Estimating Soil Organic Carbon Research Articles

A Deep Learning Approach to Estimate Soil Organic Carbon from Remote Sensing

Monitoring soil organic carbon (SOC) typically assumes conducting a labor-intensive soil sampling campaign, followed by laboratory testing, which is both expensive and impractical for generating useful, spatially continuous data products. The present study leverages the power of machine learning (ML) and, in particular, deep neural networks (DNNs) for segmentation, as well as satellite imagery, to estimate the SOC remotely. We propose a new two-stage pipeline for remote SOC estimation, which relies on using a DNN trained to classify land cover to perform feature extraction, while the SOC estimation is performed by a different ML model. The first stage is an image segmentation DNN with the U-Net architecture, which is trained to estimate the land cover for an observed geographical region, based on multi-spectral images taken by the Sentinel-2 satellite constellation. This estimator is subsequently used to extract the latent feature vector for each of the output pixels, by rolling back from the output (dense) layer of the U-Net and accessing the last available convolutional layer of the same dimension as our desired output. The second stage is trained on a set of feature vectors extracted at the coordinates for which manual SOC measurements exist. We tested a variety of ML models and report on their performance. Using the best extremely randomized trees model, we generated a spatially continuous map of SOC estimations for the region of Tuscany, in Italy, with a resolution of 10 m, to share with the researchers as a means of validating the results and to demonstrate the efficiency of the proposed approach, which can can easily be scaled to create a global continuous SOC map.

Upscaling Soil Organic Carbon Measurements at the Continental Scale Using Multivariate Clustering Analysis and Machine Learning

AbstractEstimates of soil organic carbon (SOC) stocks are essential for many environmental applications. However, significant inconsistencies exist in SOC stock estimates for the U.S. across current SOC maps. We propose a framework that combines unsupervised multivariate geographic clustering (MGC) and supervised Random Forests regression, improving SOC maps by capturing heterogeneous relationships with SOC drivers. We first used MGC to divide the U.S. into 20 SOC regions based on the similarity of covariates (soil biogeochemical, bioclimatic, biological, and physiographic variables). Subsequently, separate Random Forests models were trained for each SOC region, utilizing environmental covariates and SOC observations. Our estimated SOC stocks for the U.S. (52.6 ± 3.2 Pg for 0–30 cm and 108.3 ± 8.2 Pg for 0–100 cm depth) were within the range estimated by existing products like Harmonized World Soil Database, HWSD (46.7 Pg for 0–30 cm and 90.7 Pg for 0–100 cm depth) and SoilGrids 2.0 (45.7 Pg for 0–30 cm and 133.0 Pg for 0–100 cm depth). However, independent validation with soil profile data from the National Ecological Observatory Network showed that our approach (R2 = 0.51) outperformed the estimates obtained from Harmonized World Soil Database (R2 = 0.23) and SoilGrids 2.0 (R2 = 0.39) for the topsoil (0–30 cm). Uncertainty analysis (e.g., low representativeness and high coefficients of variation) identified regions requiring more measurements, such as Alaska and the deserts of the U.S. Southwest. Our approach effectively captures the heterogeneous relationships between widely available predictors and the current SOC baseline across regions, offering reliable SOC estimates at 1 km resolution for benchmarking Earth system models.

Relative and absolute difference in soil organic carbon stocks in grassland soils in Ireland: Impact of rock fragments, bulk density and calculation methods

On-farm management measures can sequester carbon by removing carbon dioxide from the atmosphere and storing it securely for long periods, in our soils. Estimates of soil organic carbon (SOC) stocks differ by calculation methods in terms of how they account for the coarse >2 mm fraction (i.e. rock fragments) of the soil. Irish grassland soils have variable volumetric stone contents and the extent of overestimation is unknown. Without best practice guidance, calculations of SOC in these soils are vulnerable to overestimation. The objectives of the present study for grassland soils in Ireland were to: 1) calculate mean SOC stocks using different methods by soil horizon and volumetric stone content class; 2) highlight relative differences in SOC stocks between these calculation methods and 3) compare normalised layer (Ly, fixed depth) versus horizon (Hz) type derived sampling data in terms of SOC stocks, to establish a best practice standardised sampling protocol. Five methods (M, 1–3 overestimate SOC stocks, whereas 4–5 are equivalent and are closest to reality) to calculate SOC stocks were used in combination with three soil databases i.e. Irish Soil Information System (I-SIS), the Heavy Soils Programme (HSP) and the Soil Quality Assessment Research Project (SQUARE). Objectives 1 and 2 were assessed using all three databases across methods, whereas objective 3 used the SQUARE database and M4. These data were sorted into volumetric stone content (%) classes (< 5%, 5–10%, 10–20%, 20–30% and > 30%) and sampling horizon (O, A, E, A/B, B, B/C, and C) with SOC stocks calculated using M1–5 equations. Results confirmed the high volumetric stone content assumption of grassland soils in Ireland (exception O horizons). Volumetric stone contents ranged from 0% to 36% for all horizons. When M1–5 are considered the average relative difference in SOC stocks between M1 and M5 was 18.9% (max: 388%), M2 and M5 was 7.7% (max: 60.7%) and M3 and M5 was 8.9 (max: 310%). In terms of Ly versus Hz sampling, no significant differences were found for the A horizons sampled. However, in Ireland, best practice should still utilise elements from both Ly (using an auger to get fixed depth soil samples; analysed for organic C) and Hz (using a soil profile pit to extract representative BD cores and to gather bulk samples for volumetric stone content analysis) approaches in the field. Laboratory methods used must fit with the needs and assumptions of equations M4/5 to calculate SOC stocks (thereby minimising overestimation and indeed underestimation).

Uncertainty Quantification of Soil Organic Carbon Estimation from Remote Sensing Data with Conformal Prediction

Soil organic carbon (SOC) contents and stocks provide valuable insights into soil health, nutrient cycling, greenhouse gas emissions, and overall ecosystem productivity. Given this, remote sensing data coupled with advanced machine learning (ML) techniques have eased SOC level estimation while revealing its patterns across different ecosystems. However, despite these advances, the intricacies of training reliable and yet certain SOC models for specific end-users remain a great challenge. To address this, we need robust SOC uncertainty quantification techniques. Here, we introduce a methodology that leverages conformal prediction to address the uncertainty in estimating SOC contents while using remote sensing data. Conformal prediction generates statistically reliable uncertainty intervals for predictions made by ML models. Our analysis, performed on the LUCAS dataset in Europe and incorporating a suite of relevant environmental covariates, underscores the efficacy of integrating conformal prediction with another ML model, specifically random forest. In addition, we conducted a comparative assessment of our results against prevalent uncertainty quantification methods for SOC prediction, employing different evaluation metrics to assess both model uncertainty and accuracy. Our methodology showcases the utility of the generated prediction sets as informative indicators of uncertainty. These sets accurately identify samples that pose prediction challenges, providing valuable insights for end-users seeking reliable predictions in the complexities of SOC estimation.

Soil organic carbon stocks in native forest of Argentina: a useful surrogate for mitigation and conservation planning under climate variability

BackgroundThe nationally determined contribution (NDC) presented by Argentina within the framework of the Paris Agreement is aligned with the decisions made in the context of the United Nations Framework Convention on Climate Change (UNFCCC) on the reduction of emissions derived from deforestation and forest degradation, as well as forest carbon conservation (REDD+). In addition, climate change constitutes one of the greatest threats to forest biodiversity and ecosystem services. However, the soil organic carbon (SOC) stocks of native forests have not been incorporated into the Forest Reference Emission Levels calculations and for conservation planning under climate variability due to a lack of information. The objectives of this study were: (i) to model SOC stocks to 30 cm of native forests at a national scale using climatic, topographic and vegetation as predictor variables, and (ii) to relate SOC stocks with spatial–temporal remotely sensed indices to determine biodiversity conservation concerns due to threats from high inter-annual climate variability.MethodsWe used 1040 forest soil samples (0–30 cm) to generate spatially explicit estimates of SOC native forests in Argentina at a spatial resolution of approximately 200 m. We selected 52 potential predictive environmental covariates, which represent key factors for the spatial distribution of SOC. All covariate maps were uploaded to the Google Earth Engine cloud-based computing platform for subsequent modelling. To determine the biodiversity threats from high inter-annual climate variability, we employed the spatial–temporal satellite-derived indices based on Enhanced Vegetation Index (EVI) and land surface temperature (LST) images from Landsat imagery.ResultsSOC model (0–30 cm depth) prediction accounted for 69% of the variation of this soil property across the whole native forest coverage in Argentina. Total mean SOC stock reached 2.81 Pg C (2.71–2.84 Pg C with a probability of 90%) for a total area of 460,790 km2, where Chaco forests represented 58.4% of total SOC stored, followed by Andean Patagonian forests (16.7%) and Espinal forests (10.0%). SOC stock model was fitted as a function of regional climate, which greatly influenced forest ecosystems, including precipitation (annual mean precipitation and precipitation of warmest quarter) and temperature (day land surface temperature, seasonality, maximum temperature of warmest month, month of maximum temperature, night land surface temperature, and monthly minimum temperature). Biodiversity was influenced by the SOC levels and the forest regions.ConclusionsIn the framework of the Kyoto Protocol and REDD+, information derived in the present work from the estimate of SOC in native forests can be incorporated into the annual National Inventory Report of Argentina to assist forest management proposals. It also gives insight into how native forests can be more resilient to reduce the impact of biodiversity loss.

Factors controlling the spatial heterogeneity of soil organic carbon concentrations and stocks in a boreal forest

Boreal soils play a crucial role in the global terrestrial carbon cycle, serving as significant carbon reservoirs. However, this region is experiencing rapid climate change which poses a threat to the stability of the soil carbon pool. To understand the carbon balance in the boreal region, it is essential to obtain accurate estimates of soil organic carbon (SOC) stocks. The objective of this study was to investigate factors controlling the spatial heterogeneity of SOC concentrations and stocks in a boreal forested catchment and to quantify spatial variability of SOC concentrations and stocks using a random forest-based regression kriging approach. Soil carbon content, gravel content and bulk density were obtained from the surface soil layer (0–30 cm) of 106 samples across a 390 ha catchment in northwestern Ontario, Canada. Spatial attributes including elevation, wetness index, distance to nearest ridge, relative position index, plan curvature, profile curvature and normalized difference vegetation index were used in the random forest-based regression kriging to estimate spatial heterogeneity of SOC stocks and concentrations across the catchment. Our results showed that shallow soils in well-drained upland areas contained low to moderate SOC concentrations (1.2–16.2%) while depressional wetlands had high concentrations (19.2–54.0%). Topographic variables, including wetness index were the strongest predictors of SOC concentrations, but SOC stocks were very heterogeneous across the study site (12.9–139.0 Mg ha−1) and were not well explained by topographic or vegetation variables. The gravel content was highly variable (0–29.0%), and without accounting for this fraction, SOC stocks were overestimated by 52.5% in this landscape. These results emphasize the substantial catchment scale variability in SOC stocks in boreal landscapes and underscore the significance of considering both landscape-scale drivers and intrinsic soil properties when estimating soil carbon in this region. By understanding these factors, we can better manage and protect the important carbon stores in boreal soils.

Soil organic carbon estimation using remote sensing data-driven machine learning.

Soil organic carbon (SOC) is a crucial component of the global carbon cycle, playing a significant role in ecosystem health and carbon balance. In this study, we focused on assessing the surface SOC content in Shandong Province based on land use types, and explored its spatial distribution pattern and influencing factors. Machine learning methods including random forest (RF), extreme gradient boosting (XGBoost), and support vector machine (SVM) were employed to estimate the surface SOC content in Shandong Province using diverse data sources like sample data, remote sensing data, socio-economic data, soil texture data, topographic data, and meteorological data. The results revealed that the SOC content in Shandong Province was 8.78 g/kg, exhibiting significant variation across different regions. Comparing the model error and correlation coefficient, the XGBoost model showed the highest prediction accuracy, with a coefficient of determination (R²) of 0.7548, root mean square error (RMSE) of 7.6792, and relative percentage difference (RPD) of 1.1311. Elevation and Clay exhibited the highest explanatory power in clarifying the surface SOC content in Shandong Province, contributing 21.74% and 13.47%, respectively. The spatial distribution analysis revealed that SOC content was higher in forest-covered mountainous regions compared to cropland-covered plains and coastal areas. In conclusion, these findings offer valuable scientific insights for land use planning and SOC conservation.

Changes in Soil Organic Carbon Stocks Due to Land Use Changes in the Extended São Francisco River Basin

As the largest carbon reservoir in terrestrial ecosystems, soils play a critical role in food production, mitigation and adaption to climate change, and sustainability of agroecosystems. In this context, the aim of this study is to estimate variations in soil carbon stocks resulting from land use and management changes in different biomes that compose the Bacia Estendida do Rio São Francisco (BESF - Extended São Francisco River Basin), between 1985 and 2017. For this, remote sensing data and information from the IBGE agriculture and livestock census were used, in addition to emission factors to estimate soil organic carbon (SOC) changes. The results indicate that BESF had about 5.70 million ha degraded in the analyzed period, in addition to an increase of 0.72 Tg C year-1 in SOC stocks. The sub-medium São Francisco River basin recorded the highest SOC gain, with an increase of 0.54 Tg C ha-1; on the other hand, the sub-medium São Francisco River sub-basin had the greatest SOC losses, with an estimated reduction of 0.07 Tg C year-1. In short, this study provides important evidence on changes in SOC stocks in the region, emphasizing the importance of native vegetation conversion to agriculture and livestock systems under sustainable soil management for mitigating greenhouse gas emissions and maintaining soil quality.

Regional topsoil organic carbon content in the agricultural soils of Slovakia and its drivers, as revealed by the most recent national soil monitoring data

Soil organic carbon (SOC) is a primary constituent of soil organic matter and plays an important role in the regulation of many soil processes, including greenhouse gas emissions. Recently, SOC also became an indicator for monitoring climate change mitigation policies in the agricultural sector. The availability of up-to-date SOC inventories is thus crucial in terms of supporting SOC–related actions at country or sub-country scales. Currently, the National Monitoring System of the Agricultural Soils of Slovakia (CMS-P), whose network of 318 monitoring sites was last surveyed in 2018, is the only available source of up-to-date topsoil SOC data for agricultural land in Slovakia. Although very useful at the national scale, the number of CMS-P observations it contains is too limited for much needed sub-national SOC inventories. We hypothesized that with the aid of well-chosen macro-scale drivers of topsoil SOC accumulation in agricultural land in Slovakia, and by mapping those drivers geographically, we could upscale the CMS-P observations and produce a regional estimate of topsoil SOC. Altitude, land cover, topsoil texture, and soil type were assumed to be the key factors controlling topsoil SOC accumulation in Slovakia, and based on these, the country was classified into 14 macro-scale geographical regions. Typical ranges and mid-class values of 0–30cm topsoil SOC concentrations (%) and stocks (t ha−1) were calculated for each macro-scale region from CMS-P data. The average topsoil SOC content in agricultural land was estimated to be 2.13% (72.9 t ha−1). The highest topsoil SOC stock (> 90 t ha−1) was estimated for the lowlands of Slovakia, and the lowest (< 50 t ha−1) for the shallow and stony soils of mountain regions. When aggregated to 78 administrative regions at LAU1 level, the area-weighted averages ranged between 39.20 t ha−1 and 80.0 t ha−1, with the highest values (> 65 t ha−1) being in LAU1 regions in the south-west, south-east, and north of Slovakia where arable land is most prevalent. Total SOC storage in 0–30cm topsoil of agricultural land in Slovakia was estimated at 118.39 Mt, with two-thirds of this amount stored in arable soils in 33 south-west, south-east, and south LAU1 administrative regions. As there is no alternative and up-to-date dataset on topsoil SOC content in Slovakia, the upscaling algorithm presented in this study is an important step toward utilizing CMS-P data for sub-national SOC inventories. It may also offer a new way of providing inputs to help predict future or alternative regional topsoil SOC accumulation trajectories in Slovakian agricultural land using process-based or statistical models.

Preliminary Results in Innovative Solutions for Soil Carbon Estimation: Integrating Remote Sensing, Machine Learning, and Proximal Sensing Spectroscopy

This paper explores the application and advantages of remote sensing, machine learning, and mid-infrared spectroscopy (MIR) as a popular proximal sensing spectroscopy tool in the estimation of soil organic carbon (SOC). It underscores the practical implications and benefits of the integrated approach combining machine learning, remote sensing, and proximal sensing for SOC estimation and prediction across a range of applications, including comprehensive soil health mapping and carbon credit assessment. These advanced technologies offer a promising pathway, reducing costs and resource utilization while improving the precision of SOC estimation. We conducted a comparative analysis between MIR-predicted SOC values and laboratory-measured SOC values using 36 soil samples. The results demonstrate a strong fit (R² = 0.83), underscoring the potential of this integrated approach. While acknowledging that our analysis is based on a limited sample size, these initial findings offer promise and serve as a foundation for future research. We will be providing updates when we obtain more data. Furthermore, this paper explores the potential for commercialising these technologies in Australia, with the aim of helping farmers harness the advantages of carbon markets. Based on our study’s findings, coupled with insights from the existing literature, we suggest that adopting this integrated SOC measurement approach could significantly benefit local economies, enhance farmers’ ability to monitor changes in soil health, and promote sustainable agricultural practices. These outcomes align with global climate change mitigation efforts. Furthermore, our study’s approach, supported by other research, offers a potential template for regions worldwide seeking similar solutions.

Sentinel-1 Imagery Used for Estimation of Soil Organic Carbon by Dual-Polarization SAR Vegetation Indices

Despite optical remote sensing (and the spectral vegetation indices) contributions to digital soil-mapping studies of soil organic carbon (SOC), few studies have used active radar remote sensing mission data like that from synthetic aperture radar (SAR) sensors to predict SOC. Bearing in mind the importance of SOC mapping for agricultural, ecological, and climate interests and also the recently developed methods for vegetation monitoring using Sentinel-1 SAR data, in this work, we aimed to take advantage of the high operationality of Sentinel-1 imaging to test the accuracy of SOC prediction at different soil depths using machine learning systems. Using linear, nonlinear, and tree regression-based methods, it was possible to predict the SOC content of soils from western Bahia, Brazil, a region with predominantly sandy soils, using as explanatory variables the SAR vegetation indices. The models fed with SAR sensor polarizations and vegetation indices produced more accurate results for the topsoil layers (0–5 cm and 5–10 cm in depth). In these superficial layers, the models achieved an RMSE in the order of 5.0 g kg−1 and an R2 ranging from 0.16 to 0.24, therefore explaining about 20% of SOC variability using only Sentinel-1 predictors.

Potential of globally distributed topsoil mid-infrared spectral library for organic carbon estimation

Accurate monitoring of soil organic carbon (SOC) is critical for sustainable management of soil for improving its quality, function, and carbon sequestration. As a nondestructive, efficient, and low-cost technique, mid-infrared (MIR) spectroscopy has shown a great potential in rapid estimation of SOC, despite limited studies of the global scale. The objective of this work was to use a globally distributed topsoil MIR spectral library with 33,039 samples to predict SOC using different modeling methods. Effects of nine fractional-order derivatives (FODs) on the predicted accuracy of SOC were evaluated using four regression algorithms (i.e., ratio index-based linear regression, RI-LR; partial least squares regression, PLSR; Cubist; convolutional neural network, CNN). Square-root transformation to SOC data was performed to minimize the skewness and non-linearity. Results indicated FOD to capture the subtle spectral details related to SOC, leading to improved predictions that may not be possible by the raw absorbance and common integer-order derivatives. Concerning the RI-LR models, the optimal validation result for SOC was obtained by 0.75-order derivative, with the ratio of performance to inter-quartile distance (RPIQ) of 1.85. Regarding the full-spectrum modeling for SOC, the CNN outperformed PLSR and Cubist models, irrespective of raw absorbance or eight FODs; the best-performing CNN model was achieved by 1.25-order derivative (validation RPIQ = 6.33). It can be concluded that accurate estimation of SOC using large and diverse MIR spectral library at the global scale combined with deep-learning CNN model is feasible. This global-scale database is extremely valuable for us to deal with the shortage of soil data and to monitor the soils at different geographical scales.

Evaluation and mapping soil organic carbon in seasonally frozen ground on the Tibetan Plateau

Seasonally frozen ground (SFG) significantly contributes to global carbon sinks. Global warming and anthropogenic-induced disturbances threaten the carbon storage capacity of SFG. Challenges in evaluating the SFG carbon storage potential include the lack of understanding of the control mechanisms of soil organic carbon (SOC) variations and timely spatial estimates of SOC. In this study, we investigated SOC stocks in SFG in the Tibet Autonomous Region, China, in 2020 and 2021. We employed partial least squares structural equation modeling (PLS-SEM) to explore the effect of complex processes (interacting roles of climate, plant physiology and phenology, freeze–thaw cycle, soil environment, and C inputs) on SOC and mapped SOC stocks in the topmost 30 cm. We identified four causal pathways: (1) an indirect pathway representing the effect of climate on plant physiology and phenology through changes in freeze–thaw cycles and soil environment, (2) an indirect pathway representing the effect of climate on C inputs through changes in freeze–thaw cycles, soil environment and plant physiology and phenology, (3) an indirect pathway representing the effect of climate on freeze–thaw cycles, and (4) an indirect pathway representing the effect of climate on the soil environment through changes in freeze–thaw cycles. C inputs exerted the greatest control on SOC. The effect of these factors decreased with increasing soil depth. We used PLS-SEM to generate maps of SOC stocks in SFG at a 500 m resolution with a moderate accuracy. The estimated mean SOC stocks in the 0–30 cm layer reached 6.87 kg m−2, with a 95% confidence interval ranging from 6.2 to 7.5 kg m−2. Our results indicated that it is critical to consider the depth dependence of the direct and indirect effects of environmental factors when assessing the control mechanisms of SOC variations. In this work, we also demonstrated that spatially explicit SOC estimates based on timely investigations are important for assessing C stocks against the background of considerable environmental changes across the Tibetan Plateau.

Improving the Accuracy of Soil Organic Carbon Estimation: CWT-Random Frog-XGBoost as a Prerequisite Technique for In Situ Hyperspectral Analysis

Rapid and accurate measurement of the soil organic carbon (SOC) content is a pre-condition for sustainable grain production and land development, and contributes to carbon neutrality in the agricultural industry. To provide technical support for the development and utilization of land resources, the SOC content can be estimated using Vis-NIR diffuse reflectance spectroscopy. However, the spectral redundancy and co-linearity issues of Vis-NIR spectra pose extreme challenges for spectral analysis and model construction. This study compared the effects of different pre-processing methods and feature variable algorithms on the estimation of the SOC content. To this end, in situ hyperspectral data and soil samples were collected from the lakeside oasis of Bosten Lake in Xinjiang, China. The results showed that the combination of continuous wavelet transform (CWT)-random frog could rapidly estimate the SOC content with excellent estimation accuracy (R2 of 0.65–0.86). The feature variable selection algorithm effectively improved the estimation accuracy (average improvement of (0.30–0.48); based on their ability to improve model estimation on average, the algorithms can be ranked as follows: particle swarm optimization (PSO) &gt; ant colony optimization (ACO) &gt; random frog &gt; Boruta &gt; simulated annealing (SA) &gt; successive projections algorithm (SPA). The CWT-XGBoost model based on random frog showed the best results, with R2 = 0.86, RMSE = 2.44, and RPD = 2.78. The feature bands accounted for only 0.57% of the Vis-NIR bands, and the most important sensitive bands were distributed at 755–1195 nm, 1602 nm, 1673 nm, and 2213 nm. These findings are of significance for the extraction of precise information on lakeside oases in arid areas, which would aid in achieving human–land sustainability.

Estimating soil organic carbon content of multiple soil horizons in the middle and upper reaches of the Heihe River Basin

Estimating the spatial distribution of soil organic carbon (SOC) content is essential for evaluating ecosystem functions and establishing global climate change models. Most of the previous SOC mapping studies have focused on estimating SOC content at fixed depths. However, soils at the same depth are not always belonging to the same soil horizon. Therefore, the characteristics of soil horizons should be considered. In this study, the SOC content of multiple soil horizons in the middle and upper reaches of the Heihe River Basin was estimated using multivariate factors and ensemble learning (EL). The results show that EL obtains the best SOC content estimation performance, where the R2 values are 0.58, 0.57 and 0.69 for the O-horizon, A-horizon, and B-horizon, respectively. Environmental factors show a high degree of contribution to estimate SOC content, and mean annual precipitation (MAP) presents the highest contribution amongst all the factors. For the middle and upper reaches of the Heihe River Basin, the SOC content is high in the southwest and low in the northeast. Amongst the multiple soil horizons, the O-horizon usually showed higher SOC content than the A-horizon and the B-horizon. But the SOC content of the A-horizon is higher than that of the O-horizon for farmland. From 2012 to 2013, the SOC content of forestland and farmland increased in the O-horizon and A-horizon. The application of EL improves the mapping performance of SOC content in different soil horizons, and the involvement of multivariable factors reveals the distribution of SOC content in the Heihe River Basin. This study creates maps of SOC content in different soil horizons and analyses the effect of land covers on SOC content, which provides insights for ecological conservation and regional sustainable development.

Using Soil Testing Data to Monitor Changes in Maine Agricultural Soil Organic Carbon from 1995 to 2021

ABSTRACT Estimates of soil organic matter (SOM) are commonly part of soil testing, and typically use the loss-on-ignition (LOI) method. To explore the utility of existing soil testing databases as a resource for estimating soil organic carbon (SOC) concentrations over time, we determined the relationship between SOM estimated by LOI and SOC measured using IR detection of CO2 for Maine soils. We then examined trends in SOC concentrations in Maine agricultural soils over the past 27 years using data from 97,713 standard soil samples received by the Maine Soil Testing Service. For 91 agricultural soil samples representative of Maine cropping systems, we found that the standard LOI temperature used by the Maine Soil Testing Service, 375°C, produced LOI values that correlated more strongly with SOC measured by IR detection of CO2 than the three higher temperatures we studied, although all temperatures produced data that was well correlated (R2 values from 0.95 to 0.98). Soil texture did not influence the correlation. Our analysis of soil testing data from 1995 to 2021 suggested increases in SOC concentrations in soils planted to grain and hay, and possibly corn, broadly corresponding with changing agricultural practices occurring over this time frame. Of our four focal crop categories, only potato soils decreased in SOC concentration from 1995 to 2021, underscoring the unique challenges associated with potato soil management relative to other cropping systems in Maine. Our study confirms the potential utility of soil testing databases in assessing changes in SOC concentrations in agricultural soils over decades.

Assessment of above ground biomass and soil organic carbon in the forests of Nepal under climate change scenario

IntroductionMany factors, such as climate, topography, forest management, or tree/forest attributes, influence soil organic carbon (SOC) and above-ground tree biomass (AGTB). This study focuses on assessing relationship between various predictor variables and response variables (SOC and AGTB) in the perspective of climate change scenario. The study was conducted throughout in Nepal using forest resource assessment data (2010–2014).MethodsOur study applied a random forest model to assess the status of SOC and AGTB under future climate change scenarios using 19 bioclimatic variables accompanied by other variables such as altitude, aspect, basal area, crown cover development status, distance to settlement forest types, number of trees, macro-topography, management regime, physiographic zones, slope, and soil depth. The study used 737 (70%) samples as a training data for model development while 312 (30%) samples as a testing data for model validation.Results and discussionThe respective RMSE, RMSE% and adjusted R2 of the Random Forest Model for SOC estimation were found to be 9.53 ton/ha, 15% and 0.746 while same for the AGTB were 37.55 ton/ha, 21.74% and 0.743. Particularly, changes in temperature and precipitation showed an effect on the amount of SOC and AGTB in the projected scenario i.e., CMIP6, SSP2 4.5 for 2040–2060. The study found the amount of SOC decreased by 3.85%, while AGTB increased by 2.96% in the projected scenario. The proposed approach which incorporates the effect of bioclimatic variables can be a better option for understanding the dynamics of SOC and AGTB in the future using climatic variables.

Satellite-based estimation of soil organic carbon in Portuguese grasslands

Introduction:Soil organic carbon (SOC) sequestration is one of the main ecosystem services provided by well-managed grasslands. In the Mediterranean region, sown biodiverse pastures (SBP) rich in legumes are a nature-based, innovative, and economically competitive livestock production system. As a co-benefit of increased yield, they also contribute to carbon sequestration through SOC accumulation. However, SOC monitoring in SBP require time-consuming and costly field work.Methods:In this study, we propose an expedited and cost-effective indirect method to estimate SOC content. In this study, we developed models for estimating SOC concentration by combining remote sensing (RS) and machine learning (ML) approaches. We used field-measured data collected from nine different farms during four production years (between 2017 and 2021). We utilized RS data from both Sentinel-1 and Sentinel-2, including reflectance bands and vegetation indices. We also used other covariates such as climatic, soil, and terrain variables, for a total of 49 inputs. To reduce multicollinearity problems between the different variables, we performed feature selection using the sequential feature selection approach. We then estimated SOC content using both the complete dataset and the selected features. Multiple ML methods were tested and compared, including multiple linear regression (MLR), random forests (RF), extreme gradient boosting (XGB), and artificial neural networks (ANN). We used a random cross-validation approach (with 10 folds). To find the hyperparameters that led to the best performance, we used a Bayesian optimization approach.Results:Results showed that the XGB method led to higher estimation accuracy than the other methods, and the estimation performance was not significantly influenced by the feature selection approach. For XGB, the average root mean square error (RMSE), measured on the test set among all folds, was 2.78 g kg−1(r2equal to 0.68) without feature selection, and 2.77 g kg−1(r2equal to 0.68) with feature selection (average SOC content is 13 g kg−1). The models were applied to obtain SOC content maps for all farms.Discussion:This work demonstrated that combining RS and ML can help obtain quick estimations of SOC content to assist with SBP management.

China's wetland soil organic carbon pool: New estimation on pool size, change, and trajectory.

Robust estimates of wetland soil organic carbon (SOC) pools are critical to understanding wetland carbon dynamics in the global carbon cycle. However, previous estimates were highly variable and uncertain, due likely to the data sources and method used. Here we used machine learning method to estimate SOC storage and their changes over time in China's wetlands based on wetland SOC density database, associated geospatial environmental data, and recently published wetland maps. We built a database of wetland SOC density in China that contains 809 samples from 181 published studies collected over the last 20 years as presented in the published literature. All samples were extended and standardized to a 1-m depth, on the basis of the relationship between SOC density data from soil profiles of different depths. We used three different machine learning methods to evaluate their robustness in estimating wetland SOC storage and changes in China. The results indicated that random forest model achieved accurate wetland SOC estimation with R2 being .65. The results showed that average SOC density of top 1 m in China's wetlands was 25.03 ± 3.11 kg C m-2 in 2000 and 26.57 ± 3.73 kg C m-2 in 2020, an increase of 6.15%. SOC storage change from 4.73 ± 0.58 Pg in 2000 to 4.35 ± 0.61 Pg in 2020, a decrease of 8.03%, due to 13.6% decreased in wetland area from 189.12 × 103 to 162.8 × 103 km2 in 2020, despite the increase in SOC density during the same time period. The carbon accumulation rate was 107.5 ± 12.4 g C m-2 year-1 since 2000 in wetlands with no area changes. Climate change caused variations in wetland SOC density, and a future warming and drying climate would lead to decreases in wetland SOC storage. Estimates under Shared Socioeconomic Pathway 1-2.6 (low-carbon emissions) suggested that wetland SOC storage in China would not change significantly by 2100, but under Shared Socioeconomic Pathway 5-8.5 (high-carbon emissions), it would decrease significantly by approximately 5.77%. In this study, estimates of wetland SOC storage were optimized from three aspects, including sample database, wetland extent, and estimation method. Our study indicates the importance of using consistent SOC density and extent data in estimating and projecting wetland SOC storage.

Improving the generalization error and transparency of regression models to estimate soil organic carbon using soil reflectance data

Despite the success of using soil spectroscopy in studies to predict soil attributes, like soil organic carbon (SOC), recent work has revealed several limitations to this approach: a tendency for model overfitting and a lack of transparency of machine learning (ML) methods. Thus, we aimed to both test the ability to improve the generalizability of the models to predict SOC using a cross-validation (CV) strategy oriented to soil profiles and to test the gain in model interpretability by using the least absolute shrinkage and selection operator (LASSO) regression method instead of the commonly used partial least squares (PLS) method. We used one soil spectral library composed of 127 soil profiles (n = 701), from Northeast Brazil, containing reflectance data from the visible, near, and short-wave infrared (VNIR) and the mid-infrared (MIR) spectral regions. We tuned the ML models to predict SOC via two CV strategies: the standard k-fold CV and the leave-soil-profile-out (LSPO) CV. We found that LSPO CV can produce models with better generalizability, as they lose less accuracy than the ones trained with k-fold CV. We conclude that disregarding the autocorrelation of SOC within the soil profile can produce models that are prone to overfitting. In addition, LASSO used 105 covariables from VNIR and 190 from MIR for a total of 8604 and 13,336 covariables, respectively. Moreover, a few LASSO covariables correlated with SOC and are associated with both electronic transitions and vibrational bonds in organic compounds, so the possibility and ease of identifying spectral bands and their correlation with organic carbon indicate that the LASSO models presented more transparent models than the PLS models.