Written in collaboration with  Dr. Bogdan Lita is a materials scientist/physicist with experience in using test results to improve processes and products. He holds a PhD in Materials Science and Engineering from University of Michigan, an MS in Physics from Central Michigan University, and a BSc in Physics (Optics) from the University of Bucharest. Bogdan’s focus has been supporting the development of the ASD spectroradiometers by designing and carrying out tests that improve performance.

Malvern Panalytical has a compelling case to show how a non-continuous, typically removable “jumpered” fiber optic cable has a lot more noise than a permanent fiber optic cable…

Airborne remote sensing image data are routinely used to characterize vegetation, minerals, and other materials on the surface of the Earth. Field spectra collected for either hyperspectral sensor calibration or for direct comparison with hyperspectral image data are best collected using ambient solar illumination since this allows the use of the same illumination source between the airborne and the ground instruments. In addition, the ability to use ambient solar illumination enables measurement of geologic surfaces and vegetation in their natural and intact states.

The ability to collect this field data is enabled by low spectroradiometer noise. Low noise instrumentation results in high signal-to-noise ratio (SNR) measurements which are beneficial in field measurements across the whole spectral range (350-2500 nm), but especially in the shortwave infrared (SWIR; 1000-2500 nm) where solar illumination is much weaker than in the visible-to-near infrared (VNIR; 350-1000 nm) region. Here, we discuss the importance of using a continuous optical fiber bundle, or cable, that collects and guides the diffuse reflectance or radiance coming from the sample into the optical spectrographs.

Background

The fiber optic cable is made up of fifty-seven (57) uniformly distributed silica glass fibers. Nineteen (19) of these fibers are 100-micron diameter and are branched to the VNIR portion of the instrument. The remaining thirty-eight (38) fibers are 200-micron diameter and are evenly divided between the two SWIR scanners. All fifty-seven fibers are enclosed in a flexible plastic tube. The fibers are protected by a flexible stainless steel conduit. This is important, as each broken fiber results in a 5% or greater loss. The bend radius should be as large as possible to prevent breakage of the fibers.

Fig. 1 – Performing a visual verification of the fiber optic cable using LED and magnifier

Instrument configuration

The ASD FieldSpec® spectroradiometer combines three spectrographs to measure the diffuse reflectance from a wide portion of the solar spectrum, 350 to 2500 nm. A Si photodiode array spectrograph is used to cover the 350 to 1000 nm spectral range (UV/VNIR), while two rotating grating spectrographs provide coverage for the wavelength range from 1000 to 2500 nm (SWIR 1 and 2). The UV/VNIR detector is a 512 element low dark current NMOS CCD array operated at ambient temperature. Single element InGaAs detectors, thermoelectrically cooled, are used in the two SWIR spectrographs. All three spectrographs utilize concave holographic gratings as the wavelength dispersing elements. The light input to the FieldSpec spectrographs is through a fiber optic bundle, typically 1.5 meters in length. While the use of longer fiber optic cables is possible, the performance will be degraded at wavelengths beyond 2200 nm. As shown in Fig. 2, the optical fibers carrying the light to the spectrographs are packaged as a single bundle exterior to the instrument. Once inside the instrument, the fibers are separated into three bundles which then deliver the collected light to each of the three spectrographs.

Fig. 2 – Schematic showing the routing of the fiber inside the ASD FieldSpec

The arrangement of the fiber optic input that feeds directly into the ASD instrument (i.e. a permanent, or “non-jumpered” cable) holds a few distinct advantages

  1. The fiber optic input allows the user to quickly move and aim the very lightweight fiber optic probe from point to point without having to move the entire spectroradiometer.
  2. Since the fiber optic is connected directly into the spectroradiometer, there is none of the signal losses otherwise associated with detachable couplings, as explained below.

Signal-to-noise ratio (SNR)

Signal-to-noise ratio (SNR) is typically defined as the mean signal level divided by standard deviation of the fluctuations of the signal. The introduction of a “jumper,” or an additional non-continuous, typically removable, fiber cable in between the sample and spectrographs causes significant degradation of SNR. This degradation is primarily due to the loss of light (signal) when it traverses the glass-air-glass interface introduced by the “jumper,” and misalignment between individual fibers in the bundle at this interface. A permanent continuous non-removable cable (as used in ASD instrumentation) has no such interfaces. Permanent and “jumpered” cable spectroradiometer SNRs are shown in Fig. 3. Overall the reduction of SNR due to the introduction of the jumper varies but is more pronounced at shorter wavelengths, for example VNIR. In a field spectroradiometer, the jumper SNR reduction is more detrimental in the SWIR range because the solar irradiance is much lower in the SWIR range than that in the VNIR range.

Fig. 3 – Spectroradiometer signal-to-noise ratio (SNR) plot showing permanent (blue) and “jumpered” (red) spectroradiometer laboratory SNR plotted on a linear scale.

Conclusion

The ASD FieldSpec is a field spectroradiometer especially suited to field measurements using solar illumination across the whole spectral range (350-2500 nm) but especially in the shortwave infrared (1000-2000 nm) where solar illumination is much weaker than in visible to near-infrared (350-1000 nm). The reduction of SNR due to the introduction of the jumper varies, but is more pronounced at shorter wavelengths; the jumper SNR reduction is more detrimental in the SWIR range because the solar irradiance is much lower in the SWIR range than that in the VNIR range.