Determination the Factors Explaining Variability of Physical Soil Organic Carbon Fractions using Artificial Neural Network

Ayoubi, Shamsollah; Karchegani, Parisa Mokhtari

doi:10.3923/ijss.2012.1.14

Cited by 3 publications

(2 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ANN, analog to the biological neural network, is more complex and uses synaptic weights to establish a connection among predictor variables through multiple layers of networks and predicts the classification (Grunwald, 2022). Some of the features of the neural network such as no assumption of a prior relationship, availability of multiple training algorithm detections of complex nonlinear relationships, and detection of all possible interactions between predictor variables (Tu, 1996), made it another popular soil classification model (Alvarez et al, 2011;Ayoubi & Karchegani, 2012;Baligh et al, 2020;Hossein Alavi et al, 2010;Koekkoek & Booltink, 1999;Minasny et al, 2004Minasny et al, , 2016Ozturk et al, 2011). Application of multiple models for the classification of soil can be useful in arriving at more robust prediction as well providing a comparison among different models.…”

Section: Introductionmentioning

confidence: 99%

Characterization and classification of fine-resolution soil profile for precision agriculture using random forest and self-organizing map

Elias,

Sharma,

Goel

2024

Preprint

View full text Add to dashboard Cite

The availability of high throughput soil profile information is an important component in precision agriculture to perform efficient soil management for sustainable production. We collected 14 soil physiochemical features from Nagpur, Pune, and Haveri, representing target environments of safflower cultivation, and also from our experiment station at Delhi, at fine resolution and created graphical maps to depict the variability. Additionally, we evaluated the predictive ability of two statistical learning models, random forest (RF) and self-organizing maps (SOM) against multinomial regression models for correctly classifying the soil profile. Clustering was performed around the medoids produced from the dissimilarity matrices of these models using partitioning around medoids (PAM) model. The robustness, versatility, and predictive ability of models in correctly classifying the soil profile to clusters were then tested using cross-validation which was repeated 100 times. This study was performed using training data with proportionate size varying from 60 to 95%, and increasing the unit area of observation up to nine times (or decreasing the total number of observations up to a ninth). RF model was found to be the best performing with average prediction accuracy above 85% in all settings which reached close to 100% in some settings. The predictive ability of all the models was maintained even when only the most influencing six variables were used for classification. The optimal training population size for prediction was found to be 70 to 80%. Based on our study, it is recommended to i) collect fine resolution edaphic features from a marginal farm before crop season, ii) use RF or SOM model to identify the most influencing features distinguishing the soil samples iii) expand the area of sample collection, find values for the most influencing features, and use RF model to correctly predict the class to which the new set of the soil belongs to.

show abstract

Section: Introductionmentioning

confidence: 99%

Characterization and classification of fine-resolution soil profile for precision agriculture using random forest and self-organizing map

Elias,

Sharma,

Goel

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…ANN, RF, and Multinomial Logistic Regression (MnLR) have gradually become the most commonly used models for soil classification and soil prediction (Zeraatpisheh et al 2020). ANN has good prediction accuracy for the soil enzyme activity (Tajik et al 2012), SOC (Ayoubi & Karchegani 2012), soil aggregate stability (Besalatpour et al 2013), soil hydraulic properties (Azadmard et al 2020), and Atterberg consistency (Zolfaghari et al 2015). Although some have questioned the credibility of ML (Rossiter 2018), it is widely acknowledged that modelling soil processes through ML improves our understanding of soil properties and processes (Rudin & Wagstaff https://doi.org/10.17221/94/2022-SWR 2014Brungard et al 2015;Fajardo et al 2016;Rossiter 2018;Ma et al 2019;Mjolsness & Decoste 2001) when parallel genetic algorithms are applied.…”

mentioning

confidence: 99%

Past, present and future of the applications of machine learning in soil science and hydrology

Wang¹,

Yang²,

Lv³

et al. 2023

Soil Water Res.

View full text Add to dashboard Cite

Machine learning can handle an ever-increasing amount of data with the ability to learn models from the data. It has been widely used in a variety of disciplines and is gaining increasingly more attention nowadays. As it is challenging to map soil and hydrological information that are characterised with high spatial and temporal variability, applications of machine learning in soil science and hydrology (AMLSH) have become popularised. To better understand the current state of AMLSH research, a scientific and quantitative approach was performed to statistically analyse publication information from 1973 to 2021 archived in the Scopus database using scientometric analysis tools, including VOSviewer, CiteSpace, and the open--source R package "bibliometrix". The results show a significant increase in the number of publications on AMLSH since 2006. The major contributions were identified based on country origins (China, the USA, and India), institutions (Hohai University, Islamic Azad University, and Wuhan University), and journals (Journal of Hydrology, Remote Sensing, and Geoderma). The keywords analysis of the AMLSH research demonstrates four research hotspots: neural network, artificial intelligence, machine learning, and soil. The most frequently utilised machine learning (ML) methods are neural networks, decision trees, random forests and other methods for image processing and predictive analysis. McBratney et al. 2003 is the most highly cited article. Our research sheds light on the research process on AMLSH and concludes with future research perspectives.

show abstract

Long-term cultivation and landscape position effects on aggregate size and organic carbon fractionation on surface soil properties in semi-arid region of Iran

Safadoust

Doaei

Mahboubi

et al. 2015

Arid Land Research and Management

View full text Add to dashboard Cite

This study was carried out to investigate the effects of long-term cultivation and landscape position on organic carbon content and soil aggregation. Sampling sites were determined based upon land use at the end of 50 years soil use and management, cultivated=annual wheat cropping and grazed pasture, and landscape position in Chaharmahal-va-Bakhtiary province, southwest Iran. Soil samples were collected from the 0-5 cm and 5-15 cm depths in two adjacent fields that have the same slope and aspect. The soil was silty clay at the summit and footslope positions, and was a silty clay loam at the backslope. Wet-sieving analysis and aggregate-size fractionation methods were used to separate the samples into three aggregate fractions (i.e., 2-4.75, 0.25-2, and 0.053-0.25 mm). The treatments were arranged in a factorial design. Land use significantly affected the water-stable aggregate fractions, so that the wet soil stability of the macroaggregates (i.e., 2-4.75 mm) was higher in the pasture, whereas it was greater for the meso-aggregates (i.e., 0.25-2 mm) in the cultivated soils. Cultivation decreased both the wet-aggregate stability and percent of macroaggregates whereas long-term pasture enhanced aggregation. Soil organic carbon (SOC) content within aggregates and primary particles was also significantly influenced by landscape position, land use, and the depth of sampling. The SOC content was higher in clay than those in silt and sand contents. The SOC content decreased as depth increased in all fractions. In general, the highest and lowest wet-stable aggregates were observed on the footslope and backslope positions, respectively.

show abstract

Determination the Factors Explaining Variability of Physical Soil Organic Carbon Fractions using Artificial Neural Network

Cited by 3 publications

References 24 publications

Characterization and classification of fine-resolution soil profile for precision agriculture using random forest and self-organizing map

Characterization and classification of fine-resolution soil profile for precision agriculture using random forest and self-organizing map

Past, present and future of the applications of machine learning in soil science and hydrology

Long-term cultivation and landscape position effects on aggregate size and organic carbon fractionation on surface soil properties in semi-arid region of Iran

Contact Info

Product

Resources

About