Estimating soil organic carbon density in plains using landscape metric-based regression Kriging model

Wu, Zihao; Wang, Bozhi; Huang, Junlong; An, Zihao; Jiang, Ping; Chen, Yiyun; Liu, Yanfang

doi:10.1016/j.still.2019.104381

Cited by 29 publications

(18 citation statements)

References 66 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Nevertheless, it remains a challenge to apply VIS-NIR spectroscopy to estimate SOC in anthropogenic soils that are characterized by high heterogeneity in the relationship between VIS-NIR spectra and SOC. The study area, Jianghan plain, has been under a long-term period of human activities with a highly fragmented landscape [52,60]. Our study reported that there was strong heterogeneity in the relationship between SOC and VIS-NIR spectra.…”

Section: The Effect Of Spectral Variable Selection Techniques On Model Accuracymentioning

confidence: 90%

“…The land-use types include cropland, woodland, and meadows. Cropland patches are highly fragmented, and some of them are close to settlements and various water bodies (breeding, ponds, irrigated canals, lakes, and rivers) [52]. Diverse land management practices are carried out in our study area according to our field survey.…”

Section: Sampling Area and Soil Samplesmentioning

confidence: 99%

See 1 more Smart Citation

Estimation of Organic Carbon in Anthropogenic Soil by VIS-NIR Spectroscopy: Effect of Variable Selection

Hong

et al. 2020

Remote Sensing

Self Cite

View full text Add to dashboard Cite

Visible and near-infrared reflectance (VIS-NIR) spectroscopy is widely applied to estimate soil organic carbon (SOC). Intense and diverse human activities increase the heterogeneity in the relationships between SOC and VIS-NIR spectra in anthropogenic soil. This fact results in poor performance of SOC estimation models. To improve model accuracy and parsimony, we investigated the performance of two variable selection algorithms, namely competitive adaptive reweighted sampling (CARS) and random frog (RF), coupled with five spectral pretreatments. A total of 108 samples were collected from Jianghan Plain, China, with the SOC content and VIS-NIR spectra measured in the laboratory. Results showed that both CARS and RF coupled with partial least squares regression (PLSR) outperformed PLSR alone in terms of higher model accuracy and less spectral variables. It revealed that spectral variable selection could identify important spectral variables that account for the relationships between SOC and VIS-NIR spectra, thereby improving the accuracy and parsimony of PLSR models in anthropogenic soil. Our findings are of significant practical value to the SOC estimation in anthropogenic soil by VIS-NIR spectroscopy.

show abstract

Section: The Effect Of Spectral Variable Selection Techniques On Model Accuracymentioning

confidence: 90%

Section: Sampling Area and Soil Samplesmentioning

confidence: 99%

Estimation of Organic Carbon in Anthropogenic Soil by VIS-NIR Spectroscopy: Effect of Variable Selection

Hong

et al. 2020

Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Regression Kriging is a widely used spatial interpolation technique in soil science, which combines a linear regression of dependent variable such as SOC stocks with environmental variables with kriging of the regression residuals ( Hengl et al, 2007 ; Keskin and grunwald, 2018 ; Wu et al, 2019 ). In this method, the SOC stocks at an unsampled location are predicted by adding the interpolated regression residuals into the regression predicted SOC stocks.…”

Section: Methodsmentioning

confidence: 99%

Ensemble Machine Learning Approach Improves Predicted Spatial Variation of Surface Soil Organic Carbon Stocks in Data-Limited Northern Circumpolar Region

et al. 2020

View full text Add to dashboard Cite

Various approaches of differing mathematical complexities are being applied for spatial prediction of soil properties. Regression kriging is a widely used hybrid approach of spatial variation that combines correlation between soil properties and environmental factors with spatial autocorrelation between soil observations. In this study, we compared four machine learning approaches (gradient boosting machine, multinarrative adaptive regression spline, random forest, and support vector machine) with regression kriging to predict the spatial variation of surface (0-30 cm) soil organic carbon (SOC) stocks at 250-m spatial resolution across the northern circumpolar permafrost region. We combined 2,374 soil profile observations (calibration datasets) with georeferenced datasets of environmental factors (climate, topography, land cover, bedrock geology, and soil types) to predict the spatial variation of surface SOC stocks. We evaluated the prediction accuracy at randomly selected sites (validation datasets) across the study area. We found that different techniques inferred different numbers of environmental factors and their relative importance for prediction of SOC stocks. Regression kriging produced lower prediction errors in comparison to multinarrative adaptive regression spline and support vector machine, and comparable prediction accuracy to gradient boosting machine and random forest. However, the ensemble median prediction of SOC stocks obtained from all four machine learning techniques showed highest prediction accuracy. Although the use of different approaches in spatial prediction of soil properties will depend on the availability of soil and environmental datasets and computational resources, we conclude that the ensemble median prediction obtained from multiple machine learning approaches provides greater spatial details and produces the highest prediction accuracy. Thus an ensemble prediction approach can be a better choice than any single prediction technique for predicting the spatial variation of SOC stocks.

show abstract

“…Critiques of DSM-based approaches suggest they may not adequately represent soil processes or changes in climate or management that influence SOC (31,32). Moreover, the relationship between SOC and environmental covariates is scale-dependent (33) but quantitative ranking of covariates contributing to SOC is usually lacking at the large spatial scale (34). To have confidence in predictions of SOC that are based on DSM we must gain a better understanding of strengths and weaknesses of SCORPAN-type efforts and explore the key controlling factors.…”

Section: Introductionmentioning

confidence: 99%

Digital Mapping of Agricultural Soil Organic Carbon Using Soil Forming Factors: A Review of Current Efforts at the Regional and National Scales

Xia

McSweeney²,

Wander³

2022

Front. Soil Sci.

View full text Add to dashboard Cite

To explore how well large spatial scale digital soil mapping can contribute to efforts to monitor soil organic carbon (SOC) stocks and changes, we reviewed regional and national studies quantifying SOC within lands dominated by agriculture using SCORPAN approaches that rely on soil (S), climate (C), organisms (O), relief (R), parent material (P), age (A), and space (N) covariates representing soil forming factors. After identifying 79 regional (> 10,000 km2) and national studies that attempted to estimate SOC, we evaluated model performances with reference to soil sampling depth, number of predictors, grid-distance, and spatial extent. SCORPAN covariates were then investigated in terms of their frequency of use and data sources. Lastly, we used 67 studies encompassing a variety of spatial scales to determine which covariates most influenced SOC in agricultural lands using a subjective ranking system. Topography (used in 94% of the cases), climate (87%), and organisms (86%) covariates that were the most frequently used SCORPAN predictors, aligned with the factors (precipitation, temperature, elevation, slope, vegetation indices, and land use) currently identified to be most influential for model estimate at the large spatial extent. Models generally succeeded in estimating SOC with fits represented by R2 with a median value of 0.47 but, performance varied widely (R2 between 0.02 and 0.86) among studies. Predictive success declined significantly with increased soil sampling depth (p < 0.001) and spatial extent (p < 0.001) due to increased variability. While studies have extensively drawn on large-scale surveys and remote sensing databases to estimate environmental covariates, the absence of soils data needed to understand the influence of management or temporal change limits our ability to make useful inferences about changes in SOC stocks at this scale. This review suggests digital soil mapping efforts can be improved through greater use of data representing soil type and parent material and consideration of spatio-temporal dynamics of SOC occurring within different depths and land use or management systems.

show abstract

Estimating soil organic carbon density in plains using landscape metric-based regression Kriging model

Cited by 29 publications

References 66 publications

Estimation of Organic Carbon in Anthropogenic Soil by VIS-NIR Spectroscopy: Effect of Variable Selection

Estimation of Organic Carbon in Anthropogenic Soil by VIS-NIR Spectroscopy: Effect of Variable Selection

Ensemble Machine Learning Approach Improves Predicted Spatial Variation of Surface Soil Organic Carbon Stocks in Data-Limited Northern Circumpolar Region

Digital Mapping of Agricultural Soil Organic Carbon Using Soil Forming Factors: A Review of Current Efforts at the Regional and National Scales

Contact Info

Product

Resources

About