Prediction of Topsoil Texture Through Regression Trees and Multiple Linear Regressions

Abstract

Users of soil survey products are mostly interested in understanding how soil properties vary in space and time. The aim of digital soil mapping (DSM) is to represent the spatial variability of soil properties quantitatively to support decision-making. The goal of this study is to evaluate DSM techniques (Regression Trees-RT and Multiple Linear Regressions-MLR) and the ability of these tools to predict mineral fraction content under a wide variability of landscapes. The study site was the entire Guapi-Macacu watershed (1,250.78 km²) in the state of Rio de Janeiro in the Southeast region of Brazil. Terrain attributes and remote sensing data (with 30 m of spatial resolution) were used to represent landscape co-variables selected as an input in predictive models in order to develop the explanatory variables. The selection of sampling sites was based on the Latin Hypercube algorithm. A representative set of one hundred points with feasible field access was chosen. Different input databases were tested for prediction of mineral fraction content (harmonized and original data). The Spline algorithm was used to harmonize data according to the GlobalSoil.Net consortium standards. The results showed better performance from the RT models, using input from an average of six covariates; the simplest MLR model used twice as many input variables, creating more complex models without gaining precision. Furthermore, better R² values were obtained using RT models, irrespective of harmonization of soil data. The harmonized dataset from the 0.00-0.05 and 0.05-0.15 m layers, in general, presented better results for the clay and silt, with R 2 values of 0.52 (0.00-0.05 m) and 0.69 (0.05-0.15 m), respectively. Prediction of sand content showed better results when the original depth data was used as an input, although all regression tree models had R 2 values greater than 0.52. The RT models provided a better statistical index than MLR for all predicted properties; however, the variance between models suggests similarity of performance. Regarding harmonization of soil data, both input databases (harmonized or not) can be used to predict soil properties, since the variance of model performance was low and generalization of the soil maps showed similar trends. The products obtained from the digital soil mapping approach make it possible to integrate the factor of uncertainties, providing easier interpretation for soil management and land use decisions.

Digital mapping of soil texture classes for efficient land management in the Piedmont plain of Iran

“…The topographic wetness index, which is an important topography variable that affects the amount of moisture accumulation in the soil (Mehrabi-Gohari et al, 2019), can be considered as an important variable in modelling the silt fraction. In terms of showing areas with high moisture accumulation potential, its effect on the estimation of silt content has been reported in other studies (Moore et al, 1993;Nabiollahi et al, 2014;Pinheiro et al, 2018). Hateffard et al (2019) reported that the most important environmental variable in the spatial estimation of the topsoil fractions is the information obtained from the 5th Band of the Landsat 8 OLI satellite under similar geographical conditions.…”

Identifying the Underlying Factors and Variables Governing Macronutrients in Cultivated Tropical Peatland Using Regression Tree Approach

“…In both datasets, RT-based total N and P models gained slight performance improvements compared to MLR (Table 2). The RT models reportedly outperformed MLR and its derivatives (e.g., stepwise backward or forward eliminations, general linear model/GLM) in predicting sorption and retention of heavy metals (Vega et al, 2009), topsoil texture (Pinheiro et al, 2018), soil heavy metal content (Qiu et al, 2016), and base saturation percentages (Rawal et al, 2019). The higher capability of RT compared to MLR was also reported in other fields, such as airborne bacterial hazard assessment (Yoo et al, 2018) and landslide susceptibility (Pourgashemi & Rahmati, 2018).…”

“…The regression tree (RT) algorithm is one of the firstly developed ML (Breiman et al, 1984; see review by Loh, 2014). RT was considered successful in modeling several chemical and physical characteristics of mineral soils compared to linear-based models and their derivatives (Vega et al, 2009;Pinheiro et al, 2018;Qiu et al, 2016;Rawal et al, 2019). Despite their widespread usage in other fields of study, RT-based pedotransfer models' capability to estimate soil nutrients using tropical peat data was understudied.…”

Spatial prediction of soil organic carbon stocks in an arid rangeland using machine learning algorithms

“…These model's inputs were selected due to RFE (Table 2). After implementing the model and determining sampling points by the model, the objective function was optimized with 18,000 repeats (Pinheiro et al, 2018) for digital mapping of soil texture components, used cLHS method with 20,000 repeats and five covariates include spatial position, elevation, slope, curvature, and land use. Finally, the amount of soil organic carbon was measured by Walkley and Black (1934) method, and the gravel was calculated by volume method, bulk density was measured by cylinder method, and then the SOCS was calculated considering the surface horizon thickness using Eq.…”