resolutions. The red spectrum is technically defined as the wavelength range between 620 nm and 750 nm.

 

Sentinel-2 utilizes a single band with a width of 30 nm, specifically ranging from 650 nm to 680 nm, to capture the red band. However, by imaging a band narrower than the officially defined red range, it overlooks a substantial amount of spectral information, specifically the lower 30 nm and upper 70 nm. Moreover, by employing a 30 nm wide band to capture red light, it combines subtle spectral variations that would be distinguished as separate bands by a more sensitive sensor.

 

In contrast, Pixxel’s Firefly satellites divide the complete red portion into 42 bands, each spanning 3 nm in width.

 

To visualize this, we can look at a non-scientific example: The Sentinel-2 sensor simply detects red (for the sake of this example, we’ll exclude the detection of red edge bands). Meanwhile, the Firefly sensor records 42 shades of red as distinct bands, e.g., Scarlet Red, Cherry Red, Crimson Red, Ruby Red, Burgundy Red, Maroon Red, Wine Red, Vermilion Red, Carmine Red, Raspberry Red, Blood Red, Rose Red, Tomato Red, Fire Engine Red, Brick Red, Mahogany Red, Coral Red, Rust Red, Garnet Red, Terra Cotta Red, Candy Apple Red, Venetian Red, Raspberry Rose, Merlot Red, Persimmon Red, Flame Red, Cranberry Red, Amaranth Red, Cardinal Red, Sangria Red, and so on.

 

We can compare spectral resolution to the spatial dimension of the data: Medium-resolution data (e.g., 30 m) tends to capture more “mixed pixels” compared to finer data (e.g., 0.5 m). Likewise, multispectral sensors capture more “mixed spectra” than hyperspectral sensors.

 


Screenshot 2024-05-29 at 4.08.04 PM.png

Was this article helpful?

0 out of 0 found this helpful

Have more questions? Submit a request

Comments

0 comments

Please sign in to leave a comment.