Quantification Of Soil Properties Research Articles

Classification and Total Carbon Determination of the Soils Using RGB Digital Images Combined with Machine Learning

ABSTRACT An alternative approach for soil characterization is proposed, using digital images and machine learning. The present work investigated the Red, Green, and Blue (RGB) pattern with Artificial Neural Network (ANN) and Random Forest (RF) in the classification of Oxisol and Inceptisol collected in South Brazil. Furthermore, we also built ANN, RF, Partial Least Squares (PLS), and Support Vector Machine (SVM) models from digital images to quantify Total Carbon (TC) content in soil samples. Preliminary evaluations considered the hypothesis that alternative methods of classification and quantification of soil properties might provide results comparable to traditional methods. The TC content of 91 soil samples, 40 from Oxisol and 51 from Inceptisol were determined by dry combustion. ANN and RF as machine learning algorithms were applied to classify the two kinds of soils. The classification quality was assessed by the percentage of correctly classified instances. The results suggest using digital images and machine learning algorithms as a new methodology to soil classification. The best machine learning method was RF which can be used to classify soil types by the RGB channel colors of the soil images. The use of digital images and RF is a clean (eco-friendly), rapid, and affordable strategy analysis for TC quantification in soils. Therefore, the proposed methods represent an advancement in soil classification, such as the determination of the TC content of soils.

Assessing the spatial variability of the soil properties in geologically complex Jawali region of north-west Himalaya, India

ABSTRACT The quantification of soil properties and their spatial distribution is imperative for increasing crop productivity and suggesting the appropriate management strategies. The present study is based upon the hypothesis that the geologically complex landscape of the Indian Himalayan Region may lead to the varied distribution of the soil properties. This study appraises the influence of complex geological environment on some of the selected soil properties, viz., soil pH, electrical conductivity, organic carbon, nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg) and diethylene triamine penta acetic acid-extractable micronutrients, viz., zinc (Zn), iron (Fe), manganese (Mn) and copper (Cu) by delineating their spatial dependency using the geostatistical approach. The surface maps were prepared using the ordinary kriging technique along with semivariogram modelling. The value of the coefficient of variation for the calculated soil properties ranged from low to high (10.99% to 62.83%), indicating the heterogeneous application of fertilisers, management practices and land-use pattern. The principal component analysis aggregates the variability into four principal components. The nugget-to-sill ratio of soil properties, ranging from 25% to 75%, indicates the moderate dependency. The surface maps represent the influence of the complex geological environment on the spatial distribution of soil physico-chemical properties. The results also reveal that the variations in the soil properties were mainly due to the combined effect of intrinsic as well as extrinsic factors. The spatial distribution maps can be used for site-specific nutrient intervention for better resource management and environmental protection.

Predicting soil permanganate oxidizable carbon (POXC) by coupling DRIFT spectroscopy and artificial neural networks (ANN)

Infrared spectroscopy has transformed soil property quantification by enabling low-cost, high-throughput analysis of soils, enabling mapping and monitoring of this non-renewable resource. However, less evaluated are newly emerging indicators of soil health. Furthermore, as soil spectral libraries expand in size, commonly employed linear models such as partial least squares regression (PLSR) may be challenged by the number and diversity of spectra. Artificial neural networks (ANN) are an emerging deep learning approach that can offer advantages in quantification of soil properties by utilizing non-linear relationships among spectra and soil components. We compared ANN versus PLSR models for predicting an increasingly used soil health indicator, permanganate oxidizable C (POXC), as well as more routinely predicted soil variables (e.g., clay, soil organic C [SOC]), across a gradient of soil organic matter furnished by a deforestation chronosequence in Kenya (n = 144). Candidate ANN architectures were first methodologically evaluated and described to identify best-practices for the application of ANN to soil spectroscopy. Predictions by the resulting ANN relative to PLSR were similar or slightly improved for routinely measured variables that represent soil organic matter (SOC, C:N) and physical properties (clay, silt, sand, bulk density). The accuracy of POXC predictions were similar for ANN (RMSE 102 mg kg−1) and PLSR (RMSE 106 mg kg−1). However, models drew on shared but also distinct wavenumbers, indicating differential use of information in soil infrared spectra by non-linear versus linear chemometric models. Even in relatively small spectral datasets of similar soil types expected to favor PLSR, ANN shows comparable predictive performance. To help guide future applications of ANN in soil spectroscopy, we propose a systematic procedure to select ANN model hyperparameters.

Estimation of soil properties from the EU spectral library using long short-term memory networks

Recent advances in Deep learning altered the traditional way of data analysis in various domains. This study focuses primarily on the quantification of soil properties from the hyperspectral data obtained from LUCAS using Long Short-Term Memory Networks based deep learning model. Hyperspectral data is sequential in nature due to which LSTMs are preferred as these are predominantly used for sequential problems. In addition to this, Principal Component Analysis is also performed on the LUCAS dataset to reduce the dimensions for efficient calibration of LSTMs. Better performance is achieved using the proposed framework with best R2 of 0.94 in case of Organic Carbon and the effectiveness is shown by comparing it with existing models such as PLSR, SVR, PCR, MLR and SWR.

Digital Soil Mapping of Soil Properties in the “Mar de Morros” Environment Using Spectral Data

ABSTRACT Quantification of soil properties is essential for better understanding of the environment and better soil management. The conventional techniques of laboratory analysis are sometimes costly and detrimental to the environment. Thus, development of new techniques for soil analysis that do not generate residues, such as spectroscopy, is increasingly necessary as a viable way to estimate a wide range of soil properties. The objective of this study was to predict the levels of organic carbon (OC), clay, and extractable phosphorus [...]

Variability of soil physicochemical properties at different agroecological zones of Himalayan region: Sikkim, India

Quantification of soil properties and identification of soil water holding capacity (WHC) are essential for proper cropping and irrigation planning. The traditional approach of calculating WHC involves the application of pedo-transfer functions (PTFs), which are unavailable in many developing countries especially in the Himalayan region. Therefore, this study evaluates the effect of elevation on the physical and chemical properties of the surface soil in a Himalayan river basin in India. Furthermore, regression models were generated for calculating maximum WHC at different agroecological zones (AEZs). A total of 129 soil samples were randomly collected from three AEZs (subtropical, temperate and trans-himalayan). Laboratory analysis was done to identify the pH, organic carbon (OC), particle size distribution, particle density, bulk density (BD), soil water (air dry) (SW), total porosity (TP) and maximum WHC at the three AEZs. Pearson correlation coefficients were derived among the soil attributes. Also, stepwise (forward) multiple linear regression (MLR) models were generated for the predictability of WHC. The results illustrate a statistically significant variation in soil attributes for pH, OC, sand, silt, BD and WHC at all three AEZs. Furthermore, WHC exhibits significant correlation with BD, TP and SW through all three AEZs. Stepwise multiple linear regression analysis shows high predictability of maximum WHC while using appropriate soil attributes; with an adjusted coefficient of determination of 0.87, 0.82 and 0.78 for subtropical, temperate and trans-himalayan AEZs, respectively. Based on the results, it is concluded that soil properties vary significantly through AEZs in the Himalayan region. Also, several properties are inter-correlated at the AEZs and simple MLR models are recommended while estimating maximum WHC. The outputs of this study can be used as a guide to decide crop suitability and develop MLR models for the calculation of WHC in the areas where PTFs are not readily available.

Quantification of Soil Properties with Hyperspectral Data: Selecting Spectral Variables with Different Methods to Improve Accuracies and Analyze Prediction Mechanisms

We explored the potentials of both non-imaging laboratory and airborne imaging spectroscopy to assess arable soil quality indicators. We focused on microbial biomass-C (MBC) and hot water-extractable C (HWEC), complemented by organic carbon (OC) and nitrogen (N) as well-studied spectrally active parameters. The aggregation of different spectral variable selection strategies was used to analyze benefits for reachable estimation accuracies and to explore spectral predictive mechanisms for MBC and HWEC. With selected variables, quantification accuracies improved markedly for MBC (laboratory: RPD = 2.32 instead of 1.33 with full spectra; airborne: 2.35 instead of 1.80) and OC (laboratory: RPD = 3.08 instead of 2.36; airborne: 2.20 instead of 1.94). Patterns of selected variables indicated similarities between HWEC and OC, but significant differences between all other soil variables. This agreed to our results of indirect approaches in which both (i) wet-chemical data of OC and N and (ii) spectra fitted to measured OC and N values were used to estimate MBC and HWEC. Compared to these approaches, we found marked benefits of laboratory and airborne data for a direct spectral quantification of MBC (but not for HWEC). This suggests specificity of spectra for MBC, usable for the determination of this important soil parameter.

Microscope-based computer vision to characterize soil texture and soil organic matter

Characterization and quantification of soil properties are important for the optimum use and management our soil. While Soil texture is an important factor for decision making in agriculture, civil engineering and other industries, soil organic matter (SOM) is the backbone of soil health and quality and also affects a range of other soil properties and processes. Traditional methods for estimating these soil properties are time consuming and laborious. This paper discloses the design and development of a new cost effective in situ computer vision-based sensor system to estimate soil texture and SOM. A small and inexpensive hand-held microscope was used to develop an image acquisition system. Images acquired in laboratory with variable texture and SOM were processed using means of geospatial data analysis based computer vision algorithm. Simple linear regression predictive relationships were developed to estimate soil texture and SOM using various computed parameters of acquired imagery. Predicted sand (coefficients of determination, R2 = 0.63) and SOM (R2 = 0.83) were in good agreement with the laboratory measurement. These correspond to root mean square error of 84.7 g sand kg−1 of soil and 0.11 log (SOM %) for soils exhibiting the range of % sand from 3.4 to 59.3 and % SOM from 5.5 to 72.8, respectively. Low cost and the portability of the acquisition system and the computer vision algorithm developed in this study suggested suitability for its use in both laboratory and field conditions and shows promise as a proximal soil sensor.

Weed infestation in the stand of milk thistle and infestation in sustainable crop rotation

Weed infestation in the stand of milk thistle and infestation in sustainable crop rotation

Spectral properties and sources of variability of ecosystem components in a Mediterranean semiarid environment

Monitoring changes in soil and vegetation is crucial in fragile Mediterranean semiarid environments. To identify early perturbation warnings using radiometric techniques, the spectral variability of the ecosystem components must be understood and the ranges of their natural variability established. This paper aims to assess the potential of hyperspectral information for analysing spectral variability and to provide guidelines for using spectroscopy to monitor semiarid environments. Spectral field measurements of different soil (volcanic and calcareous), vegetation (esparto and dwarf palm) and natural mixtures (NPV and BSC) were acquired; spectral assessment focused on albedo, spectral shape and absorption bands. Results show high variability of soil background between and within soil typologies. The presence of biotic elements masks the distinctive characteristics between soil typologies, making combination of the VIS-NIR and SWIR regions necessary for correct discrimination and quantification of soil properties. Plant species are discriminated by taking advantage of their spectral properties based on the differences in absorption by chlorophyll, water and lignin. In summer, these differences are more significant, predictably due to their specific survival strategies. Radiometric techniques enable discrimination of ecosystem components, although the high variability of the landscape and subtle differences between components makes it necessary to use high-spectral resolution information.