Data sent with each ASD spectrometer may include the following calibration graphs: NEdL (Noise Equivalent Radiance), Mylar, Mercury Argon, and Kaolinite.* 1
The NEdL graph is a representation of signal to noise ratio of the instrument. This plot displays three regions that correspond to each spectrometer.*2 If you were to look at the raw data from which this NEdL plot was created, you would see that the response for each detector is optimum in the mid-range. The NEdL plot is an inverse illustration of the raw data - giving you the lowest noise level within these three ranges. ASD’s goal is to adjust each of the splice, step, and size values for the corresponding spectrometers to
insure the "best" detector response.
The purpose of the Mylar (1000-2500nm) and the Mercury Argon (350-1000nm) graphs is to confirm and illustrate specific wavelengths. It is very important that each spectrograph registers specific wavelengths accurately, in order to guarantee good spectral data. If any of these specific wavelength targets are not within +/- 1 nm of the wavelength standards, we then know each spectrometer must be re-tuned. However, the graphs of these samples are not meant to check for uniformity between each spectrograph. Both of these samples are very specular - which is good because it is those reflective properties/absorption features that allow us to confirm nm accuracy - however they are not optimal for determining uniformity due to two key reasons:
Variable surfaces (ie specular samples) result in different fields of view across the random fibers for each spectrometer and can, in turn, give rise to steps
The field of view for the Vnir region is slightly larger than the Swir regions - by about 1 degree - therefore, any slight change in viewing geometry and/or a spectra from a specular sample can also give rise to steps
The purpose of the Kaolinite (1000-2500nm) is to not only to confirm wavelength accuracy and sensitivity, but also to give us a look at uniformity. Kaolinite is a uniform, non-specular sample and should represent a smooth, clean spectrum across the full spectral range. Sensitivity and accuracy are measured by the doublet seen at 1404 nm. If you take a look at Kaolinite graph, you will see this doublet along with another absorption feature at approximately 966.8 nm. Do not confuse this absorption feature with the splice between Vnir and Swir1 - this is not the same wavelength. In fact, the splices for this unit are at approximately 1000nm (Vnir and Swir 1) and at 1800nm (Swir 1 and Swir 2) - and should be very smooth and clean in the Kaolinite graph.
Because most natural samples are non-specular, the step that sometimes appears with specular data, such as the Mylar reflectance, is negligible. To guarantee smooth, clean, and accurate data, collect a large number of spectra and average them together in post-processing using ViewSpec™ Pro. The "look" of your sample spectra can also be manipulated by altering the viewing geometry.
1 FSFRs receive all graphs; NIRs receive graphs specific to the wavelengths covered; VNIRs and HHs receive only Mercury Argon and the NEdL
2 The number of representative peaks is dependant upon the type of spectrometer: Full Range – 3 peaks, NIR – 2 peaks, VNIR and HH – 1 peak