Pervasive soil phosphorus losses in terrestrial ecosystems in China.

Future phosphorus (P) shortages could seriously affect terrestrial productivity and food security. We investigated the changes in topsoil available P (AP) and total P (TP) in China's forests, grasslands, paddy fields, and upland croplands during the 1980s-2010s based on substantial repeated soil P measurements (63,220 samples in the 1980s, 2000s, and 2010s) and machine learning techniques. Between the 1980s and 2010s, total soil AP stock increased with a small but significant rate of 0.13 kg P ha-1 year-1 , but total soil TP stock declined substantially (4.5 kg P ha-1 year-1 ) in the four ecosystems. We quantified the P budgets of soil-plant systems by harmonizing P fluxes from various sources for this period. Matching trends of soil contents over the decades with P budgets and fluxes, we found that the P-surplus in cultivated soils (especially in upland croplands) might be overestimated due to the great soil TP pool compared to fertilization and the substantial soil P losses through plant uptake and water erosion that offset the P additions. Our findings of P-deficit in China raise the alarm on the sustainability of future biomass production (especially in forests), highlight the urgency of P recycling in croplands, and emphasize the critical role of country-level basic data in guiding sound policies to tackle the global P crises.

Importance of Terrain and Climate for Predicting Soil Organic Carbon Is Highly Variable across Local to Continental Scales.

Soil organic carbon (SOC) plays a vital role in global carbon cycling and sequestration, underpinning the need for a comprehensive understanding of its distribution and controls. This study explores the importance of various covariates on SOC spatial distribution at both local (up to 1.25 km) and continental (USA) scales using a deep learning approach. Our findings highlight the significant role of terrain attributes in predicting SOC concentration distribution with terrain, contributing approximately one-third of the overall prediction at the local scale. At the continental scale, climate is only 1.2 times more important than terrain in predicting SOC distribution, whereas at the local scale, the structural pattern of terrain is 14 and 2 times more important than climate and vegetation, respectively. We underscore that terrain attributes, while being integral to the SOC distribution at all scales, are stronger predictors at the local scale with explicit spatial arrangement information. While this observational study does not assess causal mechanisms, our analysis nonetheless presents a nuanced perspective about SOC spatial distribution, which suggests disparate predictors of SOC at local and continental scales. The insights gained from this study have implications for improved SOC mapping, decision support tools, and land management strategies, aiding in the development of effective carbon sequestration initiatives and enhancing climate mitigation efforts.

Integrating proximal soil sensing data and environmental variables to enhance the prediction accuracy for soil salinity and sodicity in a region of Xinjiang Province, China.

Soil salinization and sodification, the primary causes of land degradation and desertification in arid and semi-arid regions, demand effective monitoring for sustainable land management. This study explores the utility of partial least square (PLS) latent variables (LVs) derived from visible and near-infrared (Vis-NIR) spectroscopy, combined with remote sensing (RS) and auxiliary variables, to predict electrical conductivity (EC) and sodium absorption ratio (SAR) in northern Xinjiang, China. Using 90 soil samples from the Karamay district, machine learning models (Random Forest, Support Vector Regression, Cubist) were tested in four scenarios. Modeling results showed that RS and Land use alone were unreliable predictors, but the addition of topographic attributes significantly improved the prediction accuracy for both EC and SAR. The incorporation of PLS LVs derived from Vis-NIR spectroscopy led to the highest performance by the Random Forest model for EC (CCC = 0.83, R2 = 0.80, nRMSE = 0.48, RPD = 2.12) and SAR (CCC = 0.78, R2 = 0.74, nRMSE = 0.58, RPD = 2.25). The variable importance analysis identified PLS LVs, certain topographic attributes (e.g., valley depth, elevation, channel network base level, diffuse insolation), and specific RS data (i.e., polarization index of VV + VH) as the most influential predictors in the study area. This study affirms the efficiency of Vis-NIR data for digital soil mapping, offering a cost-effective solution. In conclusion, the integration of proximal soil sensing techniques and highly relevant topographic attributes with the RF model has the potential to yield a reliable spatial model for mapping soil EC and SAR. This integrated approach allows for the delineation of hazardous zones, which in turn enables the consideration of best management practices and contributes to the reduction of the risk of degradation in salt-affected and sodicity-affected soils.

Soil data recency: The foundation for harmonizing soil data across time.

Sustainable soil resource management depends on reliable soil information, often derived from 'legacy soil data' or a combination of old and new soil data. However, the task of harmonizing soil data collected at different times remains a largely unexplored in the literature. Addressing this challenge requires incorporating the temporal dimension into mathematical and statistical models for spatio-temporal soil studies. This study aimed to create a comprehensive framework for harmonizing soil data across various time. We assessed the integration of historical and recent soil data, ranging from 4 to 48 years old data, using soil data recency analysis. To achieve this, we introduced an 'age of data' attribute, calculating the time difference between soil survey years and the present (e.g., 2022). We applied three machine learning models - Decision Trees (DT), Random Forest (RF), Gradient Boosting (GBM) - to a dataset containing 6339 sites and 28,149 depth-harmonized layers. The results consistently demonstrated robust performance across models, RF outperforming with an R-squared value of 0.99, RMSE of 1.41, and a concordance of 0.97. Similarly, DT and GBM also showed strong predictive power. Terrain-derived environmental covariates played a more important role than land use and land cover (LULC) change in predicting soil data recency. While LULC change showed soil organic carbon concentration variability across the different depths, it was a less important factor. Anthropogenic factors, such as LULC change and normalized difference vegetation index (NDVI), were not primary determinants of soil data recency. Variations in soil depth had no impact on predicting soil data recency. This study validated that terrain-derived covariates, especially elevation factors, effectively explain the quality of older soil data when predicting current soil attributes using the soil data recency concept. This approach has the potential to enhance real-time estimates, such as carbon budgets, and we emphasize its importance in global earth system models.

Improved soil carbon stock spatial prediction in a Mediterranean soil erosion site through robust machine learning techniques.

Soil serves as a reservoir for organic carbon stock, which indicates soil quality and fertility within the terrestrial ecosystem. Therefore, it is crucial to comprehend the spatial distribution of soil organic carbon stock (SOCS) and the factors influencing it to achieve sustainable practices and ensure soil health. Thus, the present study aimed to apply four machine learning (ML) models, namely, random forest (RF), k-nearest neighbors (kNN), support vector machine (SVM), and Cubist model tree (Cubist), to improve the prediction of SOCS in the Srou catchment located in the Upper Oum Er-Rbia watershed, Morocco. From an inventory of 120 sample points, 80% were used for training the model, with the remaining 20% set aside for model testing. Boruta's algorithm and the multicollinearity test identified only nine (9) factors as the controlling factors selected as input data for predicting SOCS. As a result, spatial distribution maps for SOCS were generated for all models, then compared, and further validated using statistical metrics. Among the models tested, the RF model exhibited the best performance (R2 = 0.76, RMSE = 0.52 Mg C/ha, NRMSE = 0.13, and MAE = 0.34 Mg C/ha), followed closely by the SVM model (R2 = 0.68, RMSE = 0.59 Mg C/ha, NRMSE = 0.15, and MAE = 0.34 Mg C/ha) and Cubist model (R2 = 0.64, RMSE = 0.63 Mg C/ha, NRMSE = 0.16, and MAE = 0.43 Mg C/ha), while the kNN model had the lowest performance (R2 = 0.31, RMSE = 0.94 Mg C/ha, NRMSE = 0.24, and MAE = 0.63 Mg C/ha). However, bulk density, pH, electrical conductivity, and calcium carbonate were the most important factors for spatially predicting SOCS in this semi-arid region. Hence, the methodology used in this study, which relies on ML algorithms, holds the potential for modeling and mapping SOCS and soil properties in comparable contexts elsewhere.

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.

Comparison and application of SOFM, fuzzy c-means and k-means clustering algorithms for natural soil environment regionalization in China.

Soil attributes and their environmental drivers exhibit different patterns in different geographical directions, along with distinct regional characteristics, which may have important effects on substance migration and transformation such as organic matter and soil elements or the environmental impacts of pollutants. Therefore, regional soil characteristics should be considered in the process of regionalization for environmental management. However, no comprehensive evaluation or systematic classification of the natural soil environment has been established for China. Here, we established an index system for natural soil environmental regionalization (NSER) by combining literature data obtained based on bibliometrics with the analytic hierarchy process (AHP). Based on the index system, we collected spatial distribution data for 14 indexes at the national scale. In addition, three clustering algorithms-self-organizing feature mapping (SOFM), fuzzy c-means (FCM) and k-means (KM)-were used to classify and define the natural soil environment. We imported four cluster validity indexes (CVI) to evaluate different models: Davies-Bouldin index (DB), Silhouette index (Sil) and Calinski-Harabasz index (CH) for FCM and KM, clustering quality index (CQI) for SOFM. Analysis and comparison of the results showed that when the number of clusters was 13, the FCM clustering algorithm achieved the optimal clustering results (DB = 1.16, Sil = 0.78, CH = 6.77 × 106), allowing the natural soil environment of China to be divided into 12 regions with distinct characteristics. Our study provides a set of comprehensive scientific research methods for regionalization research based on spatial data, it has important reference value for improving soil environmental management based on local conditions in China.

Digital mapping of soil organic carbon density in China using an ensemble model.

The soil organic carbon stock (SOCS) is considered as one of the largest carbon reservoirs in terrestrial ecosystems, and small changes in soil can cause significant changes in atmospheric CO2 concentration. Understanding organic carbon accumulation in soils is crucial if China is to meet its dual carbon target. In this study, the soil organic carbon density (SOCD) in China was digitally mapped using an ensemble machine learning (ML) model. First, based on SOCD data obtained at depths of 0-20 cm from 4356 sampling points (15 environmental covariates), we compared the performance of four ML models, namely random forest (RF), extreme gradient boosting (XGBoost), support vector machine (SVM), and artificial neural network (ANN) models, in terms of coefficient of determination (R2), mean absolute error (MAE), and root mean square error (RMSE) values. Then, we ensembled four models using Voting Regressor and the principle of stacking. The results showed that ensemble model (EM) accuracy was high (RMSE = 1.29, R2 = 0.85, MAE = 0.81), so that it could be a good choice for future research. Finally, the EM was used to predict the spatial distribution of SOCD in China, which ranged from 0.63 to 13.79 kg C/m2 (average = 4.09 (±1.90) kg C/m2). The SOC storage amount in surface soil (0-20 cm) was 39.40 Pg C. This study developed a novel, ensemble ML model for SOC prediction, and improved our understanding of the spatial distribution of SOC in China.

Estimation of Soil Characteristics from Multispectral Sentinel-3 Imagery and DEM Derivatives Using Machine Learning.

In this paper, different machine learning methodologies have been evaluated for the estimation of the multiple soil characteristics of a continental-wide area corresponding to the European region, using multispectral Sentinel-3 satellite imagery and digital elevation model (DEM) derivatives. The results confirm the importance of multispectral imagery in the estimation of soil properties and specifically show that the use of DEM derivatives improves the quality of the estimates, in terms of R2, by about 19% on average. In particular, the estimation of soil texture increases by about 43%, and that of cation exchange capacity (CEC) by about 65%. The importance of each input source (multispectral and DEM) in predicting the soil properties using machine learning has been traced back. It has been found that, overall, the use of multispectral features is more important than the use of DEM derivatives with a ration, on average, of 60% versus 40%.

Incorporating agricultural practices in digital mapping improves prediction of cropland soil organic carbon content: The case of the Tuojiang River Basin.

Accurate mapping of soil organic carbon (SOC) in cropland is essential for improving soil management in agriculture and assessing the potential of different strategies aiming at climate change mitigation. Cropland management practices have large impacts on agricultural soils, but have rarely been considered in previous SOC mapping work. In this study, cropland management practices including carbon input (CI), length of cultivation (LC), and irrigation (Irri) were incorporated as agricultural management covariates and integrated with natural variables to predict the spatial distribution of SOC using the Extreme Gradient Boosting (XGBoost) model. Additionally, we evaluated the performance of incorporating agricultural management practice variables in the prediction of cropland topsoil SOC. A case study was carried out in a traditional agricultural area in the Tuojiang River Basin, China. We found that CI was the most important environmental covariate for predicting cropland SOC. Adding cropland management practices to natural variables improved prediction accuracy, with the coefficient of determination (R2), the root mean squared error (RMSE) and Lin's Concordance Correlation Coefficient (LCCC) improving by 16.67%, 17.75% and 5.62%, respectively. Our results highlight the effectiveness of incorporating agricultural management practice information into SOC prediction models. We conclude that the construction of spatio-temporal database of agricultural management practices derived from inventories is a research priority to improve the reliability of SOC model prediction.

Effects of optical and radar satellite observations within Google Earth Engine on soil organic carbon prediction models in Spain.

The modeling and mapping of soil organic carbon (SOC) has advanced through the rapid growth of Earth observation data (e.g., Sentinel) collection and the advent of appropriate tools such as the Google Earth Engine (GEE). However, the effects of differing optical and radar sensors on SOC prediction models remain uncertain. This research aims to investigate the effects of different optical and radar sensors (Sentinel-1/2/3 and ALOS-2) on SOC prediction models based on long-term satellite observations on the GEE platform. We also evaluate the relative impact of four synthetic aperture radar (SAR) acquisition configurations (polarization mode, band frequency, orbital direction and time window) on SOC mapping with multiband SAR data from Spain. Twelve experiments involving different satellite data configurations, combined with 4027 soil samples, were used for building SOC random forest regression models. The results show that the synthesis mode and choice of satellite images, as well as the SAR acquisition configurations, influenced the model accuracy to varying degrees. Models based on SAR data involving cross-polarization, multiple time periods and "ASCENDING" orbits outperformed those involving copolarization, a single time period and "DESCENDING" orbits. Moreover, combining information from different orbital directions and polarization modes improved the soil prediction models. Among the SOC models based on long-term satellite observations, the Sentinel-3-based models (R2 = 0.40) performed the best, while the ALOS-2-based model performed the worst. In addition, the predictive performance of MSI/Sentinel-2 (R2 = 0.35) was comparable with that of SAR/Sentinel-1 (R2 = 0.35); however, the combination (R2 = 0.39) of the two improved the model performance. All the predicted maps involving Sentinel satellites had similar spatial patterns that were higher in northwest Spain and lower in the south. Overall, this study provides insights into the effects of different optical and radar sensors and radar system parameters on soil prediction models and improves our understanding of the potential of Sentinels in developing soil carbon mapping.

Machine learning-based digital mapping of soil organic carbon and texture in the mid-Himalayan terrain.

Mountain soils have received significant attention due to their profound influence on ecological processes and environmental factors. However, mapping these soils in digital soil mapping technique encounters several challenges, including high local variability, non-linear relationships between environmental covariates and soil properties, limited accessibility in complex topographical settings, and the absence of universally applicable covariates for soil formation. To address these issues, this study integrates soil-forming factors of the scorpan model to map soil organic carbon (SOC) and soil texture in the mid-Himalayas. By considering over 100 environmental covariates, with a focus on terrain parameters relevant to mountainous environments, the study aims to enhance the accuracy of ML regression models through augmentation techniques that overcome data insufficiency. Using augmented soil observations and covariates, a non-parametric random forest regression model is trained and applied to predict soil variables across the study area, generating a continuous fine-resolution map. The model's performance, evaluated against an unknown dataset, was significant with an R-square of 0.80, 0.79, 0.72, and 0.84 for clay, sand, silt, and SOC, respectively. Furthermore, a sensitivity analysis of the environmental covariates and their impact on the model revealed that all the soil-forming factors make a significant contribution to the model's effectiveness. The insights gained from this research contribute to a better understanding of mountain soils and facilitate the development of effective conservation and sustainable management strategies for mountainous regions.

Modeling the spatial variation of calcium carbonate equivalent to depth using machine learning techniques.

Inorganic carbon is the largest source of carbon in terrestrial surface, particularly in arid and semiarid regions, including the Chahardowli Plain in western Iran. Inorganic carbon plays an equal or greater role than organic soil carbon in these areas, although less attention has been paid in quantifying their variability. The objective of this study was to model and map calcium carbonate equivalent (CCE) presenting inorganic carbon in soil using machine learning and digital soil mapping techniques. Chahardowli Plain in foothills of the Zagros Mountains in the southeast of Kurdistan Province in Iran was taken as a case study area. CCE was measured at 0-5, 5-15, 15-30, 30-60, and 60-100 cm depths following GloalSoilMap.net project specifications. A total of 145 samples were collected from 30 soil profiles using the conditional Latin hypercube (cLHS) method of sampling. Relationships between CCE and environmental predictors were modeled using random forest (RF) and decision tree (DT) models. In general, the RF model performed slightly superior than the DT model. The mean value of CCE increased with soil depth, from 3.5% (0-5 cm) to 63.8% (30-60 cm). Remote sensing (RS) variables and terrestrial variables were equally important. The importance of RS variables was higher at the surface than terrestrial variables, and vice versa. The most significant variables were Channel Network Base Level (CNBL) variable and Difference Vegetation Index (DVI) with the same variable importance value (21.1%). In areas affected by river activities, the use of the CNBL and vertical distance to channel networks (VDCN) as variables in digital soil mapping (DSM) could increase the accuracy of soil property prediction maps. The VDCN played a principal role in soil distribution in the study area by affecting the rate of discharge and, thus, erosion and sedimentation. A high percentage of carbonate in parts of the region could exacerbate nutrient deficiencies for most crops and provide information for sustainably managing agricultural activity.

Soil organic carbon stock retrieval from Sentinel-2A using a hybrid approach.

Digital soil maps find application in numerous fields, making their accuracy a crucial factor. Mapping soil properties in homogeneous landscapes where the soil surface is concealed, as in forests, presents a complex challenge. In this study, we evaluated the spatial distribution of soil organic carbon stocks (SOCstock) under forest vegetation using three methods: regression kriging (RK), random forest (RF), and RF combined with ordinary kriging of residuals (RFOK) in combination with Sentinel-2A satellite data. We also compared their accuracies and identified key influencing factors. We determined that SOCstock ranged from 0.6 to 10.9 kg/m2 with an average value of 4.9 kg/m2. Among the modelling approaches, we found that the RFOK exhibited the highest accuracy (RMSE = 1.58 kg/m2, NSE = 0.33), while the RK demonstrated a lack of spatial correlation of residuals, rendering this method inapplicable. An analysis of variable importance revealed that the SWIR B12 band of the Sentinel-2A satellite contributed the most to RFOK predictions. We concluded that the RFOK hybrid approach outperformed the others, potentially serving as a foundation for digital soil mapping under similar environmental conditions. Therefore, it is essential to consider spatial correlations when mapping soil properties in ecosystems that are inaccessible for capturing the spectral response of the soil surface.

Estimation of Soil Organic Carbon Content in the Ebinur Lake Wetland, Xinjiang, China, Based on Multisource Remote Sensing Data and Ensemble Learning Algorithms.

Soil organic carbon (SOC), as the largest carbon pool on the land surface, plays an important role in soil quality, ecological security and the global carbon cycle. Multisource remote sensing data-driven modeling strategies are not well understood for accurately mapping soil organic carbon. Here, we hypothesized that the Sentinel-2 Multispectral Sensor Instrument (MSI) data-driven modeling strategy produced superior outcomes compared to modeling based on Landsat 8 Operational Land Imager (OLI) data due to the finer spatial and spectral resolutions of the Sentinel-2A MSI data. To test this hypothesis, the Ebinur Lake wetland in Xinjiang was selected as the study area. In this study, SOC estimation was carried out using Sentinel-2A and Landsat 8 data, combining climatic variables, topographic factors, index variables and Sentinel-1A data to construct a common variable model for Sentinel-2A data and Landsat 8 data, and a full variable model for Sentinel-2A data, respectively. We utilized ensemble learning algorithms to assess the prediction performance of modeling strategies, including random forest (RF), gradient boosted decision tree (GBDT) and extreme gradient boosting (XGBoost) algorithms. The results show that: (1) The Sentinel-2A model outperformed the Landsat 8 model in the prediction of SOC contents, and the Sentinel-2A full variable model under the XGBoost algorithm achieved the best results R2 = 0.804, RMSE = 1.771, RPIQ = 2.687). (2) The full variable model of Sentinel-2A with the addition of the red-edge band and red-edge index improved R2 by 6% and 3.2% over the common variable Landsat 8 and Sentinel-2A models, respectively. (3) In the SOC mapping of the Ebinur Lake wetland, the areas with higher SOC content were mainly concentrated in the oasis, while the mountainous and lakeside areas had lower SOC contents. Our results provide a program to monitor the sustainability of terrestrial ecosystems through a satellite perspective.

Digital Mapping of Soil Organic Carbon Based on Machine Learning and Regression Kriging.

In the last two decades, machine learning (ML) methods have been widely used in digital soil mapping (DSM), but the regression kriging (RK) model which combines the advantages of the ML and kriging methods has rarely been used in DSM. In addition, due to the limitation of a single-model structure, many ML methods have poor prediction accuracy in undulating terrain areas. In this study, we collected the SOC content of 115 soil samples in a hilly farming area with continuous undulating terrain. According to the theory of soil-forming factors in pedogenesis, we selected 10 topographic indices, 7 vegetation indices, and 2 soil indices as environmental covariates, and according to the law of geographical similarity, we used ML and RK methods to mine the relationship between SOC and environmental covariates to predict the SOC content. Four ensemble models-random forest (RF), Cubist, stochastic gradient boosting (SGB), and Bayesian regularized neural networks (BRNNs)-were used to fit the trend of SOC content, and the simple kriging (SK) method was used to interpolate the residuals of the ensemble models, and then the SOC and residual were superimposed to obtain the RK prediction result. Moreover, the 115 samples were divided into calibration and validation sets at a ratio of 80%, and the tenfold cross-validation method was used to fit the optimal parameters of the model. From the results of four ensemble models: RF performed best in the calibration set (R2c = 0.834) but poorly in the validation set (R2v = 0.362); Cubist had good accuracy and stability in both the calibration and validation sets (R2c = 0.693 and R2v = 0.445); SGB performed poorly (R2c = 0.430 and R2v = 0.336); and BRNN had the lowest accuracy (R2c = 0.323 and R2v = 0.282). The results showed that the R2 of the four RK models in the validation set were 0.718, 0.674, 0.724, and 0.625, respectively. Compared with the ensemble models without superimposed residuals, the prediction accuracy was improved by 0.356, 0.229, 0.388, and 0.343, respectively. In conclusion, Cubist has high prediction accuracy and generalization ability in areas with complex topography, and the RK model can make full use of trends and spatial structural factors that are not easy to mine by ML models, which can effectively improve the prediction accuracy. This provides a reference for soil survey and digital mapping in complex terrain areas.

Digital soil mapping of soil total nitrogen based on Landsat 8, Sentinel 2, and WorldView-2 images in smallholder farms in Yellow River Basin, China.

Predicting spatial explicit information of soil nutrients is critical for sustainable soil management and food security under climate change and human disturbance in agricultural land. Digital soil mapping (DSM) techniques can utilize soil-landscape information from remote sensing data to predict the spatial pattern of soil nutrients, and it is important to explore the effects of remote sensing data types on DSM. This research utilized Landsat 8 (LT), Sentinel 2 (ST), and WorldView-2 (WV) remote sensing data and employed partial least squares regression (PLSR), random forest (RF), and support vector machine (SVM) algorithms to characterize the spatial pattern of soil total nitrogen (TN) in smallholder farm settings in Yellow River Basin, China. The overall relationships between TN and spectral indices from LT and ST were stronger than those from WV. Multiple red edge band-based spectral indices from ST and WV were relevant variables for TN, while there were no red band-based spectral indices from ST and WV identified as relevant variables for TN. Soil moisture and vegetation were major driving forces of soil TN spatial distribution in this area. The research also concluded that farmlands of crop rotation had relatively higher TN concentration compared with farmlands of monoculture. The soil prediction models based on WV achieved relatively lower model performance compared with those based on ST and LT. The effects of remote sensing data spectral resolution and spectral range on enhancing soil prediction model performance are higher than the effects of remote sensing data spatial resolution. Soil prediction models based on ST can provide location-specific soil maps, achieve fair model performance, and have low cost. This research suggests DSM research utilizing ST has relatively high prediction accuracy, and can produce soil maps that are fit for the spatial explicit soil management for smallholder farms.

Spatial modeling of soil organic carbon using remotely sensed indices and environmental field inventory variables.

The relationship between soil organic carbon (SOC) and environmental parameters was investigated in the Galazchai Watershed, Iran. Therefore, correlating the SOC amounts with remote sensing (RS) indices, topographic variables, and soil texture was analyzed. Some 125 soil samples gather from the upper 30 cm, and the weight of each sample was about 0.5 kg. The RS indices, consisting of difference vegetation index (DVI), enhanced vegetation index (EVI), optimized soil adjusted vegetation index (OSAVI), normalized difference vegetation index (NDVI), and soil adjusted vegetation index (SAVI), were used. Topographic variables included slope, elevation, aspect, and topographical wetness index (TWI), as well as clay and silt contents. The ordinary least square (OLS) and the geographically weighted regression (GWR) were employed to develop the SOC relationship considering different combinations of the variables. Results showed that none of the combinations of variables accurately estimated SOC (R2 < 0.32 and p value > 0.001). However, EVI with GWR (R2 = 0.291) and OSAVI, clay, slope, and aspect with GWR (R2= 0.32) better estimated SOC. Therefore, results showed that the study remotely sensed indices and environmental field inventory variables could not favorably predict the SOC content. These results can be attributed to the low SOC values varying from 0.917 to 3.355%, with a mean of 2.194 ± 0.522 in the study watershed. However, studies using more uniformly distributed and denser sampling in the study area and other methods to investigate the relationship between variables are recommended.

Accuracy and uncertainty of geostatistical models versus machine learning for digital mapping of soil calcium and potassium.

Accuracy and uncertainty of models used for digital soil mapping are important for assessing confidence of predictions and reliable land use planning and management. In this study, two approaches of geostatistical (spatial) and machine learning (ML) models were evaluated for predictive mapping of soil calcium (Ca) and potassium (K). Two spatial models including empirical Bayesian kriging (EBK) and sequential Gaussian simulation (SGS) were compared with machine learning models: Cubist, random forest (RF) and support vector machine (SVM) in terms of their accuracy and uncertainty for mapping soil Ca and K. The study area is in Nowley, New South Wales, Australia, with an area of 2083 ha and a variety of soil types and farming systems. For the models training process, 240 soil samples data and for validation 102 independent samples data were used. For accuracy assessment R2, root mean square error (RMSE), concordance and bias and for uncertainty assessment confidence limits were investigated. Also, in order to compare the outcomes for the two soil properties with different measurement units, mean absolute percentage error (MAPE) and relative uncertainty (RU) as accuracy and uncertainty measures, respectively, were evaluated. Results showed that for K map SGS had the highest R2 (0.74) and lowest RMSE (1.96), followed by EBK with R2 = 0.72 and RMSE = 2.02. For Ca map, EBK model showed the highest accuracy (R2 = 0.46; RMSE = 3.21), followed by SVM and SGS with comparable accuracies. Comparing the two soil properties, Ca map showed higher MAPE and RU, compared to K map. The lowest MAPE was obtained for EBK model (for K = 39) and SGS model (for K = 44). Also, the lowest RU values were found for EBK and SGS models. Among the ML models, SVM showed lower sensitivity to higher variance in data input. In general, the spatial models outperformed the ML models with regard to both accuracy and uncertainty. An additional conclusion is that considering the data variance in the two soil properties, geostatistical models with lower RU and MAPE were relatively less susceptible to data variance, compared to ML models.

Selecting optimal calibration samples using proximal sensing EM induction and γ-ray spectrometry data: An application to managing lime and magnesium in sugarcane growing soil.

Calcium (Ca) and magnesium (Mg) are essential for growth of sugarcane leaves and roots, as well as respiration and nitrogen metabolism, respectively. To assist farmers decide suitable application rates of lime and Mg fertiliser, respectively, the Australian sugarcane industry established the Six-Easy-Steps nutrient management guidelines based on topsoil (0-0.3 m) Ca (cmol(+) kg-1) and Mg (cmol(+) kg-1). Given the heterogeneous nature of soil, digital soil mapping (DSM) methods can be employed to allow for the precise application rate to be determined. In this study, we examine statistical models (i.e., ordinary kriging [OK], linear mixed model [LMM], quantile regression forests [QRF], support vector machine [SVM], and Cubist regression kriging [CubistRK]) to predict topsoil and subsoil (0.6-0.9) Ca and Mg, employing digital data in combination (i.e., proximal sensing electromagnetic induction (EMI) [e.g., 1mPcon, 1mHcon, etc.], gamma-ray [γ-ray] spectrometry [i.e., TC, K, U and Th] and digital elevation model [DEM] derivatives). We also investigate various sampling designs (i.e., spatial coverage [SCS], feature space coverage [FSCS], conditioned Latin hypercube [cLHS] and simple random sampling [SRS]) and calibration sample size (i.e., n = 180, 150, 120, 90, 60 and 30). The predictions are assessed using Lin's concordance correlation coefficient (LCCC) and ratio of performance to interquartile distance (RPIQ) with an independent validation dataset (i.e., n = 40). The best results were for prediction of subsoil Mg, utilising CubistRK (LCCC = 0.82) with the largest calibration sample size (n = 180), followed by LMM (0.79), SVM (0.76), QRF (0.70) and OK (0.65). This was generally the case for topsoil and subsoil Ca. We also conclude that no single sampling design was universally better, and 180 samples were necessary for predicting topsoil Ca and Mg with moderate agreement (0.65 < LCCC < 0.80). However, with FSCS, a minimum of 120 samples were enough to calibrate a CubistRK model and achieve substantial (LCCC > 0.80) agreement for predicting subsoil Ca and Mg. With respect to soil use and management according to the Six-Easy-Steps, the sandy soil in the north and south require large application rate of lime (3.5 t/ha) and Mg (125 kg/ha), respectively. Conversely, varying amounts of fertiliser rates of lime (2.0, 1.5 and 1 t/ha) and Mg (50 kg/ha) were recommended where Vertosols were previously mapped.