 ## Predefined Spectral Descriptors

Spectral descriptors form the bases of all models used within . They provide the benefit of introducing basic chemical or spectroscopic knowledge into the calculation of the independent variables used for modelling relationships in images. The term "spectral descriptor" is a general term designating the specification of an independent variable. Spectral descriptors may consist of something as simple as unprocessed intensities at a particular wavelength or mass, or maybe something as complicated as a mathematical function derived from several (if not all) intensities of a spectrum. Additional background information on spectral descriptors can be found here.

Currently, supports the following spectral descriptors:

 Descriptor/Button Supported Spectral Types Explanation ABL IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Area: sum of intensities between two reference points with the baseline subtracted. The user has to specify the two reference points b1 and b2 to calculate the baseline. In order to avoid a baseline too sensitive to noise the parameter "Nb" defines the number of surrounding data points to obtain the baseline reference points. ALV IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Area: sum of all intensities between two reference points; the baseline is defined by the level of the reference point b1. In order to minimize the influence of noise in the data, the parameter "Nb" defines the number of surrounding data points used to obtain the baseline reference point. Please note that areas below the reference level are subtracted. ARW All Area: the sum of intensities between two reference points; no baseline is subtracted. BBL All The logarithm of the ratio of two peak areas (with the baseline subtracted) weighted by the area of the second peak. The baseline is calculated separately for each of the two areas. In order to minimize the influence of noise in the data, the parameter "Nb" defines the number of surrounding data points used to obtain the baseline reference point. BRW All The logarithm of the ratio of two peak areas weighted by the area of the second peak; no baseline is subtracted. CEN IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Centroid: the position of the centroid of the baseline-subtracted spectral line within the boundaries b1 and b2. The value of CEN may assume values between 0 and 1. COO IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Correlation with a positive triangle template peak. The height of the triangular template peak is always 1.0, its shape is controlled by the points [b1,0], [a1,1], and [b2,0]. Please note that in contrast to the TC descriptor this descriptor simply returns the correlation (no weighting, no significance test). DV1 IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. First Derivative: the smoothed first derivative of the spectral curve at a particular position (a1). The smoothing of the first derivative can be controlled by the number of data points used for the polynomial smoothing function and is determined by the parameter halfwidth (number of smoothing points = 2*halfwidth+1). Please note that the first derivative degrades towards zero at both ends of the spectrum. For reliable use of the DV1 descriptor the position a1 should be in the interval [HW,L-HW], with HW being the halfwidth and L being the length of the spectrum (= number of layers of the corresponding spectral group). GC IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Significant positive correlation (level of significance = 0.01) with a general Gaussian template peak, multiplied by the signal area. Please note that this descriptor is set to zero if the correlation is either negative or not significant. The signal area is calculated by subtracting the area below the baseline from the integral between the reference points b1 and b2. The height of the template peak is always 1.0, its shape is controlled by the points [b1,0] and [b2,0] and the parameter "flatness". A flatness value of 1 correponds to a Gaussian bell curve, higher values flatten the top of the curve. GCI IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Significant positive correlation (level of significance = 0.01) with a general inverse Gaussian template peak, multiplied by the signal area. Please note that this descriptor is set to zero if the correlation is either negative or not significant. The signal area is calculated by subtracting the area above the baseline from the integral between the reference points b1 and b2. The height of the template peak is always 1.0, its shape is controlled by the points [b1,0] and [b2,0] and the parameter "flatness". A flatness value of 1 correponds to a Gaussian bell curve, higher values flatten the top of the curve. LAR All This descriptor counts the values above and below a threshold within the interval [b1,b2] and calculates the logarithm of the ratio of the two counts: LAR = ln((1+cnt1)/(1+cnt2))/ln(cnt1+cnt2+1). The range of the LAR descriptor is limited to the interval [-1.0, 1.0]. PBL IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Peak: single intensity at a particular wavelength or mass; the baseline is subtracted. The user has to specify the two reference points b1 and b2 which are used to calculate the baseline. In order to avoid a baseline too sensitive to noise in the data, the parameter "Nb" defines the number of surrounding data points used to obtain the baseline reference points. PLV IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Peak: single intensity at a particular wavelength or mass; the baseline is defined by the level of a single reference point b1. In order to minimize the influence of noise the parameter "Nb" defines the number of surrounding data points used to obtain the baseline reference point. PRW All Raw peak: single unprocessed intensity at a particular wavelength or mass; the baseline is not subtracted. This kind of descriptor is most suitable for baseline corrected spectra. RBL IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Peak ratio: the logarithm of the ratio of two peak intensities at two different wavelengths weighted by the intensity of the second peak. The peak intensities are measured relative to a baseline which is obtained from two reference points. Please note that values of RBL are restricted to a minium or maximum value which can be set in the Preferences dialog. By default the RBL values are restricted to +/- 4.6052 (which is equivalent to a dynamic range of 1:100). RLV IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Peak ratio: the logarithm of the ratio of two peak intensities at two different wavelengths weighted by the intensity of the second peak. The peak intensities are measured relative to a horizontal baseline which is defined by a reference point (intersection of the cursor b1 and the signal). Please note that values of RLV are restricted to a minium or maximum value which can be set in the Preferences dialog. By default the RLV values are restricted to +/- 4.6052 (which is equivalent to a dynamic range of 1:100). RRW All Raw peak ratio: the logarithm of the ratio of two peak intensities at two different wavelengths weighted by the intensity of the second peak. Please note that values of RRW are restricted to a minium or maximum value which can be set in the Preferences dialog. By default the RRW values are restricted to +/- 4.6052 (which is equivalent to a dynamic range of 1:100). TC IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Significant positive correlation (level of significance = 0.01) with a positive triangle template peak, multiplied by the signal area. Please note that this descriptor is set to zero if the correlation is either negative or not significant. The signal area is calculated by subtracting the area below the baseline from the integral between the reference points b1 and b2. The height of the triangular template peak is always 1.0, its shape is controlled by the points [b1,0], [a1,1], and [b2,0]. TCI IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Significant positive correlation (level of significance = 0.01) with an negative triangle template peak, multiplied by the signal area. Please note that this descriptor is set to zero if the correlation is either negative or not significant. The signal area is calculated by subtracting the area below the baseline from the integral between the reference points b1 and b2. The height of the triangular template peak is always -1.0, its shape is controlled by the points [b1,0], [a1,1], and [b2,0]. VAR IR, UV/VIS/NIR, THz, Raman, X-Ray Diff. Variance of the first derivative of the spectral line within the boundaries b1 and b2. The value of VAR is normalized to the total variance of the spectrum and may assume values between 0 and approx. 1, in certain rare cases the value can become 2 or higher.

 Hint: The following video explains the basic ideas behind spectral descriptors.