# Also in the Article

2.2. Laboratory Analysis
This protocol is extracted from research article:
Application of Low-Cost MEMS Spectrometers for Forest Topsoil Properties Prediction
Sensors (Basel), Jun 7, 2021;

Procedure

All chemical analysis was performed by the laboratory of Sachsenforst public enterprise, following german-wide standards for forest soil chemical analysis [33]. Total carbon (C) and nitrogen (N) fractions were determined. For total carbon content the dry combustion method with elementary analysis was applied [34].

Before the spectral analysis was conducted, all samples were dried and sieved with a 2 mm mesh to ensure homogeneous material. To protect volatile compounds of the soil during the drying process, the temperature for the samples was set to 40 $∘$C.

The spectral measurements were acquired following a self-developed protocol, which was created based on literature research and several trial measures. Aims of the protocol were to capture each samples variability, to enable a fail-safe procedure and ensure comparability of the measurements. Per soil sample, two petri dishes were filled with soil material. Each dish was then measured five times, rotating the dish after every measure to increase the acquired variability. Further, this procedure balances the values within the measured area [35]. According to the protocol, the procedure results in ten spectra per sample.

The R language for statistical computing was used for all processing and calculation steps in this study [36].

During the spectral measuring, light scatter can result in effects like baseline shifts or other anomalies. To antagonize these effects, numerous suitable preprocessing techniques exist [37]. First, we smoothed our data using the Savitzky–Golay-Filter. It calculates the sum over a given window, computed as follows in Equation (1):

where $xj∗$ is the new value, N is a normalizing coefficient, k is the gap size on each side of j and $ch$ are pre-computed coefficients, that depend on the chosen polynomial order and degree [38,39]. We used the third polynomial and a window size of 11.

We further normalized our spectra using standard normal variate (SNV). It is calculated as displayed in Equation (2):

where $xi$ is the signal of a sample i, $xi¯$ is its mean and $si$ its standard deviation [37,40]. The above described and applied pre-processing steps were carried out by means of the the R package prospectr [39].

Note: The content above has been extracted from a research article, so it may not display correctly.

Q&A