Soil Science-Informed Machine Learning

Geoderma, Journal Year: 2024, Volume and Issue: 452, P. 117094 - 117094

Published: Nov. 14, 2024

Language: Английский

---

National-scale mapping of soil organic carbon stock in France: New insights and lessons learned by direct and indirect approaches

Soil & Environmental Health, Journal Year: 2023, Volume and Issue: 1(4), P. 100049 - 100049

Published: Nov. 11, 2023

Soil organic carbon (SOC) plays a crucial role in soil health and global cycling, therefore accurate estimates of its spatial distribution are important for managing mitigating climate change. Digital mapping shows potential to provide high-resolution SOC across scales. To convert content density (SOCD), two inference trajectories exist predicting SOCD digital mapping: the direct approach (calculate-then-model) indirect (model-then-calculate). However, there is lack comprehensive exploration regarding differences their performance estimates, particularly regions characterized by diverse pedoclimatic conditions. bridge this knowledge gap, we evaluated approaches based on model France. Using 916 topsoils (0−20 cm) from LUCAS 2018 24 environmental covariates, random forest forward recursive feature selection were used build predictive models using approaches. The results show that, full both showed moderate (R2 = 0.28−0.32). By utilizing model, number predictors was reduced 9, enhancing 0.35) with no improvement 0.28). mean French topsoil 5.29 6.14 kg m-2 approaches, resulting stock (SOCS) 2.8 3.3 Pg, respectively. We found that clearly underestimated high (>9 m-2), while performed much better SOCD. Our findings serve as valuable reference mapping, thereby providing scientific basis maintaining health.

Language: Английский

---

Digital soil mapping in the Russian Federation: A review

Geoderma Regional, Journal Year: 2024, Volume and Issue: 36, P. e00763 - e00763

Published: Jan. 18, 2024

Language: Английский

---

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

Journal of Environmental Management, Journal Year: 2024, Volume and Issue: 364, P. 121311 - 121311

Published: June 13, 2024

Language: Английский

---

BIS-4D: mapping soil properties and their uncertainties at 25 m resolution in the Netherlands

Earth system science data, Journal Year: 2024, Volume and Issue: 16(6), P. 2941 - 2970

Published: June 25, 2024

Abstract. In response to the growing societal awareness of critical role healthy soils, there has been an increasing demand for accurate and high-resolution soil information inform national policies support sustainable land management decisions. Despite advancements in digital mapping initiatives like GlobalSoilMap, quantifying variability its uncertainty across space, depth time remains a challenge. Therefore, maps key properties are often still missing on scale, which is also case Netherlands. To meet this challenge fill data gap, we introduce BIS-4D, modeling platform BIS-4D delivers texture (clay, silt sand content), bulk density, pH, total nitrogen, oxalate-extractable phosphorus, cation exchange capacity their uncertainties at 25 m resolution between 0 2 3D space. Additionally, it provides organic matter space 1953 2023 same range. The statistical model uses machine learning informed by observations amounting 3815 855 950, depending property, 366 environmental covariates. We assess accuracy mean median predictions using design-based inference probability sample location-grouped 10-fold cross validation (CV) prediction interval coverage probability. found that clay, pH was highest, with efficiency coefficient (MEC) ranging 0.6 0.92 depth. Silt, matter, nitrogen (MEC 0.27 0.78), especially phosphorus −0.11 0.38) were more difficult predict. One main limitations cannot be used quantify spatial aggregates. provide example good practice help users decide whether suitable intended purpose. An overview all can Supplement. Openly available code input enhance reproducibility future updates. readily downloaded https://doi.org/10.4121/0c934ac6-2e95-4422-8360-d3a802766c71 (Helfenstein et al., 2024a). fills previous gap national-scale GlobalSoilMap product Netherlands will hopefully facilitate inclusion as routine integral part decision systems.

Language: Английский

---

Handheld In Situ Methods for Soil Organic Carbon Assessment

Sustainability, Journal Year: 2024, Volume and Issue: 16(13), P. 5592 - 5592

Published: June 29, 2024

Soil organic carbon (SOC) assessment is crucial for evaluating soil health and supporting sequestration efforts. Traditional methods like wet digestion dry combustion are time-consuming labor-intensive, necessitating the development of non-destructive, cost-efficient, real-time in situ measurements. This review focuses on handheld methodologies SOC estimation, underscoring their practicality reasonable accuracy. Spectroscopic techniques, visible near-infrared, mid-infrared, laser-induced breakdown spectroscopy, inelastic neutron scattering each offer unique advantages. Preprocessing such as external parameter orthogonalization standard normal variate, employed to eliminate moisture content particle size effects estimation. Calibration methods, partial least squares regression support vector machine, establish relationships between spectral reflectance, properties, SOC. Among 32 studies selected this review, 14 exhibited a coefficient determination (R2) 0.80 or higher, indicating potential accurate estimation using approaches. Each study meticulously adjusted factors range, pretreatment method, calibration model improve accuracy content, highlighting both methodological diversity continuous pursuit precision direct field Continued research validation imperative ensure across diverse environments. Thus, underscores devices with good leveraging that influence its precision. Crucial optimizing farming, these measurements, empowering land managers enhance promote sustainable management agricultural landscapes.

Language: Английский

---

The validity domain of sensor fusion in sensing soil quality indicators

Geoderma, Journal Year: 2023, Volume and Issue: 438, P. 116657 - 116657

Published: Sept. 4, 2023

Soil health has gained increasing attention under the rapid development of industrialization and requirement for green agriculture. Therefore, up-to-date soil information related to is urgently needed ensure food security biodiversity protection. Previous studies have shown potential proximal sensing in measuring information, while it remains challenging get cost-efficient robust estimates multiple indicators simultaneously via sensor fusion. In this study, we investigated visible near-infrared (vis-NIR), mid-infrared (MIR) spectroscopy as well three model averaging methods predicting properties, including organic matter (SOM), pH, cation exchange capacity (CEC). The are not only used fusion but also high-level fusion, which include Granger-Ramanathan (GR), Bayesian Model Averaging Spectral-Guided Ensemble Modelling (S-GEM). Here, S-GEM a recently proposed algorithm that can improve spectroscopic prediction by spectral ensemble modelling. Four widely models were evaluated, partial least square regression, Cubist, memory based learning convolutional neural network. For SOM, on algorithms was comparable Sensorsingle + Modelmultiple (MIR singly S-GEM) with R2 0.86. However, MIR performed best among all (LCCC 0.92, RMSE 3.66 g kg−1 RPIQ 3.68). 10-fold cross-validation results indicated 0.84, LCCC 0.90, 0.45 3.65. CEC, Sensormultiple GR 0.66, 0.80, 3.48 cmol 2.22. Our showed failed when performance sensors differed lot (△R2 > 0.2), use single therefore suggested case. When close < recommended. outcome study provide reference determining validity domain improving accuracy prediction.

Language: Английский

---

High-resolution digital soil mapping of amorphous iron- and aluminium-(hydr)oxides to guide sustainable phosphorus and carbon management

Geoderma, Journal Year: 2024, Volume and Issue: 443, P. 116838 - 116838

Published: March 1, 2024

Amorphous iron- and aluminium-(hydr)oxides are key soil properties in controlling the dynamics of phosphorus availability carbon storage. These oxides affect potential soils to retain carbon, thus affecting ecosystem services such as crop production, water quality sequestration. In this study, we spatially predicted oxalate-extractable Fe Al (FeOX, AlOX) contents Netherlands at 25 m resolution across six depth layers between 0 200 cm quantified associated prediction uncertainty using quantile regression forest. For model training validation, geo-referenced data FeOX AlOX were used including 12,110 wet-chemical observations 102,393 NIR spectroscopy observations. Over 150 spatial covariates selected that provide information about typology, climate, organisms, land use, relief, parent material space (sampling oblique coordinates). Map was assessed by comparing predictions with an independent set 4841 samples from agricultural fields. Soil sample locations stratified random sampling, allowing us assess map design-based statistical inference. evaluated metrics Model Efficiency Coefficient (MEC), Root Mean Square Error (RMSE) (ME). differed, depending on target variable depth, MEC ranging 0.19 0.80, RMSE 13.5 56.9 mmol kg−1 ME −6.8 6.8 kg−1. Overall, better for topsoil than subsoil contents. Prediction calculating Interval Coverage Probability 90 per cent Interval, which close 0.90 all cases slightly below AlOX. Thus, uncertainties generally reliable, though underpredicted. The maps a valuable tool site-specific manure fertiliser management strategies aiming balance sequestration agriculture.

Language: Английский

---

Soil Organic Carbon Prediction Using Sentinel-2 Data and Environmental Variables in a Karst Trough Valley Area of Southwest China

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(8), P. 2118 - 2118

Published: April 17, 2023

Climate change is closely linked to changes in soil organic carbon (SOC) content, which affects the terrestrial cycle. Consequently, it essential for accounting and sustainable management predict SOC content accurately. Although there has been an extensive utilization of optical remote sensing data environmental factors few studies have explored their applicability karst areas. Therefore, remains unclear how can be accurately simulated these In this study, 160 samples, 8 covariates 14 variables were used build prediction models. Three machine learning models, i.e., support vector (SVM), random forest (RF) extreme gradient boosting (XGBoost), applied each three land use classes, including entire study area, as well farmland The with greatest influence bands, derived indices, precipitation temperature areas, band11 Pop-density farmland. results from suggest that RF XGBoost are superior SVM accuracy. Additionally, simulation accuracy model areas (R2 = 0.32, RMSE 6.81, MAE 5.63) 0.28, 4.03, 3.27) was greatest. based on different types could obtain a higher than whole area. These findings provide new insights estimation high precision

Language: Английский

---

Ensemble modelling-based pedotransfer functions for predicting soil bulk density in China

Geoderma, Journal Year: 2024, Volume and Issue: 448, P. 116969 - 116969

Published: July 15, 2024

Understanding and managing soil organic carbon stocks (SOCS) are integral to ensuring environmental sustainability the health of terrestrial ecosystems. The information bulk density (BD) is important in accurately determining SOCS while it often missing database. Using 3,504 profiles (14,170 samples) that represented diverse regions across China, we investigated effectiveness various pedotransfer functions (PTFs), including traditional PTFs, machine learning (ML), ensemble model (EM), predicting BD. results showed refitting parameter(s) PTFs was essential for BD prediction (coefficient determination (R2) 0.299–0.432, root mean squared error (RMSE) 0.156–0.162 g cm−3, Lin's concordance coefficient (LCCC) 0.428–0.605). Compared ML can greatly improve performance with R2 0.425–0.616, RMSE 0.129–0.158 cm−3 LCCC 0.622–0.765. Our also EM further by ensembling four models (R2 = 0.630, 0.126 0.775). model, filled (1207 3,112 our database built SOC stock (4,275 17,282 samples). This study be a good reference gap-filling depending on data availability, thus contribute deeper understanding C related climate change mitigation, ecological balance preservation promotion.

Language: Английский

---