Soil Aggregate Stability Mapping Using Remote Sensing and GIS-Based Machine Learning Technique

Abstract

Soil aggregate stability (SAS) is a critical parameter of soil quality and its mapping can help determine erosion hotspots. Despite this importance, SAS is less documented in available literature due to limited number of analyzes besides being a time consuming. For this reason, many researchers have turned to alternative methods that often use readily available variables such as soil parameters or remote sensing indices to estimate this variable. In that framework, the aim of the present study focused on the investigation of the feasibile use of adapted Leo Breiman’s random forest algorithm (RF) to mapping different mean weight diameter (MWD) tests as an index of SAS (mechanical breakdown (MWDmb), slow wetting (MWDsw), fast wetting (MWDfw) and the mean of the three tests (MWDmean)). The model was built with 77 samples distributed in the three watersheds of the study area located at Settat Ben-Ahmed, in Morocco and with the use of several environmental variables such as soil parameters (organic matter and clay), remote sensing indices (band 2, band 3, band 4, band 5, normalized difference vegetation index (NDVI) and transformed normalized difference vegetation index (TNDVI)), topography (elevation, slope, curvature plane and the topographic wetness index (TWI)) along with additional categorical variables as geological maps, land use and soil classes. The results showed a good level of accuracy for the training phase (75% of samples) for the different tests (R2 > 0.92, RMSE and MAE < 0.15) and were satisfactory for the testing phase (25% of samples, R2 > 0.65, RMSE and MAE < 0.31). Also, organic matter, topography and geology were the most important parameters in the spatial prediction of SAS. Finally, the maps build during this study could be of great use to identify areas of less stable soils in the perspective for taking the necessary measures to improve their quality.

Machine learning applications in vadose zone hydrology: A review

“…Specifically, ANN is widely applied in PTF and soil WRC (Bandai & Ghezzehei, 2021; More et al., 2022; Zhang & Schaap, 2017, 2019; Zhang et al., 2018) and is comparable to RF (Kim et al., 2021; Wang et al., 2022) (see Sections 2.2.1.1 and 2.2.1.2). Further, RF dominates quite a diverse set of applications over other ML options, including soil WRC‐based hydraulic parameters (Jena et al., 2021; Sedaghat et al., 2022) (Section 2.2.1.2), vadose zone flux (Crompton et al., 2019; Wells et al., 2021) (Section 2.2.1.3), vadose zone processes under extreme weather and climate conditions (Bouslihim et al., 2021; Zhong et al., 2022) (Section 2.3.1), and other anthropogenic events and natural processes (Cho et al., 2019; Kumar et al., 2021; Lei et al., 2019) (Section 2.3.2).…”

“…In particular, topography and geology were found to be two important features (Bouslihim et al, 2021). In flooded coastal regions, reconstructing ANN inputs based on embedding theory with mutual information improved its soil pore water salinity prediction performance (its R 2 score is more than 0.897) (Zheng et al, 2016).…”

“…In semiarid and arid regions, environmental variables, RS indices (NDVI), and topography features are generally used in ML (RF) applications to predict soil aggregate stability. In particular, topography and geology were found to be two important features (Bouslihim et al., 2021). In flooded coastal regions, reconstructing ANN inputs based on embedding theory with mutual information improved its soil pore water salinity prediction performance (its R 2 score is more than 0.897) (Zheng et al., 2016).…”

The Effect of Erosional Transformation of Soil Cover on the Stability of Soil Aggregates within Young Hummocky Moraine Landscapes in Northern Poland

“…Prior studies that have noted the importance of the examination of aggregate stability as a part of spatial characteristics of soil cover did not take into account such complicated areas as the young hummocky moraine landscapes in the European boreal zone. However, more and more research studies have made attempts to access aggregate stability in different landscapes and carry out the differences in aggregate stability on a regional scale [61][62][63][64]. In contemporary research, the prediction of soil aggregate distribution in a landscape is based on available soil data, which is why the expansion of additional data can be an important task in such areas as a young hummocky moraine landscape.…”

Research on quantitative inversion of ion adsorption type rare earth ore based on convolutional neural network

“…Commonly used statistical models (Ma et al, 2018) are: univariate regression, multiple linear regression (Cheng, 2007;Guo et al, 2018;Cheng et al, 2019a), partial least squares (Li et al, 2005;Liu and Zhang, 2007;Xu et al, 2018), neural networks (Schiller and Doerffer, 1999;Yu et al, 2012;Cao et al, 2017;Lin et al, 2018) support vector machines, random forests and other methods (Durbha et al, 2007;Abdel-Rahman et al, 2013;Xu et al, 2014;Vincenzi et al, 2015;Jiang, 2017;LI et al, 2017;Wang et al, 2018). With the development of computer technology, artificial intelligence science and deep learning methods have been developed, and some scholars began to apply deep learning methods to quantitative inversion, and have achieved good inversion results (Wang et al, 2017;Tan et al, 2018;Liu et al, 2020;Bouslihim et al, 2021). At present, most of the remote sensing quantitative inversion research is based on the measured spectral data.…”