Affect of Different Preprocessing Methods on Principal Component Analysis for Soil Classification

Abstract

Fast classification of soil with different texture is essential for site-specific application of different inputs into farmland. Total 203 soil samples with five textures were collected from Silsoe Experimental Farm, Cranfield University, England. Using a Vis/NIR spectrophotometer (Tech5, Germany), spectra of soil samples were recorded for the study. Amongst the pre-processing methods, smoothing with moving average(MA), multiplicative scatter correction(MSC), standard normal variation(SNV), de-trending(DT), baseline correction(BC) and derivatives( 1 st and 2 nd ) were mainly investigated. PCA was applied for evaluation of the efficiency of different pre-processing methods on soil spectra. The sore plot of PCs shows that 1 st derivative can help separate all textures much more effective than other methods. According to the cumulative variance of first 8 PCs, the various combinations of MA, MSC, DT and BC can be regarded as good methods. The worst is 2 nd derivative due to its inducing much more noise. The study suggests that 1 st derivatives should be firstly concerned amongst various pre-processing methods for the classification of soil textures.

Efficient Fuzzy Controller to Increase Soybean Productivity

“…As a consequence of exceeding field capacity, there will be infiltration and nutrient loss eventually. Then soil type must be identified and based on each kind, the proposed system can adjust the amount of water required [10] [11].…”

Near Infrared Spectra and Soft Independent Modelling of Class Analogy for Discrimination of Chernozems, Luvisols and Vertisols

“…SIMCA models were developed using different data pretreatments that included smoothing, multiplicative scatter correction (MSC), standard normal variate (SNV), baseline correction (BC), normalise and first and second derivatives. 24,37 The first and second derivative and smoothing transforms are based on a Savitzky-Golay 38 polynomial filter. The number of significant PCs in the model of each class was evaluated using leave-one-out cross-validation.…”

“…These SIMCA models correctly classified 100% of the samples in the test set included as "unknown" samples ( Table 2). Yang et al 24 also reported that first derivatives of NIR spectra produce the best results for the PCA classification of soil texture as compared to MSC, SNV, BC and second-derivative pre-processing methods.…”

“…Generally soil colour, soil texture, carbon content, iron content and form, carbonate, structure and consistence are used along with a range of chemical and physical characteristics to distinguish and identify soil horizons and to group soils according to the soil classification system -soil taxonomy. As far as we are aware, several studies have reported relatively successful soil qualitative analysis such as soil textural classification, 23,24 soil colour classification 25 and, a few of them, soil taxonomy [26][27][28] using visible and NIR soil spectra. The first piece of research that demonstrated that the soil horizons and soil classes can be distinguished in a soil classification system with visible-NIR spectra was performed by Viscarra Rossel and Webster.…”

Creation of Soil Spectral Library for Marathwada Region

“…The soil spectral signature is the radiation reflected as a function of the wavelength [9]. Recent researchers have indicated that visible and near-infrared, diffuse reflectance spectroscopy (VINR-DRS) and specific band from hyperspectral sensor data [10,[11][12][13][14][15]. Near-Infrared spectroscopy can be used to analyze heavy metal contents in soil such as zinc (Zn), copper (Cu), lead (Pb), chrome (Cr), and nickel (Ni) [16].…”