Tree-based algorithms for spatial modeling of soil particle distribution in arid and semi-arid region.

Accurate estimation of particle size distribution across a large area is crucial for proper soil management and conservation, ensuring compatibility with capabilities and enabling better selection and adaptation of precision agricultural techniques. The study investigated the performance of tree-based models, ranging from simpler options like CART to sophisticated ones like XGBoost, in predicting soil texture over a wide geographic region. Models were constructed using remotely sensed plant and soil indexes as covariates. Variable selection employed the Boruta approach. Training and testing data for machine learning models consisted of particle size distribution results from 622 surface soil samples collected in southeastern Turkey. The XGBoostClay model emerged as the most accurate predictor, with an R2 value of 0.74. Its superiority was further underlined by a 21.36% relative improvement in XGBoostClay RMSE compared to RFClay and 44.5% compared to CARTClay. Similarly, the R2 values for XGBoostSilt and XGBoostSand models reached 0.71 and 0.75 in predicting sand and silt content, respectively. Among the considered covariates, the normalized ratio vegetation index and slope angle had the highest impact on clay content (21%), followed by topographic position index and simple ratio clay index (20%), while terrain ruggedness index had the least impact (18%). These results highlight the effectiveness of Boruta approach in selecting an adequate number of variables for digital mapping, suggesting its potential as a viable option in this field. Furthermore, the findings of this study suggest that remote sensing data can effectively contribute to digital soil mapping, with tree-based model development leading to improved prediction performance.

Improvement of spatial estimation for soil organic carbon stocks in Yuksekova plain using Sentinel 2 imagery and gradient descent-boosted regression tree.

Carbon sequestration in earth surface is higher than the atmosphere, and the amount of carbon stored in wetlands is much greater than all other land surfaces. The purpose of this study was to estimate soil organic carbon stocks (SOCS) and investigate spatial distribution pattern of Yuksekova wetlands and surrounding lands in Hakkari province of Turkey using machine learning and remote sensing data. Disturbed and undisturbed soil samples were collected from 10-cm depth in 50 locations differed with land use and land cover. Vegetation, soil, and moisture indices were calculated using Sentinel 2 Multispectral Sensor Instrument (MSI) data. Significant correlations (p≤0.01) were obtained between the indices and SOCS; thus, the remote sensing indices (ARVI 0.43, BI -0.43, GSI -0.39, GNDI 0.44, NDVI 0.44, NDWI 0.38, and SRCI 0.51) were used as covariates in multi-layer perceptron neural network (MLP) and gradient descent-boosted regression tree (GBDT) machine learning models. Mean absolute error, root mean square error, and mean absolute percentage error were 3.94 (Mg C ha -1), 6.64 (Mg C ha-1), and 9.97%, respectively. The simple ratio clay index (SRCI), which represents the soil texture, was the most important factor in the SOCS estimation variance. In addition, the relationship between SRCI and Topsoil Grain Size Index revealed that topsoil clay content is a highly important parameter in spatial variation of SOCS. The spatial SOCS values obtained using the GBDT model and the mean SOCS values of the CORINE land cover classes were significantly different. The land cover has a significant effect on SOC in Yuksekova plain. The mean SOCS for continuously ponded fields was 45.58 Mg C ha-1, which was significantly different from the mean SOCS of arable lands. The mean SOCS in arable lands, with significant areas of natural vegetation, was 50.22 Mg C ha-1 and this amount was significantly higher from the SOCS of other land covers (p<0.01). The wetlands had the highest SOCS (61.46 Mg C ha-1), followed by the lands principally occupied by natural vegetation and used as rangelands around the wetland (50.22 Mg C ha-1). Environmental conditions had significant effect on SOCS in the study area. The use of remote sensing indices instead of using single bands as estimators in the GBDT algorithm minimized radiometric errors, and reliable spatial SOCS information was obtained by using the estimators. Therefore, the spatial estimation of SOCS can be successfully determined with up-to-date machine learning algorithms only using remote sensing predictor variables. Reliable estimation of SOCS in wetlands and surrounding lands can help understand policy and decision makers the importance of wetlands in mitigating the negative impacts of global warming.

Combining spatial autocorrelation with artificial intelligence models to estimate spatial distribution and risks of heavy metal pollution in agricultural soils.

Information on spatial distribution and potential sources of heavy metals in agricultural lands is very important for human health and food safety. In this study, pollution degree of lead (Pb), cadmium (Cd), and nickel (Ni) in Yüksekova Plain, located on the border in the southeastern part of Turkey, was evaluated by geoaccumulation index (Igeo), modified contamination factor (mCdeg), and Nemerow pollution index (PINemerow) combined with spatial autocorrelation using deep learning algorithms. A total of 304 soil samples were collected from two different depths (0-20 and 20-40 cm) in the study area, which covered 17.5 thousand ha land. Covariates were determined for spatial distribution models of Pb, Cd, and Ni by factor analysis (FA). Spatial distribution models for surface soils were developed using pedovariables (silt, sand, clay lime, organic matter, electrical conductivity, pH, Ca, and Na) determined by the FA and Igeo and mCdeg values by artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) models. The estimation success of models for different depths was assessed by root mean square error (RMSE), mean absolute percent error (MAPE), and Taylor diagrams. The RMSE and MAPE values showed a strong correlation between heavy metal contents and the covariates. The RMSE values of ANN-Ni0-20, ANN-Ni20-40, ANN-Pb0-20, ANN-Cd0-20, and ANN-Cd20-40 models (0.01240, 0.07257, 0.0039, 0.00045, 0.00044, and 0.04607, respectively) confirmed the success of the models. Likewise, the MAPE values between 0.2 and 8.5% indicated that all models were very good predictors. In addition, the Taylor diagrams showed that the estimation performance of ANFIS and ANN models are compatible. The IgeoNi and IgeoPb values in both models at both depths indicated that strongly to extremely polluted (4-5) areas are quite high in the study area, while the IgeoCd values revealed that unpolluted areas are widespread. The mCdeg index value showed a moderate to high contamination at the first depth, while very high contamination at the second depth in most of the study area. Spatial distribution of PINemerow revealed that moderate pollution (2-3) is common in both soil depths of the study area. The PINemerow of subsurface layer was between 0.91 and 1 (warning limit class) in a small part of the study area. The results showed that vertical mobility of heavy metals is closely related to pedovariables. In addition, the ANN and ANFIS models are capable of exhibiting the heterogeneity in the spatial distribution pattern of high variation in the data. Thus, the locations with extreme contamination have been accurately determined. The pollution indices calculated considering the commonly used international reference values revealed that heavy metal pollution in some part of the study area reached the detrimental levels for human health and food safety. The results suggested that the pollution indices were more successful than simple heavy metal concentrations in interpreting the pollution risk levels. High-resolution spatial information reported in this study can help policy makers and authorities to reduce heavy metal emissions of pollutants or, if possible, to eliminate the pollution.

Accuracy Assessment of Kriging, artificial neural network, and a hybrid approach integrating spatial and terrain data in estimating and mapping of soil organic carbon.

This study aimed to produce a soil organic carbon (SOC) content map with high accuracy and spatial resolution using the most effective factors in the model. The spatial SOC estimation success of Inverse Distance Weighting (IDW), Ordinary Kriging (OK), Empirical Bayesian Kriging (EBK), Multi-Layered Perception Network (MLP) and MLP-OK Hybrid models were compared to obtain the most reliable model in estimating the SOC content. The study area was located in Besni district in the Southeastern Anatolia Region of Turkey. Total of 132 surface (0-30 cm) soil samples were collected from the covers 1330 km2 land and analyzed for SOC, lime, clay and sand content and soil reaction included in the estimation models. Mean annual precipitation and temperature, elevation, compound topographic index, enhanced vegetation and normalized difference vegetation index, were also used as the inputs in the modelling. The spatial distribution of SOC was determined using a MLP and a two-stage ensemble model (MLP-OK) combining the estimation of OK residuals. Soil surveys and covariates were used to train and validate the MLP-OK hybrid model. The MLP-OK model provided a more accurate estimation of SOC content with minimal estimation errors (ME: -0.028, 45 MAE: 0.042, RMSE: 0.066) for validation points compared to the other models. The MLP-OK model outperformed other models by 75.09 to 77.92%. The MLP-OK model estimated the lower and upper limits of the estimated and the measured values in a consistent manner compared to the other models. The spatial distribution map of SOC content obtained by ANN-kriging approach was significantly affected by ancillary variables, and revealed more detail than other interpolation methods in the northern, central, southwestern and southeastern parts of the study area. The results revealed that the assembling of MLP with OK model can contribute to obtain more reliable regional, national and global spatial soil information.