No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Modelling and mapping soil organic carbon stocks in Brazil
Abstract
Brazil has extensive forests and savannas on deep weathered soils and plays a key role in the discussions about carbon sequestration, but the distribution of soil organic carbon (SOC) stocks up to 1 m depth has not been investigated in Brazil using machine learning techniques. In this study, we applied a methodological framework to optimize the prediction of SOC stocks for the entire Brazilian territory and determine how the environmental heterogeneity of Brazil influences the SOC stocks distribution. We used a legacy dataset of 8227 soil profiles which consisted of 37,693 samples. For each profile, the vertical distribution of SOC and bulk density were interpolated to standard depths (0–5, 5–15, 15–30, 30–60 and 60–100 cm) using mass preserving equal-area quadratic splines. The covariates database was composed of 74 variables including bioclimatic (temperature and precipitation) data, soil and biome maps, vegetation indexes and morphometric maps derived from a digital elevation model, with a 1 km spatial resolution. To obtain the best prediction performance, we tested four machine learning algorithms: Random Forests, Cubist, Generalized Linear Model Boosting and Support Vector Machines. Random Forests showed the best performance in predicting SOC stocks for all depths, with the highest performance at 30–60 cm for training (R2 = 0.32) and validation (R2 = 0.33); hence, it was selected for the spatial prediction of SOC stocks. The most important covariates selected by Random Forests using the recursive feature elimination were: soil class, sum of monthly mean temperature (SAMT), precipitation, slope height and vegetation indexes (NDVI, GPP). In total, Brazilian soils store approximately 71.3 PgC within the top 100 cm, where the first 0–30 cm contains almost 36 PgC. Approximately 31% of the total SOC stocks (22.2 PgC) occurs in protected areas (2.6 million km2), which are not subjected to land use pressure and carbon losses. Although the Amazon biome has the highest amount of stored SOC (36.1 PgC), its soils do not represent a good potential for carbon accumulation. Among soil classes, the Luvisols showed the lowest SOC density (6.45 kg m−2) and the Histosols presented the highest values (14.87 kg m−2). More than 57% of the total SOC was found in nutrient-poor, deep-weathered Ferralsols and Acrisols, which are the dominant soils in Brazil. The presented methodological framework covers all steps of prediction process, building maps with known accuracy and has great potential to be used in future soil carbon inventories at large scales. Concerning conservation issues, the results highlight the importance of nature reserves for protecting SOC in the long-term.
... SOC modelling often employs machine learning techniques [4] based on several soil parameters and properties (textural data, N, pH, Ca 2+ , Mg 2+ , Al, Fe), and their covariates related to climate, organisms, topography, parent material, time and site [5,6]. However, the availability of these extensive datasets can often be limited or lacking in certain study sites, particularly in regions with limited financial resources and skilled labor for database construction [7], as commonly observed in tropical savannas of countries like Brazil. ...
... These soils displayed textural data values with large-scale variability, from 92 to 515 g•kg −1 , and high standard deviations ( Table 1). The dataset covers all the textural variation characteristics of Oxisols [4]. The layer 0-0.3 m had higher SOC values (1.2 g•kg −1 < SOC < 47.1 g•kg −1 ) than the layer 0.3-1.0 ...
... However, the models' performance depends on how the variables related to the input data could intervene into the triangle results. We recognize that there are more robust methods [4,39,40], involving other variables that exert influence on SOC and, therefore, resulting in a higher accuracy. Nevertheless, we believe that approach proposed in this work proved to be more than enough for areas where intensive soil survey and analytical effort are not capable of being accomplished. ...
There are many studies about soil organic carbon (SOC) around the world but, in extensive territories, it is more difficult to obtain data due to the number of variables involved in the models and their high cost. In large regions with poor infrastructure, low-cost SOC models are needed. With this in mind, our objective was to estimate the SOC using a simple model based on soil textural data. The work was focused on savanna soil and validated the model in the Brazilian Savanna. Two models were constructed, one for topsoil (0-0.3 m) and other for subsoil (0.3-1.0 m). The SOC models can be carried out in a textural triangle together with SOC values. The results showed that subsoil models were more accurate than topsoil models, but both had good performance. The models give support to SOC-related preliminary research in gross and fast estimates, requiring only reduced financial contribution to calculate SOC in a large region of interest.
... More accurate quantification of SOC stock can help to refine carbon inventories at regional and national scales, improve our understanding of its spatial distribution, and inform us about the level of uncertainties in national or global soil carbon maps (Lemercier et al., 2022). The spatial distribution of SOC stock can be performed from a local to global scale using digital soil mapping -DSM (McBratney et al., 2003), a technique that links soil sample information with explanatory variables (e.g., remote sensing variables) using machine learning algorithms (Chen et al., 2018;Chen et al., 2022;Li et al., 2021b;Gomes et al., 2019;Grundy et al., 2015;McBratney et al., 2003;Minasny and McBratney, 2016;Siqueira et al., 2023;Siqueira et al., 2024). ...
... The Amazon is one of the most extensive tropical forest areas in the world, which in addition to harboring a high diversity of species, also stores carbon in the soil and biomass, playing an important role in mitigating climate change (Barros Ferraz et al., 2005). Due to the operational difficulty of accessing the Amazon region, this biome contains one of the lowest densities of soil sampling points in Brazil, which generates great uncertainty in the prediction of SOC stock for this region (Gomes et al., 2019;Vasques et al., 2017). Quantifying and spatializing the SOC stock in these areas is essential for understanding carbon inputs and outputs from deforestation. ...
... Despite the lack of sampling points for the Amazon biome, the government of Rondônia carried out a detailed survey of the state's soils in the 1990s, with >2900 profiles distributed throughout the state's territory. This database is now available online but has not been used in current SOC stock projections at the national (Gomes et al., 2019;Vasques et al., 2017) and global (Hengl et al., 2017;Poggio et al., 2021) levels. Therefore, mapping the SOC stock at the regional level with a broader database has the potential to improve its quantification and distribution and better guide public policymakers in the context of fostering strategies for the conservation of SOC stock. ...
... Soil is a major carbon reservoir with great potential to store carbon twice the potential of the atmosphere or vegetation (Schlesinger 1997, Yang et al. 2010. The stocks of soil organic carbon (SOC) act as major carbon inventories in the environment that help in sequestering atmospheric CO 2 , acting as its prominent sink (Gomes et al. 2019) and contributing significantly to plummeting the effects of existing and impending climate change (Batjes 1998, IPCC 2014. The atmospheric CO 2 plays a pivotal part in sustaining the temperature of the Earth's surface globally (Dinakaran et al. 2014). ...
... Humid conditions or moisture content in soil favors the formation of SOC by increasing the activity of soil microbes, litter disintegration, and formation of SOM and also favors stabilizing of minerals as a result of the amplified breakdown of parent material (Chaplot et al. 2010, Doetterl et al. 2015. For example, the Amazon biome within Brazil region has a humid tropical climate with mean annual precipitation (MAP) higher than 3100 mm and mean annual temperature (MAT) ranging between 25.9 to 27.7°C stores 36.1 PgC carbon stock (Gomes et al. 2019). Whereas the semi-arid eastern region of Rajasthan in India has a dry climate with average 500-1000 mm annual rainfall and an estimated carbon stock of 2129.9 ...
... It is assumed that volcanic ash is responsible for such high carbon density of Japanese forest soil because it has higher efficiency in absorbing, retaining, and stabilizing organic matter. Another study done by Kyuma (1990) and Gomes et al. (2019) to estimate SOC stock in Brazilian soil revealed that Brazil is majorly covered by the Amazon biome (49.29%), having a humid tropical climate, with MAP ≥ 3100 mm and MAT of 25.9-27.7°C, Cerrado biome (22% of terrain) and Atlantic Forests. ...
Soil is one of the largest carbon reservoirs sequestering more carbon than vegetation and atmosphere. Due to the enormous potential of soil to sequester atmospheric CO2, it becomes a feasible option to alleviate the current and impending effects of changing climate. Soil is a vulnerable resource globally because it is highly susceptible to global environmental problems such as land degradation, biodiversity loss, and climate change. Therefore, protecting and monitoring worldwide soil carbon pools is a complicated challenge. Soil organic carbon (SOC) is a vital factor affecting soil health since it is a major component of SOM and contributes to food production. This review attempts to summarize the information on carbon sequestration, storage, and carbon pools in the major terrestrial ecosystems and underpin soil carbon responses under climate change and mitigation strategies. Topography, pedogenic, and climatic factors mainly affect carbon input and stabilization. Humid conditions and low temperature favor high soil organic carbon content. Whereas warmer and drier regions have low SOC stocks. Tropical peatlands and mangrove ecosystems have the highest SOC stock. The soil of drylands stores 95% of the global Soil Inorganic Carbon (SIC) stock. Grasslands include rangelands, shrublands, pasturelands, and croplands. They hold about 1/5th of the world’s total soil carbon stocks.
... For instance, to investigate the impact of temperature on the distribution of SOCS, various temperature scenarios can be tested and their effects on SOCS can be measured. This approach has effectively assessed the spatial variation of SOCS across diverse land use and climate change scenarios in Brazil, Australia, Chile, United States, and China (Strey et al., 2016;Adhikari et al., 2019;Gomes et al., 2019;Wang et al., 2022). However, despite its the validity of this method cannot be tested without verification data for future scenario and uncertainties may exist (Strey et al., 2016;Gray and Bishop, 2016;Reyes Rojas et al., 2018). ...
... In areas characterized by complex and dynamic terrain, topographic variables were found to be reliable predictors of SOCS (Yang et al., 2016). The terrain indirectly affects the soil though the material and energy redistribution, as demonstrated in studies conducted by Román-Sánchez et al., (2018) and Gomes et al. (2019). This study confirmed previous findings that ELE was the dominant variable affecting spatial variability in SOCS (Yang et al., 2016;Blackburn et al., 2022). ...
... The diminishing influence of environmental variables on SOCS spatial variability with soil depth can be attributed to inherent physical and chemical properties of the soil (Yang et al., 2016). SA indirectly affects the physical and chemical properties of soil in mountainous regions through its regulation of lighting conditions, precipitation, and temperature (Gomes et al., 2019). Rezaei and Gilkes (2005) conducted that the SOC varied with depths and slope direction, showing a significant increase in deep SOC on shady slopes. ...
... In the RFR model, the predicted value of an observation is calculated by averaging over all of the trees. This helps to avoid the limitations of a single decision tree and improves the performance by reducing the model variance [44,45]. In this study, 70% of the soil samples were utilized as the training set, and 30% of soil samples were utilized as the testing set. ...
... To ensure the optimal and stable model, five-fold cross validation was employed to evaluate the performance of the RFR models. The performance of the RFR models of the SBD and SOC predictions were evaluated using the coefficient of determination (R 2 , equation (2)), mean absolute error (MAE, equation (3)), root mean square error (RMSE, equation (4)), and BIAS (equation (5)) [40,45]. ...
SOC stock (SOCS) changes caused by land use changes are still unclear, and understanding this response is essential for many environmental policies and land management practices. In this study, we investigated the temporal-spatial and vertical distribution characteristics of SOCS on the Western Sichuan Plateau (WSP) using the sparrow search algorithm-random forest regression (SSA-RFR) models with excellent model applicability and accuracy. The temporal-spatial varia-tions in the SOCS were modeled using 1080 soil samples and a set of nine environmental covariates. We analyzed the effect of land use changes on the SOCS on the WSP. The total SOCS increased by 18.03 Tg C from 1990 to 2020. The results of this study confirmed that the SOCS in the study area has increased significantly since 2010, with an increase of 27.88 Tg C compared to the total SOCS in 2010. We found that the spatial distribution of the SOCS increased from southeast to northwest, and the vertical distribution of the SOCS in the study area decreased with increasing soil depth. Forest and grassland are the main sources of SOCS, the total SOCS in the forest and grassland accounted for 37.53 and 59.39% of the total SOC pool in 2020, respectively. Expansion of the wetlands, forest, and grassland areas could increase SOCS in the study area.
... To ensure the accuracy and reliability of our analysis, the redundant covariates were eliminated using the Recursive Feature Elimination (RFE, (Brungard et al., 2015;Gomes et al., 2019;Nussbaum et al., 2017;Poggio et al., 2021;Yang et al., 2022;Zhang et al., 2023)) algorithm during the covariate selection process, with the default hyperparameters. It is worth noting that RFE is also based on data not used for testing. ...
Digital soil mapping relies on statistical relationships between soil profile observations and environmental covariates at the sample locations. However, inherent limitations of legacy soil profiles, such as inaccurate georeferencing, could frequently introduce location errors into these soil profiles that affect the quality of digital soil mapping. To address this challenge, this study focuses on reducing the location error of legacy soil profiles and evaluating the resulting impact on digital soil mapping. We improved the agreement between the detailed descriptive information of legacy soil profiles and the relatively accurate environmental covariates (such as elevation, slope, and land use) to reduce the location errors of the legacy soil profiles. Quantile regression forest models were constructed to predict soil properties and their uncertainty using legacy soil profiles before and after location error correction. Our results demonstrate that for the majority of soil variables, correcting positional errors in legacy soil profiles to some extent enhances the accuracy of the digital soil mapping. The largest improvement was found for soil organic carbon at 0–5 cm depth interval, with 21 % increase of MEC. The impact of reduced location error is particularly noteworthy in regions characterized by complex terrain. In addition, the details of the predicted maps of legacy soil profiles were improved after correcting for positional errors, which better represent the spatial variation of soil properties across China. Besides, we also found that elevation was the primary controlling factor for correcting location error of legacy soil profiles. This research presents a step towards producing high-resolution and high-quality spatial soil datasets, which can provide essential support for soil management and ensure future soil security.
... In addition to management practices, CS levels and rates of soil SEQ vary depending on different factors, such as source material, pedogenetic processes, soil texture, amount of organic matter (OM) cycling and input, and climatic conditions, with higher CS generally being achieved in conditions of lower temperatures and higher rainfall (Jenny, 1941;Hengl et al., 2015;Gomes et al., 2019). At least 50 years of soil maintenance are required to achieve the maximum possible CS, but the rate of increase will not necessarily be constant throughout this period (Lal et al., 1998). ...
Since the industrial revolution, human activities have emitted approximately 2,500 Gt of CO2, increasing the concentration of atmospheric CO2 by 50 % compared to pre-industrial levels. To better understand the potential for mitigating greenhouse gas (GHG) emissions through proper management of degraded pasture areas, we conducted a systematic literature review and identified 23 publications reporting carbon sequestration values for pastures managed under different conditions in the south and southeast regions of Brazil. From this dataset, 17 publications considered to be in line with the research premises were selected to estimate the potential for soil carbon sequestration (SEQ) through pasture recovery in the southern region of Brazil, using conservative and regenerative agricultural management practices. Results show that managed pastures can sustain carbon sequestration rates of around 2.50 Mg C ha⁻¹ yr⁻¹ over approximately 20 years. However, due to the numerous variables influencing SEQ rates, the limited number of publications, and the lack of data for some variables among them, a more extensive analysis of publications and data is needed to establish causal and preponderance relationships regarding the effect of each variable on the found SEQ rates. Under current pasture occupation conditions in Brazil’s south region, it is estimated these areas could sequester between 0.433 and 1.273 Gt CO2 at the end of 20 years if managed under appropriate practices. These numbers are not representative to reduce atmospheric CO2 concentration from legacy emissions and significantly mitigate physical impacts of climate change, reinforcing the importance of prioritizing the reduction of global GHG emissions as the primary mitigation strategy. On the other hand, from the perspective of mitigating the national agricultural sector’s annual GHG emissions, this potential cannot be considered negligible. Carbon sequestration by soils under agricultural management can play a vital role in mitigating climate change, integrating the set of necessary solutions and actions for a Paris Agreement goals compatible trajectory of limiting global warming to between 1.5 and 2 °C by the end of the century.
Keywords
climate change; soil texture; degradation level; Atlantic Forest Biome; Pampas Biome
... Minasny et al., 2013;Taghizadeh-Mehrjardi et al., 2020; Žížala et al., 2021 Žížala et al., Grimm et al., 2008Hounkpatin et al., 2021;Nabiollahi et al., 2019 Rentschler et al., 2019Were et al., 2015Gomes et al., 2019 Hengl et al., 2015;Taghizadeh-Mehrjardi et al., 2016;Wiesmeier et al., 2011Adhikari et al., 2019Funes et al., 2019;Kučera et al., 2020; Ottoy et ب) و عمق تا خاک ذخیره آلی کربن مقادیر و مکانی توزیع نحوه 100 گرفت. قرار بررسی مورد متری سانتی اندازه مقادیر مکانی پراکنش . ...
Investigation of soil organic carbon stock (SOCS) in agricultural lands and the role of factors affecting its variability and digital modeling are important for predicting possible scenarios of future carbon stock. The purpose of this study was to investigate the spatial variability and to estimate SOCS at 0 to 100 cm depth based on two generation of machine learning approaches in a part of Qazvin plain. SOCS of about 211 legacy soil data were prepared. The environmental variables including 11 geomorphometric variables and 25 spectral indices with 10-meter spatial resolution were used. Further, the dataset was divided into two parts: 70% of data were chosen as training and 30% of data for model validation. Two algorithm were used for SOCS modeling in the study area. Validation results indicated that the QRF had a higher coefficient of determination than the RF. According to the results of the relative importance of environmental variables, DEM and Valley depth parameters are more important in the spatial modeling of SOCS than other variables. Generally, it is suggested to investigate hybrid models in the process of modeling secondary soil characteristics.
... The SSA incorporates randomness and local search mechanisms to enhance its ability to explore the solution space and improve the convergence of the algorithm [43]. The performance of the RFR models of the SBD and SOC predictions was evaluated using the coefficient of determination (R 2 , Equation (2)), mean absolute error (MAE, Equation (3)), root mean square error (RMSE, Equation (4)), and BIAS (Equation (5)) [44,45]. Both RFR and the SSA are implemented in Python 3.8. ...
Soil organic carbon stock (SOCS) changes caused by land use changes are still unclear, and understanding this response is essential for many environmental policies and land management practices. In this study, we investigated the temporal–spatial and vertical distribution characteristics of the SOCS in the Western Sichuan Plateau (WSP) using the sparrow search algorithm–random forest regression (SSA-RFR) models with excellent model applicability and accuracy. The temporal–spatial variations in the SOCS were modeled using 1080 soil samples and a set of nine environmental covariates. We analyzed the effect of land use changes on the SOCS in the WSP. The total SOCS increased by 18.03 Tg C from 1990 to 2020. The results of this study confirmed a significant increase in the SOCS in the study area since 2010. There was a 27.88 Tg C increase in the SOCS in 2020 compared to the total SOCS in 2010. We found that the spatial distribution of the SOCS increased from southeast to northwest, and the vertical distribution of the SOCS in the study area decreased with increasing soil depth. Forests and grasslands are the main sources of SOCS the total SOCS in the forest and grassland accounted for 37.53 and 59.39% of the total soil organic carbon (SOC) pool in 2020, respectively. The expansion of the wetlands, forest, and grassland areas could increase the SOCS in the study area. A timely and accurate understanding of the dynamics of SOC is crucial for developing effective land management strategies to enhance carbon sequestration and mitigate land degradation.
... Different machine learning algorithms produce different levels of accuracy with the same dataset. Previous research on Moso bamboo forest biomass and carbon storage has indicated that models constructed using the XGBoost algorithm are more precise Gomes et al., 2019). Concurrently, selective harvesting of Moso bamboo forests serves as a crucial strategy to maintain productivity and production levels, leading to enhanced carbon sequestration and the preservation of aboveground productivity (Kuehl et al., 2013;Zheng et al., 2022). ...
Estimating the carbon sequestration potential of Moso bamboo ( Phyllostachys pubescens ) forests and optimizing management strategies play pivotal roles in enhancing quality and promoting sustainable development. However, there is a lack of methods to simulate changes in carbon sequestration capacity in Moso bamboo forests and to screen and optimize the best management measures based on long-term time series data from fixed-sample fine surveys. Therefore, this study utilized continuous survey data and climate data from fixed sample plots in Zhejiang Province spanning from 2004 to 2019. By comparing four different algorithms, namely random forest, support vector machine, XGBoost, and BP neural network, to construct aboveground carbon stock models for Moso bamboo forests. The ultimate goal was to identify the optimal algorithmic model. Additionally, the key driving parameters for future carbon stocks were considered and future aboveground carbon stocks were predicted in Moso bamboo forests. Then formulated an optimal management strategy based on these predictions. The results indicated that the carbon stock model constructed using the XGBoost algorithm, with an R ² of 0.9895 and root mean square error of 0.1059, achieved the best performance and was considered the optimal algorithmic model. The most influential driving parameters for vegetation carbon stocks in Moso bamboo forests were found to be mean age, mean diameter at breast height, and mean culm density. Under optimal management measures, which involve no harvesting of 1–3 du bamboo, 30% harvesting of 4 du bamboo, and 80% harvesting of bamboo aged 5 du and above. Our predictions show that aboveground carbon stocks in Moso bamboo forests in Zhejiang Province will peak at 36.25 ± 8.47 Tg C in 2046 and remain stable from 2046 to 2060. Conversely, degradation is detrimental to the long-term maintenance of carbon sequestration capacity in Moso bamboo forests, resulting in a peak aboveground carbon stock of 29.50 ± 7.49 Tg C in 2033, followed by a continuous decline. This study underscores the significant influence of estimating carbon sequestration potential and optimizing management decisions on enhancing and sustaining the carbon sequestration capacity of Moso bamboo forests.
... First, we excluded highly correlated variables using a correlation coefficient limit of ±0.9 Silva et al., 2016). A method based on Recursive Feature Elimination (RFE) (Gomes et al., 2019) was subsequently implemented considering only the variables not excluded in the first step to select the best subset of variables. This method is a reverse selection algorithm that calculates the importance of the resource in each iteration, classifying them from most important to least important, removing a user-defined subset at each stage Johnson, 2013a, 2013b). ...
... Annual precipitation data was obtained through WorldClim 2.1 (Fick and Hijmans, 2017) to represent climate as a soil-forming factor. Soil organic carbon stock was obtained from Gomes et al. (2019) due to the influence of organisms for soil characteristics. ...
Most Brazilian municipalities do not have regulated areas for solid waste disposal in civil construction. Usually, residues are disposed of vacant lots and dumps, posing risks to the population health and the environment. Soils are the primary means for the disposal or recycling of waste, highlighting the importance of well-characterized soils and their respective landscape. This study aimed to establish a land suitability system for solid residues in civil construction and apply such information in a case study in Southeastern Brazil. An unprecedented digital soil map with a resolution of 30 m was created using the random forest classifier algorithm and soil field prospection information. A guide listing favorable soil-landscape attributes that most prevent soil erosion, water bodies or water table contamination was elaborated and discussed. Thus, such information was linked through a suitability system to classify areas with potential for receiving waste on a daily volume basis as follows: large size: >500 m³ day⁻¹, medium size: >100 m³ day⁻¹ and <300 m³ day⁻¹, and small size <100 m³ day⁻¹. Topography and soil depth were the most limiting factors of the areas in the case study. The proposed attributes as criteria for the suitability system complement the current state legislation. A total of 236 ha closer to the urban perimeter connected by roads in good condition were classified as suitable for managing medium- and small-scale daily volume, whose destination might reduce transportation and installation costs in the study area.
Keywords
land-use planning; soil survey; random forest; environmental legislation
... Akpa et al. (2016) found that the RF model performed better when compared to cubist and BRT in prediction of SOC stock. Gomes et al. (2019) also reported better predictive performance of RF model by comparison with cubist, generalized linear model boosting and SVM in prediction of SOC stock. Uncertainty can be indicated by an indicator called PICP which indicates % of all observed values fitting within their prediction limits. ...
Usefulness of nitrification inhibitors (NIs) has been identified in reducing the soil
nitrogen losses of applied N fertilizers. We conducted an incubation experiment to
evaluate the effects of soil applied NIs on nitrogen transformations (NH & NO ) and
NI in sandy clay loam soil. Powdered extracts of pomegranate rind, melia fruit, neem
cake were prepared and added in soil at concentration of 20 and 40 g kg-1 soil, respec-
tively, while calcium carbide was added at 30 g kg-1 soil. Sources of N, P and K were
urea, single super phosphate (SSP), murate of potash (MOP) at the rate of 1.98, 3.50
and 0.88 g-1pot, respectively. Sole application of urea was taken as the control
treatment. Treated soils were incubated at 25oC for 42 days. Results revealed that at the
end of incubation, highest NH +-N retention (126.30 mg N kg-1) and total soil nitrogen 4
(TSN) (152.72 mg kg-1) was recorded under CaC @ 30 g kg-1 soil treatment. Maximum 2
NO --N accumulation (42.26 mg N kg-1) was associated with melia fruit @ 20 g kg-1 soil 3
treatment. Regarding the nitrification inhibition, treatment of neem cake 40 g kg-1 soil
recorded maximum nitrification inhibition (44.31 %). Amongst the different nitrification
inhibitors tested, lowest levels of nitrified N (20.41 and 22.05%) were recorded
under the application of CaC @30 g kg-1 soil and neem cake @40 g kg-1, while the 2
maximum (54.5%) was observed in treatment comprising of urea alone (control). Ecofriendly
and cost-effective plant based nitification inhibitors were identified superior
as compared to synthetic on nitrification inhibition.
... Although DSM methodologies [8][9][10][11] have been employed to map SOC stocks on large scales, there is still a gap at the hyper-local level of high spatial resolution which needs to be addressed. This is particularly important for improving the accuracies in the ground-truth stages with field observations [12]. ...
Mapping soil organic carbon (SOC) stock can serve as a resilience indicator for climate change. As part of the carbon dioxide (CO2) sink, soil has recently become an integral part of the global carbon agenda to mitigate climate change. We used hyperspectral remote sensing to model the SOC stock in the Sele River plain located in the Campania region in southern Italy. To this end, a soil spectral library (SSL) for the Campania region was combined with an aerial hyperspectral image acquired with the AVIRIS–NG sensor mounted on a Twin Otter aircraft at an altitude of 1433 m. The products of this study were four raster layers with a high spatial resolution (1 m), representing the SOC stocks and three other related soil attributes: SOC content, clay content, and bulk density (BD). We found that the clay minerals’ spectral absorption at 2200 nm has a significant impact on predicting the examined soil attributes. The predictions were performed by using AVIRIS–NG sensor data over a selected plot and generating a quantitative map which was validated with in situ observations showing high accuracies in the ground-truth stage (OC stocks [RPIQ = 2.19, R2 = 0.72, RMSE = 0.07]; OC content [RPIQ = 2.27, R2 = 0.80, RMSE = 1.78]; clay content [RPIQ = 1.6 R2 = 0.89, RMSE = 25.42]; bulk density [RPIQ = 1.97, R2 = 0.84, RMSE = 0.08]). The results demonstrated the potential of combining SSLs with remote sensing data of high spectral/spatial resolution to estimate soil attributes, including SOC stocks.
... PhytOC stability in wheat straw has never been comprehensively estimated on a large spatial scale. Recently, machine learning has emerged as a promising tool in prediction research since it can provide profound insights into multi-dimensional and non-linear information on data (Cheng et al., 2023;Lucas et al., 2019). This prediction is of great interest to researchers, as it can provide a novel perspective on the influence of environmental factors on carbon stability and storage (Cheng et al., 2023;Yang et al., 2022). ...
... Instead, we employed all measured soil characteristics as potential indicators to elucidate the ecological restoration process at both soil depths. We considered that each soil characteristic exposes a specific ecological and bio-physical-chemical processes, including carbon sequestration, nutrient cycling, soil fertility, and microbial activities (Bandyopadhyay and Maiti, 2021;Barliza et al., 2019;Č ížková et al., 2018;Gomes et al., 2019;Hou et al., 2019;Uzarowicz and Skiba, 2011), which are related to soil quality and functionality. Several studies have used PCA to retain MDS (Bandyopadhyay and Maiti, 2021;Biswas et al., 2017;Mukhopadhyay et al., 2014Mukhopadhyay et al., , 2016Nabiollahi et al., 2018;Yu et al., 2018aYu et al., , 2018b. ...
The soil quality index (SQI) serves as a general ecological restoration indicator, however, statistics approaches that accurately assess the minimum data set (MDS) for SQI remain susceptible. The present study aims to evaluate the short-term reclamation results at the Ferro-Carvão stream and propose a system for ecological restoration monitoring, by selecting influential attributes and indexing soil quality. We hypothesized that the reclamation activities at the Ferro-Carvão stream, referred to as the "Marco zero" (MZ) area, can bring its soil quality to levels comparable to those of the native area. We collected soil samples at 0-20 and 20-40 cm depths from transects of MZ and reference sites (R1 and R2). Principal component analysis showed the MDS for each soil depth. Permutational analysis of variance, in conjunction with Nonmetric Multidimensional Scaling, exposed relationships between transects of areas. An additive non-linear factorial algorithm allowed SQI assessment. The results indicated a similar soil quality between transects of areas at 0-20 cm depth, whereas a dissimilarity at 20-40 cm. To sum up, reclamation activities allowed MZ-constructed Technosol to present a soil quality similar to native areas. The soil quality assessment at both depths offered insights into reclamation activities' immediate and long-term impacts on the Ferro-Carvão stream. This robust framework effectively monitors ecological restoration progress and guides future efforts in post-mining and post-dam collapse sites.
... We used Eq. (1) to calculate the per unit area SOC stock (kg m -2 ) (Gomes et al., 2019;Guo et al., 2019;Han et al., 2018;Mayer et al., 2019;Wiesmeier et al., 2012) for each sampled soil sample at the relative sampled depth, and Eq. (2) was used to calculate the total SOC stocks (Tg C) for the whole study area. ...
Accurate estimates of soil organic carbon (SOC) stocks are important in understanding terrestrial carbon cycling. Based on the fundamental theorem of surfaces, an alternative method, high accuracy surface modelling (HASM) combined with soil depth information was applied to predict the spatial pattern of SOC stocks in Hebei Province, China. In this study, we collected 434 soil samples and key environmental covariates related to soil-forming factors (soil, climate, organisms, topography, and soil depth information) in the study area, and compared the accuracy of 16 spatial prediction models (including single models, hybrid models, and HASM combined with single or hybrid models) on the spatial distribution of SOC stocks. The results confirmed that the method of HASM combined with the generalized additive model (GAM) with soil depth covariate (HASM_GAMD) achieved a better performance than other methods at soil depths of 0-30, 0-100 and 0-200 cm. The root-mean-square error and coefficient of determination values of predicting the spatial pattern of SOC stocks by the HASM_-GAMD model demonstrated a 43% and 49% improvement, respectively, compared with models without depth information. The prediction uncertainty of the HASM_GAMD model based on 90% prediction interval was lower than that of other models. The HASM_GAMD model excels in addressing not only the nonlinear relationship between covariates and SOC stocks, but also in incorporating point observation data that varies with soil depth. Furthermore, the model conducts modelling by integrating surface and optimal control theories. Results obtained from the HASM_GAMD demonstrated that the SOC stocks in Hebei Province amounted to 1449.08 Tg C. Our study introduces an alternative model for modelling of SOC stocks and our findings are a valuable reference for assessing carbon stocks in Hebei Province to support sustainable land management and climate change mitigation. Abbreviations: HASM, high accuracy surface modelling; RK, regression kriging; GAM, generalized additive model without soil depth covariate; RF, random forest without soil depth covariate; GAMD, generalized additive model with depth covariate; RFD, random forest with depth covariate; GAM_RK (RF_RK, GAMD_RK and RFD_RK), regression kriging with the trend prediction by GAM (RF, GAMD and RFD); HASM_GAM (HASM_RF, HASM_GAMD and HASM_RFD), HASM combined with GAM (RF, GAMD and RFD); HASM_GAM_R (HASM_RF_R, HASM_GAMD_R and HASM_RFD_R), GAM (RF, GAMD and RFD) predicted the trends of the SOC stock and HASM replaced ordinary kriging in the hybrid models to predict the residual surfaces; ME, mean error; RMSE, root-mean-square error; R 2 , coefficient of determination ; LCCC, Lin's concordance correlation coefficient.
... It was also confirmed in many other regions of the word. Luvisols showed the lowest SOC stocks in the top 100 cm (6.45 kg m −2 ) among main Reference Soil Groups occurred in Brasil -even lower than highly weathered tropical soils as Plinthosols Ferralsols or Acrisols (Gomes et al., 2019). Jarmain et al. (2023) also confirm that Luvisols have one of the lowest carbon retention capacities compared with other soils. ...
... Instead, we employed all measured soil characteristics as potential indicators to elucidate the ecological restoration process at both soil depths. We considered that each soil characteristic exposes a specific ecological and bio-physical-chemical processes, including carbon sequestration, nutrient cycling, soil fertility, and microbial activities (Bandyopadhyay and Maiti, 2021;Barliza et al., 2019;Č ížková et al., 2018;Gomes et al., 2019;Hou et al., 2019;Uzarowicz and Skiba, 2011), which are related to soil quality and functionality. Several studies have used PCA to retain MDS (Bandyopadhyay and Maiti, 2021;Biswas et al., 2017;Mukhopadhyay et al., 2014Mukhopadhyay et al., , 2016Nabiollahi et al., 2018;Yu et al., 2018aYu et al., , 2018b. ...
The soil quality index (SQI) serves as a general ecological restoration indicator, however, statistics approaches that accurately assess the minimum data set (MDS) for SQI remain susceptible. The present study aims to evaluate the short-term reclamation results at the Ferro-Carvão stream and propose a system for ecological restoration monitoring, by selecting influential attributes and indexing soil quality. We hypothesized that the reclamation activities at the Ferro-Carvão stream, referred to as the “Marco zero” (MZ) area, can bring its soil quality to levels comparable to those of the native area. We collected soil samples at 0–20 and 20–40 cm depths from transects of MZ and reference sites (R1 and R2). Principal component analysis showed the MDS for each soil depth. Permutational analysis of variance, in conjunction with Nonmetric Multidimensional Scaling, exposed relationships between transects of areas. An additive non-linear factorial algorithm allowed SQI assessment. The results indicated a similar soil quality between transects of areas at 0–20 cm depth, whereas a dissimilarity at 20–40 cm. To sum up, reclamation activities allowed MZ-constructed Technosol to present a soil quality similar to native areas. The soil quality assessment at both depths offered insights into reclamation activities' immediate and long-term impacts on the Ferro-Carvão stream. This robust framework effectively monitors ecological restoration progress and guides future efforts in post-mining and post-dam collapse sites.
... Effective programs that incentivize producers to adopt strategies that increase SOC stocks require cost-effective methods for monitoring and quantifying changes in SOC stocks. Recent advances in soil mapping that incorporate remote sensing data with point measurements [2][3][4][5][6][7] suggest that there is potential to make cost-effective SOC stock mapping feasible. The current methodologies for assessing SOC stocks for carbon credits lack detail on required sampling amounts and designs beyond recommending landscape stratification for sampling designs for stock assessments [8]. ...
Soil organic carbon (SOC) sequestration assessment requires accurate and effective tools for measuring baseline SOC stocks. An emerging technique for estimating baseline SOC stocks is predictive soil mapping (PSM). A key challenge for PSM is determining sampling density requirements, specifically, determining the economically optimal number of samples for predictive soil mapping for SOC stocks. In an attempt to answer this question, data were used from 3861 soil organic carbon samples collected as part of routine agronomic soil testing from a 4702 ha farming operation in Saskatchewan, Canada. A predictive soil map was built using all the soil data to calculate the total carbon stock for the entire study area. The dataset was then subset using conditioned Latin hypercube sampling (cLHS), both conventional and stratified by slope position, to determine the total carbon stocks with the following sampling densities (points per ha): 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, and 0.8. A nonlinear error function was then fit to the data, and the optimal number of samples was determined based on the number of samples that minimized soil data costs and the value of the soil carbon stock prediction error. The stratified cLHS required fewer samples to achieve the same level of accuracy compared to conventional cLHS, and the optimal number of samples was more sensitive to carbon price than sampling costs. Overall, the optimal sampling density ranged from 0.025 to 0.075 samples per hectare.
... Initially, the RFE calculates the importance of the set of variables in the complete model. In this case, the variables used were those not excluded in the first step (Gomes et al. 2019). The algorithm then removes the least important predictor and readjusts the model. ...
The Atlantic Forest fragments have suffered from the impacts of climate change, resulting in the increased production of coarse woody debris (CWD), which needs to be evaluated in space and time to generate accurate estimates of carbon accumulation. Thus, the goals of this study were (i) to quantify the CWD volume, necromass, carbon stock, and annual increment of carbon (AIcarb) over a period of 4 years; and (ii) to select the optimal combination of climatic, topographic, edaphic, and intrinsic forest variables to accurately predict AIcarb using machine learning and multivariate analysis. The CWD volume, necromass, and carbon stock increased between 2017 and 2020. The AIcarb was 1.09 MgC ha⁻¹ year⁻¹ (2017–2018), 1.24 MgC ha⁻¹ year⁻¹ (2018–2019), and 2.31 MgC ha⁻¹ year⁻¹ (2019–2020). Statistical analysis indicated that climate variables had greater weight in the CWD carbon increment in the 2018–2019 and 2019–2020 periods, while edaphic, topographical, and intrinsic forest variables were more important for the 2017–2018 period. Our findings showed that the carbon increase in CWD was linked to temporal and spatial variables within forests. These results demonstrate the importance of this parameter in the carbon cycle of forest ecosystems and highlight that there should be greater international research efforts to quantify this carbon pool.
... Land-use change can lead to disruption of soil aggregates, expose soil organic matter (SOM) to microbial decomposition, reduce nutrient stock in biomass, and aggravate emission of carbon dioxide (Najera et al., 2020). In the Caatinga biome, stock of soil carbon and quality of SOM have been prone to accelerated soil erosional processes resulting from anthropic and climate actions more than those under Cerrado or Atlantic Forest biomes (Gomes et al., 2019;Gonçalves & Vital, 2019). Although some farmers intercrop legumes and grasses with no-tillage systems (NTS) to minimize risks of soil degradation, unsustainable management practices and excessive crop fertilization are still widely practiced (Fracetto et al., 2012). ...
Land-use change has driven soil carbon stock losses in ecosystems worldwide. Implementing agricultural crops and exploiting forest resources trigger the breakdown of soil aggregates, thus exposing organic matter to microbial decomposition and enhancing carbon dioxide emissions, especially in biomes more susceptible to climate extremes as in the tropical semiarid regions. This study was based on the hypothesis that the undisturbed soil from the dry forest (Caatinga biome under natural revegeta-tion in Brazilian semiarid) would have an improvement in the mass of macroaggre-gates and recover more than 50% of the soil C stock within 10 years. Thus, a field experiment was conducted to investigate soils from the Caatinga biome under native vegetation, "cowpea cropping" for over 30 years, and soil under natural revegetation for over 10 years, after conventional soil cultivation of maize and cowpea, to determine soil and soil-aggregates carbon stocks and to estimate the recovery rate of these stocks. The proportional mass of aggregates of different sizes and the total stock of particulate organic carbon (POC) were also quantified. The results showed that soil under preserved native vegetation of dry forest Caatinga biome had higher total soil C stock (50.9 Mg ha À1) than that under cowpea cropping (23.2 Mg ha À1) and natural revegetation (45.1 Mg ha À1). The proportional mass of large macroaggre-gates was higher in soil under native vegetation for all depths. However, soil under cowpea cropping had lower C stocks in macroaggregates, and recovered roughly 63% of the original C stocks, while revegetation recovered 78% of the stock in 10 years. Although the conventional management system for cowpea monoculture aggravated losses in soil carbon stock by more than 50% of the original C stocks, dry forest under natural revegetation recovered 79% of this stock and almost 100% of POC stock in 10 years ($12 Mg ha À1). Furthermore, soil under undisturbed Caatinga dry forest achieved C stock levels equivalent to that of the global average range for semiarid tropical environments. The high recovery rate of C stock in forest soil under natural revegetation indicates the resilience potential of organisms responsible for structural protection of aggregates and the encapsulated soil organic matter content.
... Their findings indicated that the RF model demonstrated superior accuracy in SOCS modeling. In addition, in Brazil, Gomes et al. (2019) also predicted the horizontal and vertical spatial distribution of SOCS using bioclimatic data, soil and biome maps, vegetation indices, and morphometric maps, suggesting that random forests emerged as the top-performing machine learning model. Meanwhile, in Iran, Emadi et al. (2020) reported that the deep neural network (DNN) model outperformed other algorithms, exhibiting the lowest prediction error and uncertainty in predicting SOCS. ...
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.
Graphical Abstract
... However, the impact of precipitation on SOC and TN is debatable. Currently, it has been observed that an increase in precipitation is beneficial for increasing or has no significant effect on SOC (Gomes et al. 2019;McDonough et al. 2020;Rocci et al. 2021). Similar results have been obtained regarding the impact of precipitation on TN (Jongen et al. 2013;Wang et al. 2014), or that water even inhibited TN accumulation in certain tall grass prairies (Luo et al. 2004). ...
Purpose
Soil aggregates regulate soil water and temperature, soil fertilizer, and leaf gas exchange. In desert steppes, precipitation restricts the growth and development of plants, and it affects the availability of soil carbon and nitrogen, thereby influencing soil aggregate stability. However, studies on precipitation influence on the stability of aggregates are limited.
Materials and methods
Here, we conducted a 2-year field experiment in a desert steppe of Siziwang Banner, Inner Mongolia, to test the effect of a changing precipitation gradient “reducing precipitation by 50% (W-50%), natural precipitation (W), increasing precipitation by 50% (W+50%), and increasing precipitation by 100% (W+100%)” on the depth distribution, stability of soil aggregates and aggregate-associated organic carbon content (OC), and total nitrogen (TN) contents. We used a wet sieving method yielding silt and clay (SCA, < 0.053 mm), microaggregates (MIA, 0.053–0.25 mm), small macroaggregates (SMA, 0.25–2 mm), and large macroaggregates (LMA, > 2 mm).
Results and discussion
Our results indicated that the topsoil (0–30 cm) was dominated by SCA and MIA. Increasing precipitation increased soil aggregate stability and reduced soil erodibility by increasing water-stable aggregates (WSA, > 0.25 mm). In this study, the comprehensive soil aggregate stability score was the highest at W+100%. Although LMA serve as the main carriers of SOC and TN, MIA-associated OC and TN had the highest contribution rate to SOC and TN. This study revealed that bulk soil properties including MBN, BD, MBC, and pH significantly influenced aggregate stability. Additionally, WSA-associated OC were found to be the most crucial contributors to soil aggregate stability.
Conclusions
Overall, our study indicates that increasing precipitation is beneficial to WSA accumulation and highlighted the vital role of microbial biomass and WSA-associated OC on maintaining soil aggregate stability under precipitation change.
... Additionally, various studies including those by Khanal et al. (2018), Žížala et al. (2019), andZeraatpisheh et al. (2020) have acknowledged the substantial contributions of digital elevation models (DEMs) and their derivatives in this field. The establishment of statistical links has been also made feasible through a variety of methods, ranging from machine learning algorithms like random forest (Laamrani et al., 2019) and support vector machine (Gomes et al., 2019) to traditional techniques like multiple linear regression (MLR) (Forkuor et al., 2017;Zeraatpisheh et al., 2019). The diversity in predicting variables and models underscores the region-specific nature of SQ assessment and predictions, highlighting the need for a deep understanding of the unique conditions in each region. ...
Soil quality (SQ) modeling and mapping is a leading research field aiming to provide reproducible and cost-effective yet accurate SQ predictions at the landscape level. This endeavor was conducted in a complex watershed in northern Iran. We classified the region into spectrally and topographically homogenous land units (average area of 48 ± 23 ha) using object-based segmentation analysis. Following the physicochemical analysis of soil samples from 98 stations, the Nemoro soil quality index (SQIn) was produced using the minimum dataset procedure and a non-linear sigmoid scoring function. SQIn values averaged 0.21 ± 0.06 and differed statistically between major land uses. To predict and map SQIn for each land unit, the best-performing regression model (F(3, 84) = 45.57, p = 0.00, R² = 0.62) was built based on the positive contribution of the mean Landsat 8-OLI band 5, and negative influence of land surface temperature retrieved from Landsat 8-OLI band 10 and surface slope (t-test p-values < 0.01). Results showed that dense-canopy woodlands located in low-slope land units exhibit higher SQIn while regions characterized by either low-vegetation or steep-sloped land units had SQ deficits. This study provides insights into SQ prediction and mapping across spatially complex large-scale landscapes.
... At the early stage, most studies firstly estimate average carbon density based on soil profiles and then obtain carbon content using the area ratio of different soil types and vegetation types [6,7]. However, due to the high variability of soil carbon models used in DSM include multi-linear regression (MLR), random forest (RF), support vector machine (SVM), boosted regression tree (BRT), extreme learning machine (ELM) and artificial neural network (ANN) [34][35][36][37][38]. Previous studies have suggested that ML can achieve better results compared with traditional methods for SOC prediction [39,40]. ...
The accurate mapping of soil organic carbon (SOC) distribution is important for carbon sequestration and land management strategies, contributing to mitigating climate change and ensuring agricultural productivity. The Heihe River Basin in China is an important region that has immense potential for SOC storage. Phenological variables are effective indicators of vegetation growth, and hence are closely related to SOC. However, few studies have incorporated phenological variables in SOC prediction, especially in alpine areas such as the Heihe River Basin. This study used random forest (RF) and extreme gradient boosting (XGBoost) to study the effects of phenological variables (e.g., Greenup, Dormancy, etc.) obtained from MODIS (i.e., Moderate Resolution Imaging Spectroradiometer) product (MCD12Q2) on SOC content prediction in the middle and upper reaches of Heihe River Basin. The current study also identified the dominating variables in SOC prediction and compared model performance using a cross validation procedure. The results indicate that: (1) when phenological variables were considered, the R2 (coefficient of determination) of RF and XGBoost were 0.68 and 0.56, respectively, and RF consistently outperforms XGBoost in various cross validation experiments; (2) the environmental variables MAT, MAP, DEM and NDVI play the most important roles in SOC prediction; (3) the phenological variables can account for 32–39% of the spatial variability of SOC in both the RF and XGBoost models, and hence were the most important factor among the five categories of predictive variables. This study proved that the introduction of phenological variables can significantly improve the performance of SOC prediction. They should be used as indispensable variables for accurately modeling SOC in related studies.
... These factors, which define soilforming process, are climate, vegetation, human activities, terrain, parent material, and time. Numerous researchers logically used different environmental covariates to represent the SCORPAN model of soil spatial variability (Rossel and Chen 2011;Szatmári et al. 2016;Wang et al. 2018;Gomes et al. 2019). The most common covariates are existing soil property or legacy maps, geological or lithological maps, long-term mean annual rainfall, temperature and land surface temperature, remote sensing-derived satellite products (e.g., NDVI, EVI, NDWI), satellite imageries (e.g., MODIS, Landsat, and SPOT), and terrain parameters (e.g., elevation, slope, curvature, topographic wetness index). ...
Soil carbon (C) and its quality has always been a matter of global importance in a changing climate scenario and plays a crucial role in maintaining soil health, crop productivity and in mitigating climate change through C sequestration. However, the huge heterogeneity in climate, topography, land use, and management practices gives rise to wide variations in the soil C level. These variations in C across soils are best depicted by the spatial variability mapping through interpolation/extrapolation techniques, whose accuracy in turn depends on the intensity of sampling and correctness of the analysis. The intensive sampling and chemical analysis of soil samples are highly expensive and time-consuming; therefore, it is not possible to monitor soil C at large spatial coverage with limited resources in a short period of time. In recent years, digital mapping of soil C has gained importance in different parts of the world due to its cost-effectiveness as well as time-saving technique. Basically, digital mapping includes collection of preexisting/legacy observed dataset of the soil C for the area of interest and the relevant covariates, followed by calibration and developing a prediction function using the covariates, after which the prediction function is used for interpolation and/or extrapolation over the entire area of interest. Finally, validation of the calculated/derived dataset is done by comparing them with the observed dataset for a particular location. This chapter discusses the use of digital soil mapping for soil C, its techniques, previous works done, and its applications in relevant fields.
... In addition, due to continuous improvement in human living standards and accelerated demand for materials, policymakers continue to optimize the urban structure, which indirectly leads to the decline in ecosystem carbon storage. There are also quite a few studies focusing on the combination of carbon storage and predictive land-use change models, which simulate future land-use trends by setting future scenarios but do not carry out a more detailed analysis of its influencing factors [25,26,66,67]. In an attempt to compensate for this deficiency, this study focuses on analyzing the intrinsic influences on carbon storage. ...
Terrestrial carbon storage plays a vital role in limiting global climate change and achieving regional carbon neutrality. However, intensive human activities and rapid urbanization have led to a rapid decline in carbon storage. Understanding what causes carbon storage to decline and how this happens is important for the scientific regulation of urbanization and safeguarding of urban ecological security. This study takes Wuhan as an example and analyzes the quantity, structure, and spatial patterns of urban land-use changes in the context of human activities and natural conditions, and applies correlation methods to identify general relationships between influencing factors and carbon storage. The results of the study are as follows: over the 30-year period studied, the area devoted to construction land increased by 757 km2 and the carbon storage decreased by 7.68 × 106 t. Outside Wuhan’s Third Ring Road, there was a significant increase in the carbon storage, but in the areas where construction increased, there was a reduction in carbon storage. Carbon storage in the remote suburbs was significantly higher than in the city center, and the distribution pattern was characterized by significant spatial heterogeneity. Our analysis revealed that human economic and social activities have affected Wuhan’s ecosystem carbon storage to a significant extent. Policymakers should focus on industrial optimization, strictly control the red line of ecological protection, and ultimately achieve high-quality urban development.
Mapping soil properties in sub-watersheds is critical for agricultural productivity, land management, and ecological security. Machine learning has been widely applied to digital soil mapping due to a rapidly increasing number of environmental covariates. However, the inclusion of many environmental covariates in machine learning models leads to the problem of multicollinearity, with poorly understood consequences for prediction performance. Here, we explored the effects of variable selection on the prediction performance of two machine learning models for multiple soil properties in the Haihun River sub-watershed, Jiangxi Province, China. Surface soils (0–20 cm) were collected from a total of 180 sample points in 2022. The optimal covariates were selected from 40 environmental covariates using a recursive feature elimination algorithm. Compared to all-variable models, the random forest (RF) and extreme gradient boosting (XGBoost) models with variable selection improved in prediction accuracy. The R2 values of the RF and XGBoost models increased by 0.34 and 0.47 for the soil organic carbon, by 0.67 and 0.62 for the total phosphorus, and by 0.43 and 0.62 for the available phosphorus, respectively. The models with variable selection presented reduced global uncertainty, and the overall uncertainty of the RF model was lower than that of the XGBoost model. The soil properties showed high spatial heterogeneity based on the models with variable selection. Remote sensing covariates (particularly principal component 2) were the major factors controlling the distribution of the soil organic carbon. Human activity covariates (mainly land use) and organism covariates (mainly potential evapotranspiration) played a predominant role in driving the distribution of the soil total and soil available phosphorus, respectively. This study indicates the importance of variable selection for predicting multiple soil properties and mapping their spatial distribution in sub-watersheds.
Phosphorus (P) is a critical nutrient for primary production in terrestrial and aquatic ecosystems. As P mineral reserves are finite and non-renewable, there is an increasing discussion on its sustainable utilization to safeguard food security for future generations. Understanding the spatial distribution of soil P is central in advancing effective phosphorus management and fostering sustainable agricultural practices. This study aims to digitally map the stocks of available P (AP) and total P (TP) in Brazil at a fine resolution (30 m). Using the Random Forest machine learning algorithm and a database of topsoil (0-20 cm) with 28,572 samples for AP and 3154 for TP, we predicted P stocks based on environmental covariates related to soil formation processes. By dividing Brazil into two sub-regions, representing areas with native coverage and anthropogenic ones, we built independent predictive models for each sub-region. Our results show that Brazil has a TP stock of 531 Tg and an AP stock of 17.4 Tg. The largest soil TP stocks are in the Atlantic Forest biome (73.8 g.m2), likely due to higher organic carbon stocks in this biome. The largest AP stocks were in the Caatinga biome (2.51 g.m2) because of younger soils with low P adsorption capacity. We also found that fertilizer use significantly increased AP stocks in agricultural areas compared to native ones. Our results indicated that AP stocks strongly influenced Brazil's agricultural production, with a correlation coefficient ranging from 0.20 for coffee crops to 0.46 for soybean. The maps generated in this study are expected to contribute to the sustainable use of P in agriculture and environmental systems.
Soil organic carbon (SOC) is a dynamic soil property (DSP) that represents the largest portion of terrestrial carbon. Its relevance to carbon sequestration and the potential effects of land use on SOC storage, make it imperative to map across both space and time. Most regional-scale studies mapping SOC give static estimates and train different models for different periods with varying accuracies. We developed a flexible modeling approach called DSP-Scale to map SOC in both space and time. DSP-Scale uses ecological concepts and empirical data to predict DSP dynamics using inherent soil properties (static factors) and land cover changes (dynamic factors). We compiled SOC data for the 0–20 cm depth (SOC20) from 1441 points spanning a 25 million ha study area in the southeastern U.S. Coastal Plain, incorporating data from the Rapid Carbon Assessment, National Cooperative Soil Survey Soil Characterization database, and other regional studies. We developed a random forest model using climate, topography, soil survey, and land cover changes to predict SOC20 dynamics for five-year periods between 2001 and 2019. Our model explained 66 % and 59 % of the variation for the training and test data, respectively. Top predictors included mean annual precipitation, slope, and soil erosion class. Land cover 10 years before measurements of SOC20 was more important than current land cover for estimating SOC20. We estimated total SOC stocks of 207.1 and 208.3 Tg for 2001 and 2019, respectively. Highest gains of total SOC stock (0.9 Tg from 2001 to 2019) were associated with land cover change from mixed to evergreen forest. The greatest loss of total SOC stock (0.2 Tg) in the same period was associated with land cover change from pasture/hay to evergreen forest. We concluded that the DSP-Scale approach provides a flexible way to use dynamic and static factors affecting SOC stocks to predict changes in space and time at regional scales.
Background
Soil organic carbon (SOC) is a critical component of the global carbon cycle, and an accurate estimate of regional SOC stock (SOCS) would significantly improve our understanding of SOC sequestration and cycles. Zoige Plateau, locating in the northeastern Qinghai-Tibet Plateau, has the largest alpine marsh wetland worldwide and exhibits a high sensitivity to climate fluctuations. Despite an increasing use of optical remote sensing in predicting regional SOCS, optical remote sensing has obvious limitations in the Zoige Plateau due to highly cloudy weather, and knowledge of on the spatial patterns of SOCS is limited. Therefore, in the current study, the spatial distributions of SOCS within 100 cm were predicted using an XGBoost model—a machine learning approach, by integrating Sentinel-1, Sentinel-2 and field observations in the Zoige Plateau.
Results
The results showed that SOC content exhibited vertical distribution patterns within 100 cm, with the highest SOC content in topsoil. The tenfold cross-validation approach showed that XGBoost model satisfactorily predicted the spatial patterns of SOCS with a model efficiency of 0.59 and a root mean standard error of 95.2 Mg ha ⁻¹ . Predicted SOCS showed a distinct spatial heterogeneity in the Zoige Plateau, with an average of 355.7 ± 123.1 Mg ha ⁻¹ within 100 cm and totaled 0.27 × 10 ⁹ Mg carbon.
Conclusions
High SOC content in topsoil highlights the high risks of significant carbon loss from topsoil due to human activities in the Zoige Plateau. Combining Sentinel-1 and Sentinel-2 satisfactorily predicted SOCS using the XGBoost model, which demonstrates the importance of selecting modeling approaches and satellite images to improve efficiency in predicting SOCS distribution at a fine spatial resolution of 10 m. Furthermore, the study emphasizes the potential of radar (Sentinel-1) in developing SOCS mapping, with the newly developed fine-resolution mapping having important applications in land management, ecological restoration, and protection efforts in the Zoige Plateau.
Monitoring soil organic carbon (SOC) content is crucial for climate change mitigation and sustaining ecological balance. Despite the unparalleled advantages of hyperspectral data in capturing nuanced variations in soil properties through its high spectral resolution, effectively extracting useful features from numerous bands via spectral processing techniques remains a formidable challenge. This study proposes an integrated approach combining fractional-order derivative (FOD) technique and optimal band combination algorithm using ZY1-02D satellite hyperspectral data to estimate SOC in Northeast China’s Black soil region. Three modeling strategies were compared: (1) FOD-transformed reflectance (FOD spectra), (2) FOD spectra with traditional 2D spectral indices (FOD+2D SI), and (3) FOD spectra with new 3D spectral indices (FOD+3D SI). These strategies were implemented using the random forest model with the aim of the optimal SOC prediction. Results showed that the application of FOD technique for spectral transformation effectively addressed the challenges posed by overlapping peaks and baseline drift inherent in the original spectral reflectance. Additionally, FOD transformation enhanced subtle soil spectral features and yielded more pronounced spectral variations with increasing fractional order, as compared to the original spectral data and conventional integer-order derivatives (i.e., first and second-order derivatives). However, as the FOD order continued to increase beyond 1.4, the spectral curve exhibited amplified noise and distortion, thereby adversely impacting subsequent model development. The 3D spectral indices correlate more robustly with SOC than 2D indices. The model that combines 0.6-order FOD and 3D spectral indices achieved the best accuracy (R2=0.66, RMSE=2.99 g/kg and MAE=2.42 g/kg), significantly outperforming the models built by 0.6-order FOD spectra (R2=0.48, RMSE=3.65 g kg-1, and MAE=2.93 g kg-1) and 0.8-order FOD+2D SI modeling strategy (R2=0.55, RMSE=3.54 g kg-1, and MAE=2.85 g kg-1). These findings indicated that FOD and 3D spectral indices exhibit superior synergistic performance in SOC prediction, demonstrating their feasibility and providing valuable insights for large-scale soil property prediction and mapping using satellite hyperspectral data.
Integrated production systems composed of trees, crops and pastures have shown good results in improving soil quality and the capacity to store carbon in the soil, being efficient in mitigating greenhouse gas emissions. Despite this, changes in carbon stocks and soil organic matter fractions in the initial stages of implementing an agroforestry system remain unclear. This study evaluated the carbon balance and the dynamics of soil organic matter fractions in an agroforestry system conducted over a decade. Total carbon, labile carbon, carbon from particulate organic matter, organic carbon associated with minerals and inert carbon were determined at depths 0–10 cm, 10–20 cm and 20–40 cm. Soil carbon stocks were also estimated for the 0–40 cm depth. Total carbon increased in the agroforestry system compared with a low-productivity pasture. The total carbon stock in the last growing season (68.57 Mg ha−1) was close to the original soil stocks under native Cerrado vegetation (76.5 Mg ha−1). After 10 years, there was a positive balance in the soil carbon stock of both the total carbon and the soil organic matter fractions. The successional agroforestry system is a good alternative to increasing soil total carbon stocks and labile and non-labile fractions of soil organic matter.
Spatial soil databases can help model complex phenomena in which soils are a decisive factor – for example, evaluating agricultural potential or estimating carbon storage capacity. The Latin America and Caribbean Soil Information System, SISLAC, is a regional initiative promoted by the Food and Agriculture Organization's (FAO) Latin America and the Caribbean Soil Partnership to contribute to sustainable management of soil. SISLAC includes data from 49 084 soil profiles distributed unevenly across the continent, making it the region's largest soil database. In addition, there are other soil databases in the region with about 40 000 soil profiles that can be integrated into SISLAC and improve it. However, some problems hinder its usages, such as the quality of the data and their high dimensionality. The objective of this research is evaluate the quality of the SISLAC data and the other available soil databases to generate a new improved version that meets the minimum quality requirements to be used for different purposes or practical applications. The results show that 15 % of the existing soil profiles had an inaccurate description of the diagnostic horizons and 17 % of the additional profiles already existed in SISLAC; therefore, a total of 32 % of profiles were excluded for these two reasons. Further correction of an additional 4.5 % of existing inconsistencies improved overall data quality. The improved database consists of 66 746 profiles and is available for public use at https://doi.org/10.5281/zenodo.7876731 (Díaz-Guadarrama and Guevara, 2023). This revised version of SISLAC data offers the opportunity to generate information that helps decision-making on issues in which soils are a decisive factor. It can also be used to plan future soil surveys in areas with low density or where updated information is required.
Characterization of soil resources based on their morphological, physical and chemical properties. Comparison of machine learning algorithms(Random Forest, Support Vector Machine, Cubist, Artificial Neural Network) for prediction of soil properties. Mapping of predicted properties along with uncertainty using the best machine learning algorithm.
Background and Research Aims: Soil quality (SQ) is the basis for the Sustainability of Peasant Farming Systems (PFS). We hypothesized that different land uses modify soil quality through changes that can be analyzed by determining the Soil Quality Index (SQI).
Methods: Soil samples were collected from the 0-20 cm layer in five subsystems of peasant agroecosystems located in the municipalities of Solânea (A), Casserengue (B), and Serraria (C). SQI was calculated using non-linear scoring, while a principal component analysis was performed using all data (bulk and particle density, total porosity, particle size, pH, macronutrients, and soil organic carbon) to determine a Minimum Data Set (MDS).
Results: The MDS composed of P available, Ca ⁺² , Al ⁺³ , sand, silt, H+Al, base saturation (BS%), and the aluminum saturation (AS%) indicate that these parameters can serve as indicators for soil quality assessment in peasant agroecosystems. Sand and silt are related to pedogenic processes and parent material, while the remaining indicators reflect management practices. Land conversion from forest to cropland decreased nutrient availability and soil organic matter in agroecosystems A (Arenosol) and B (Luvisol) and increased the cation exchange capacity in agroecosystem C (Lixisol).
Conclusions: All agroecosystems showed low SQI values, highlighting the need to expand conservation practices in the studied agricultural subsystems, especially regarding the increase of soil organic matter. Our results contribute to improving the use and management of soils and the vulnerability assessment in peasant farming, an essential requirement for the sustainability of agroecosystems.
Implications for Conservation: Our results also demonstrated that agroforestry practices can significantly increase soil quality and soil carbon sequestration, a viable alternative for maintaining organic matter in areas susceptible to degradation.
Soil attributes such as granulometric fractions and Atterberg limits (LL: liquid limit, PL: plastic limit, and PI: plasticity index) are needed to assess off-road vehicle mobility (OVM) risks. Parameters describing these attributes are generally measured in soil samples collected from a few locations through cumbersome laboratory methods. Although diffuse reflectance spectroscopy (DRS) can rapidly yield estimates for soil attributes in samples collected from specific locations and digital soil mapping (DSM) can transform such discrete measurements into spatially-continuous inference systems, these two technologies are rarely used for assessing OVM risks. In this study, we combined the DRS and DSM approaches for deriving spatially-continuous estimates for the key vehicle mobility parameters (gravel, sand, and fine particles; Cu: coefficients of uniformity; Cc: coefficient of curvature; LL; and PI) and classified soils using the Unified Soil Classification System (USCS). A total of 204 soil samples were collected from the north-eastern Himalayan state of Sikkim for measuring these parameters along with spectral reflectance over the visible and near-infrared region. Results of the chemometric models in the DRS approach showed that the USCS parameters may be estimated with the coefficient of determination (R2) values as high as 0.72. The fine (<2 mm diameter) fraction spectra provided the best estimates for the Atterberg limits while a combination of spectra collected from fine and coarse (>2 mm diameter) fractions was effective in estimating other granulometric fractions except for sand, which was best estimated using the coarse fraction spectra. With the DSM approach allowing effective mapping of these parameters, a spatially-continuous framework to quantify soil-associated OVM risks was developed for Sikkim for the first time.
Soil organic carbon (SOC) stands out as a crucial indicator for assessing soil properties due to its direct impact on soil productivity. To delve into this, we collected 350 soil samples from depths ranging from 0 to 30 cm in the northwest region of Iran, measuring SOC levels. Concurrently, we obtained vegetation indices from Landsat 8 and Sentinel-2 satellite images. Subsequently, we employed machine learning techniques, specifically artificial neural network (ANN) and random forest (RF) models, to estimate the spatial distribution of SOC. In the subsequent phase, we classified the land units of the region based on physiography, vegetation, erosion, flooding, soil type, and depth for a more accurate comparison of estimation models. We then compared the average values of measured and predicted SOC within these land units. The evaluation of ANN and RF models, utilizing vegetation indices from Landsat 8 and Sentinel-2 satellites, revealed that the RF method, particularly when using vegetation indices from the Sentinel-2 satellite, exhibited superior accuracy in predicting SOC (R 2 = 0.8, RMSE = 0.19, and ρc = 0.81). Moreover, when comparing the estimated and predicted average values of SOC in different land units, we observed no significant difference between the measured and predicted averages using the RF method. This underscores the robustness of the RF model in accurately predicting SOC by leveraging vegetation indices extracted from Sentinel-2 satellite data. Consequently, the RF model emerges as a reliable tool for SOC evaluation, offering precise forecasts and contributing to cost and time savings by minimizing the need for extensive soil sampling and laboratory analysis.
In developing countries like Iran, where information is scarce, understanding the spatial variability of soil available phosphorous (SAP), one of the three major nutrients, is crucial for effective agricultural ecosystem management. This study aimed to predict and digitally map the spatial distribution and related uncertainty of SAP while also assessing the impact of environmental factors on SAP variability in the topsoils. A study area from northern Khuzestan province, Iran was selected as case study area. Three machine learning (ML) models, namely, Random Forest (RF), Artificial Neural Network (ANN), and Support Vector Regression (SVR), were used to develop predictive relationship between surface soil (0–10 cm) SAP content and environmental covariates derived from a digital elevation model and Landsat 8 images. A total of 250 topsoil samples were collected following the conditioned Latin Hypercube Sampling (cLHS) approach and several soil properties were measured in the laboratory. Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Lin’s Concordance Correlation Coefficient (LCCC) were used to determine the accuracy of models. The findings indicated that the RF algorithm demonstrated the most favorable performance, with a mean absolute error (MAE) of 0.85 mg SAP kg⁻¹, the lowest root mean square error (RMSE) of 0.99 mg SAP kg⁻¹, and the highest linear correlation coefficient (LCCC) values of 0.96. This suggests that the RF algorithm had the least tendency to overestimate or underestimate SAP contents compared to other methods. Consequently, the RF algorithm was selected as the optimal choice. Predictive ML models were employed to digitally map SAP contents within the region. Spatial patterns of SAP contents showed an increasing gradient from west to east. The spatial variability information provides a basis for developing sustainable production system in the area.
The grasslands in Southern Brazil have been been used for cattle ranching for more than 300 years. More recently, it has been recognized that these ecosystems provide key environmental services. Dramatic land use changes in the last decades, notably the conversion to cropland and silviculture, have brought about the risk of land degradation. We applied a combination of bibliographic research and geoprocessing operations to revise and update an earlier SOC stock estimate for the Brazilian Pampa. The calculated SOC stocks obtained in this study were compared with our original study and other SOC stock estimates. Our results were 7–27% higher than these estimates and approximately 8% lower than our previous estimate. Given the limited information on biome-wide soil C status, our soil C stock dataset and derived maps provide a useful assessment of SOC stocks in Pampa. Refining this approximation would require additional soil sampling and land use mapping to cover data gaps. This is a crucial research topic deriving from the Brazilian commitment to pursue sustainable agriculture systems in the context of mitigation of human-induced changes in the terrestrial C cycle.
This study investigated the impact of regional land abandonment in northeast Portugal. It specifically focused on carbon sequestration opportunities in the Upper Sabor River Watershed, situated in the northeast of Portugal, amidst agricultural land abandonment. The study involved mapping the distribution of soil organic carbon (SOC) across four soil layers (0–5 cm, 5–10 cm, 10–20 cm, and 20–30 cm) at 120 sampling points. The quantification of SOC storage (measured in Mg C ha−1) allowed for an analysis of its relationship with various landscape characteristics, including elevation, land use and land cover (LULC), normalized difference vegetation index (NDVI), modified soil-adjusted vegetation index (MSAVI), topographic wetness index (TWI), and erosion risk (ER). Six statistical tests were employed, including multivariate approaches like Cubist and Random Forest, within different scenarios to assess carbon distribution within the watershed’s soils. These modeling results were then utilized to propose strategies aimed at enhancing soil carbon storage. Notably, a significant discrepancy was observed in the carbon content between areas at higher elevations (>1000 m) and those at lower elevations (<800 m). Additionally, the study found that the amount of carbon stored in agricultural soils was often significantly lower than in other land use categories, including forests, mountain herbaceous vegetation, pasture, and shrub communities. Analyzing bi- and multivariate scenarios, it was determined that the scenario with the greatest number of independent variables (set 6) yielded the lowest RMSE (root mean squared error), serving as a key indicator for evaluating predicted values against observed values. However, it is important to note that the independent variables used in set 4 (elevation, LULC, and NDVI) had reasonably similar values. Ultimately, the spatialization of the model from scenario 6 provided actionable insights for soil carbon conservation and enhancement across three distinct elevation levels.
The book chapter deals with the concept, methodology and application of Digital Soil Mapping along with case study (digital soil moisture estimation) in GIS and Remote Sensing Perspective
Brazil is home to the largest tracts of tropical vegetation in the world, harbouring high levels of biodiversity and carbon. Several biomass maps have been produced for Brazil, using different approaches and methods, and for different purposes. These maps have been used to estimate historic, recent, and future carbon emissions from land use change (LUC). It can be difficult to determine which map to use for what purpose. The implications of using an unsuitable map can be significant, since the maps have large differences, both in terms of total carbon storage and its spatial distribution. This paper presents comparisons of Brazil's new ‘official’ carbon map; that is, the map used in the third national communication to the UNFCCC in 2016, with the former official map, and four carbon maps from the scientific literature. General strengths and weaknesses of the different maps are identified, including their suitability for different types of studies. No carbon map was found suitable for studies concerned with existing land use/cover (LULC) and LUC outside of existing forests, partly because they do not represent the current LULC sufficiently well, and partly because they generally overestimate carbon values for agricultural land. A new map of aboveground carbon is presented, which was created based on data from existing maps and an up-to-date LULC map. This new map reflects current LULC, has high accuracy and resolution (50 m), and a national coverage. It can be a useful alternative for scientific studies and policy initiatives concerned with existing LULC and LUC outside of existing forests, especially at local scales when high resolution is necessary, and/or outside the Amazon biome. We identify five ongoing climate policy initiatives in Brazil that can benefit from using this map.
Accurate and detailed spatial soil information about within-field variability is essential for variable-rate applications of farm resources. Soil total nitrogen (TN) and total carbon (TC) are important fertility parameters that can be measured with on-line (mobile) visible and near infrared (vis-NIR) spectroscopy. This study compares the performance of local farm scale calibrations with those based on the spiking of selected local samples from both fields into an European dataset for TN and TC estimation using three modelling techniques, namely gradient boosted machines (GBM), artificial neural networks (ANNs) and random forests (RF). The on-line measurements were carried out using a mobile, fiber type, vis-NIR spectrophotometer (305–2200 nm) (AgroSpec from tec5, Germany), during which soil spectra were recorded in diffuse reflectance mode from two fields in the UK. After spectra pre-processing, the entire datasets were then divided into calibration (75%) and prediction (25%) sets, and calibration models for TN and TC were developed using GBM, ANN and RF with leave-one-out cross-validation. Results of cross-validation showed that the effect of spiking of local samples collected from a field into an European dataset when combined with RF has resulted in the highest coefficients of determination (R2) values of 0.97 and 0.98, the lowest root mean square error (RMSE) of 0.01% and 0.10%, and the highest residual prediction deviations (RPD) of 5.58 and 7.54, for TN and TC, respectively. Results for laboratory and on-line predictions generally followed the same trend as for cross-validation in one field, where the spiked European dataset-based RF calibration models outperformed the corresponding GBM and ANN models. In the second field ANN has replaced RF in being the best performing. However, the local field calibrations provided lower R2 and RPD in most cases. Therefore, from a cost-effective point of view, it is recommended to adopt the spiked European dataset-based RF/ANN calibration models for successful prediction of TN and TC under on-line measurement conditions.
Forest soils are important components in the global C-cycle. Regional soil organic carbon (SOC) stock assessments are hampered by low spatial densities of routine soil inventories. In this study, we estimated the SOC stock in the upper meter of forest in the region of Flanders (N. Belgium) using four digital soil mapping techniques, i.e. multiple linear regression (MLR), boosted regression trees (BRT), artificial neural networks (ANN) and least-squares support vector machines (LS-SVM), and compared them with the results of a multi-level spatial generalisation (MLSG) approach.
Moso bamboo (Phyllostachys heterocycla (Carr.) Mitford cv. Pubescens) is an important timber substitute in China. Site specific stand management requires an accurate estimate of soil organic carbon (SOC) stock for maintaining stand productivity and understanding global carbon cycling. This study compared ordinary kriging (OK) and inverse distance weighting (IDW) approaches to study the spatial distribution of SOC stock within 0–60 cm using 111 soil samples in Moso bamboo forests in subtropical China. Similar spatial patterns but different spatial distribution ranges of SOC stock from OK and IDW highlighted the necessity to apply different approaches to obtain accurate and consistent results of SOC stock distribution. Different spatial patterns of SOC stock suggested the use of different fertilization treatments in Moso bamboo forests across the study area. SOC pool within 0–60 cm was 6.46 and 6.22 Tg for OK and IDW; results which were lower than that of conventional approach (CA, 7.41 Tg). CA is not recommended unless coordinates of the sampling locations are missing and the spatial patterns of SOC stock are not required. OK is recommended for the uneven distribution of sampling locations. Our results can improve methodology selection for investigating spatial distribution of SOC stock in Moso bamboo forests.
In this article, the predictive accuracy of 48 published pedotransfer functions (PTFs) for predicting soil bulk density (BD) was evaluated using soil database from the United States. In addition, new PTF for predicting BD using inputs of organic carbon content (OC) was proposed and validated. The results showed that 8 of the evaluated PTFs showed high performances (EF > 0.50, RMSE < 0.18 Mg m À3), 7 PTFs showed moderate performances (0.20 < EF < 0.50, 0.17 < RMSE < 0.20), 9 PTFs showed low performances (0 < EF < 0.20), while the rest of PTFs showed poor performances (EF < 0, RMSE > 0.24). New PTF has been developed using the entire database. The proposed PTF (BD = 1.449e À0.03OC , R 2 = 0.6802) requires only the organic carbon content as input. The developed PTF was compared to the best 10 performing PTFs evaluated in the study using dataset of 45,195 samples. The results showed that the new PTF had the best performance (EF = 0.59, RMSE = 0.13 Mg m À3). Ó 2016 Ain Shams University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Soil bulk density (ρ b) data are needed for a wide range of environmental studies. However, ρ b is rarely reported in soil surveys. An alternative to obtain ρ b for data-scarce regions, such as the Rio Doce basin in southeastern Brazil, is indirect estimation from less costly covari-ates using pedotransfer functions (PTF). This study primarily aims to develop region-specific PTFs for ρ b using multiple linear regressions (MLR) and random forests (RF). Secondly, it assessed the accuracy of PTFs for data grouped into soil horizons and soil classes. For that purpose, we compared the performance of PTFs compiled from the literature with those developed here. Two groups of data were evaluated as covariates: 1) readily available soil properties and 2) maps derived from a digital elevation model and MODIS satellite imagery, jointly with lithological and pedological maps. The MLR model was applied step-wise to select significant predictors and its accuracy assessed by means of cross-validation. The PTFs developed using all data estimated ρ b from soil properties by MLR and RF, with R 2 of 0.41 and 0.51, respectively. Alternatively, using environmental covariates, RF predicted ρ b with R 2 of 0.41. Grouping criteria did not lead to a significant increase in the estimates of ρ b. The accuracy of the 'regional' PTFs developed for this study was greater than that found with the 'compiled' PTFs. The best PTF will be firstly used to assess soil carbon stocks and changes in the Rio Doce basin.
Concerted political attention has focused on reducing deforestation, and this remains the cornerstone of most biodiversity conservation strategies. However, maintaining forest cover may not reduce anthropogenic forest disturbances, which are rarely considered in conservation programmes. These disturbances occur both within forests, including selective logging and wildfires, and at the landscape level, through edge, area and isolation effects. Until now, the combined effect of anthropogenic disturbance on the conservation value of remnant primary forests has remained unknown, making it impossible to assess the relative importance of forest disturbance and forest loss. Here we address these knowledge gaps using a large data set of plants, birds and dung beetles (1,538, 460 and 156 species, respectively) sampled in 36 catchments in the Brazilian state of Pará. Catchments retaining more than 69-80% forest cover lost more conservation value from disturbance than from forest loss. For example, a 20% loss of primary forest, the maximum level of deforestation allowed on Amazonian properties under Brazil's Forest Code, resulted in a 39-54% loss of conservation value: 96-171% more than expected without considering disturbance effects. We extrapolated the disturbance-mediated loss of conservation value throughout Pará, which covers 25% of the Brazilian Amazon. Although disturbed forests retained considerable conservation value compared with deforested areas, the toll of disturbance outside Pará's strictly protected areas is equivalent to the loss of 92,000-139,000 km(2) of primary forest. Even this lowest estimate is greater than the area deforested across the entire Brazilian Amazon between 2006 and 2015 (ref. 10). Species distribution models showed that both landscape and within-forest disturbances contributed to biodiversity loss, with the greatest negative effects on species of high conservation and functional value. These results demonstrate an urgent need for policy interventions that go beyond the maintenance of forest cover to safeguard the hyper-diversity of tropical forest ecosystems.
Tropical forests are significant carbon sinks and their soils' carbon storage potential is immense. However, little is known about the soil organic carbon (SOC) stocks of tropical mountain areas whose complex soil-landscape and difficult accessibility pose a challenge to spatial analysis. The choice of methodology for spatial prediction is of high importance to improve the expected poor model results in case of low predictor-response correlations. Four aspects were considered to improve model performance in predicting SOC stocks of the organic layer of a tropical mountain forest landscape: Different spatial predictor settings, predictor selection strategies, various machine learning algorithms and model tuning. Five machine learning algorithms: random forests, artificial neural networks, multivariate adaptive regression splines, boosted regression trees and support vector machines were trained and tuned to predict SOC stocks from predictors derived from a digital elevation model and satellite image. Topographical predictors were calculated with a GIS search radius of 45 to 615 m. Finally, three predictor selection strategies were applied to the total set of 236 predictors. All machine learning algorithms-including the model tuning and predictor selection-were compared via five repetitions of a tenfold cross-validation. The boosted regression tree algorithm resulted in the overall best model. SOC stocks ranged between 0.2 to 17.7 kg m-2, displaying a huge variability with diffuse insolation and curvatures of different scale guiding the spatial pattern. Predictor selection and model tuning improved the models' predictive performance in all five machine learning algorithms. The rather low number of selected predictors favours forward compared to backward selection procedures. Choosing predictors due to their indiviual performance was vanquished by the two procedures which accounted for predictor interaction.
There is a need for accurate estimate of soil organic carbon (SOC) stocks for understanding the role of alpine soils in the global carbon cycle. We tested a method for mapping digitally the continuous distribution of the SOC stock in three dimensions in the northeast of the Tibetan Plateau. The approach integrated the spatial distribution of the mattic epipedon which is a special surface horizon widespread and rich in organic matter in Tibetan grasslands. Prediction models resulted in high prediction accuracy. An average SOC stock in the mattic epipedon was estimated to be 4.99 kg m−2 in a mean depth of 14 cm. The amounts of SOC in the mattic epipedon, the upper 30 cm and 50 cm accounted for about 21%, 80% and 89%, respectively, of the total SOC stock in the upper 1 m depth. Compared with previous estimates, our approach resulted in more reliable predictions. The mattic epipedon was proven to be an important factor for modelling the realistic distribution of the SOC stock in Tibetan grasslands. Vegetation-related covariates have the most important influence on the distribution of the mattic epipedon and the SOC stock in the alpine grassland soils of northeast Tibetan Plateau.
For baselining and to assess changes in soil organic carbon (C) we need efficient soil sampling designs and methods for measuring C stocks. Conventional analytical methods are time-consuming, expensive and impractical, particularly for measuring at depth. Here we demonstrate the use of proximal soil sensors for estimating the total soil organic C stocks and their accuracies in the 0–10cm, 0–30cm and 0–100cm layers, and for mapping the stocks in each of the three depth layers across 2837ha of grazing land. Sampling locations were selected by probability sampling, which allowed design-based, model-assisted and model-based estimation of the total organic C stock in the study area. We show that spectroscopic and gamma attenuation sensors can produce accurate measures of soil organic C and bulk density at the sampling locations, in this case every 5cm to a depth of 1m. Interpolated data from a mobile multisensor platform were used as covariates in Cubist to map soil organic C. The Cubist map was subsequently used as a covariate in the model-assisted and model-based estimation of the total organic C stock. The design-based, model-assisted and model-based estimates of the total organic C stocks in the study area were similar. However, the variances of the model-assisted and model-based estimates were smaller compared to those of the design-based method. The model-based method produced the smallest variances for all three depth layers. Maps helped to assess variability in the C stock of the study area. The contribution of the spectroscopic model prediction error to our uncertainty about the total soil organic C stocks was relatively small. We found that in soil under unimproved pastures, remnant vegetation and forests there is good rationale for measuring soil organic C beyond the commonly recommended depth of 0–30cm.
Soil bulk density is an important soil parameter directly related to a number of soil properties and processes and required to estimate element stocks in soils on an area basis. The measure of ρb is expensive and time-consuming and thus is often excluded from ordinary analyses. It is thus necessary the development of proper pedotransfer functions (PTF) to estimate ρb from parameters ordinarily included in soil analyses. In this study we used a geochemical database of 115 epipedons from Galicia (NW Spain) to test 3 different statistical methods – multiple linear regression, random forest and neural networks – in order to develop a PTF linking bulk density to organic matter content and soil textural fractions. Random forest was the model that presented the highest predictive performance (R-squared=0.90; RMSE=0.14; ME=0.03). This PTF was used to generalize a map of ρb covering the study area. Soil bulk density in Galicia is mainly related to the soil carbon content, peat soils being the features with lower ρb in this study area.
80% of arable land in Africa has low soil fertility and suffers from physical soil problems. Additionally, significant amounts of nutrients are lost every year due to unsustainable soil management practices. This is partially the result of insufficient use of soil management knowledge. To help bridge the soil information gap in Africa, the Africa Soil Information Service (AfSIS) project was established in 2008. Over the period 2008-2014, the AfSIS project compiled two point data sets: the Africa Soil Profiles (legacy) database and the AfSIS Sentinel Site database. These data sets contain over 28 thousand sampling locations and represent the most comprehensive soil sample data sets of the African continent to date. Utilizing these point data sets in combination with a large number of covariates, we have generated a series of spatial predictions of soil properties relevant to the agricultural management-organic carbon, pH, sand, silt and clay fractions, bulk density, cation-exchange capacity, total nitrogen, exchangeable acidity, Al content and exchangeable bases (Ca, K, Mg, Na). We specifically investigate differences between two predictive approaches: random forests and linear regression. Results of 5-fold cross-validation demonstrate that the random forests algorithm consistently outperforms the linear regression algorithm, with average decreases of 15-75% in Root Mean Squared Error (RMSE) across soil properties and depths. Fitting and running random forests models takes an order of magnitude more time and the modelling success is sensitive to artifacts in the input data, but as long as quality-controlled point data are provided, an increase in soil mapping accuracy can be expected. Results also indicate that globally predicted soil classes (USDA Soil Taxonomy, especially Alfisols and Mollisols) help improve continental scale soil property mapping, and are among the most important predictors. This indicates a promising potential for transferring pedological knowledge from data rich countries to countries with limited soil data.
To help meet the increasing need for knowledge and data on the spatial distribution of soils, readily applied multiple linear regression models were developed for key soil properties over eastern Australia. Selected covariates were used to represent the key soil forming factors of climate (annual precipitation and maximum temperature), parent material (a lithological silica index) topography (new topo-slope and aspect indices) and biota (a modified land disturbance index).
The models are presented at three depth intervals (0-10, 10-30 and 30-100 cm) and are of variable but generally moderate statistical strength, with concordance correlation coefficients in the order of 0.7 for organic carbon (OC) upper depth, pHca, sum of bases, CEC and sand, but somewhat lower (0.4-0.6) for OC lower depths, total phosphorous, clay and silt. The pragmatic models facilitate soil property predictions at individual sites using only climate and field collected data. They were also moderately effective for deriving digital soil maps over the State of New South Wales and a regional catchment. The models and derived maps compared well in predictive ability to those derived from more sophisticated techniques involving Cubist decision trees with remotely sensed covariates. The readily understood and interpreted nature of these products means they may provide a useful introduction to the more advanced digital soil modelling and mapping (DSMM) techniques. The models provide useful information and broader insights into the factors controlling soil distribution in eastern Australia and beyond, including the change in a soil property with a given unit change in a covariate.
Estimation of carbon contents and stocks are important for carbon sequestration, greenhouse gas emissions and national carbon balance inventories. For Denmark, we modeled the vertical distribution of soil organic carbon (SOC) and bulk density, and mapped its spatial distribution at five standard soil depth intervals (0-5, 5-15, 15-30, 30-60 and 60-100 cm) using 18 environmental variables as predictors. SOC distribution was influenced by precipitation, land use, soil type, wetland, elevation, wetness index, and multi-resolution index of valley bottom flatness. The highest average SOC content of 20 g kg-1 was reported for 0-5 cm soil, whereas there was on average 2.2 g SOC kg-1 at 60-100 cm depth. For SOC and bulk density prediction precision decreased with soil depth, and a standard error of 2.8 g kg-1 was found at 60-100 cm soil depth. Average SOC stock for 0-30 cm was 72 t ha-1 and in the top 1 m there was 120 t SOC ha-1. In total, the soils stored approximately 570 Tg C within the top 1 m. The soils under agriculture had the highest amount of carbon (444 Tg) followed by forest and semi-natural vegetation that contributed 11% of the total SOC stock. More than 60% of the total SOC stock was present in Podzols and Luvisols. Compared to previous estimates, our approach is more reliable as we adopted a robust quantification technique and mapped the spatial distribution of SOC stock and prediction uncertainty. The estimation was validated using common statistical indices and the data and high-resolution maps could be used for future soil carbon assessment and inventories.
Köppen's climate classification remains the most widely used system by geographical and climatological societies across the world, with well recognized simple rules and climate symbol letters. In Brazil, climatology has been studied for more than 140 years, and among the many proposed
methods Köppen's system remains as the most utilized. Considering Köppen's climate classification importance for Brazil (geography, biology, ecology, meteorology, hydrology, agronomy, forestry and environmental sciences), we developed a geographical information system to identify
Köppen's climate types based on monthly temperature and rainfall data from 2,950 weather stations. Temperature maps were spatially described using multivariate equations that took into account the geographical coordinates and altitude; and the map resolution (100 m) was similar to the
digital elevation model derived from Shuttle Radar Topography Mission. Patterns of rainfall were interpolated using kriging, with the same resolution of temperature maps. The final climate map obtained for Brazil (851,487,700 ha) has a high spatial resolution (1 ha) which allows to observe
the climatic variations at the landscape level. The results are presented as maps, graphs, diagrams and tables, allowing users to interpret the occurrence of climate types in Brazil. The zones and climate types are referenced to the most important mountains, plateaus and depressions, geographical
landmarks, rivers and watersheds and major cities across the country making the information accessible to all levels of users. The climate map not only showed that the A, B and C zones represent approximately 81%, 5% and 14% of the country but also allowed the identification of Köppen's
climates types never reported before in Brazil.
Carbon stored in soils worldwide exceeds the amount of carbon stored in phytomass and the atmosphere. Despite the large quantity of carbon stored as soil organic carbon (SOC), consensus is lacking on the size of global SOC stocks, their spatial distribution, and the carbon emissions from soils due to changes in land use and land cover. This article summarizes published estimates of global SOC stocks through time and provides an overview of the likely impacts of management options on SOC stocks. We then discuss the implications of existing knowledge of SOC stocks, their geographical distribution and the emissions due to management regimes on policy decisions, and the need for better soil carbon science to mitigate losses and enhance soil carbon stocks.
We present a piecewise linear decision tree model for predicting percent of soil organic C (SOC) in the agricultural zones of Australia generated using a machine learning approach. The inputs for the model are a national database of soil data, national digital surfaces of climate, elevation, and terrain variables, Landsat multispectral scanner data, lithology, land use, and soil maps. The model and resulting map are evaluated, and insights into biogeological surficial processes are discussed. The decision tree splits the overall data set into more homogenous subsets, thus in this case, it identifies areas where SOC responds closely to climatic and other environmental variables. The spatial pattern of SOC corresponds well to maps of estimated primary productivity and bioclimatic zones. Topsoil organic C levels are highest in the high rainfall, temperate regions of Tasmania, Victoria, and Western Australia, along the coast of New South Wales and in the wet tropics of Queensland; and lowest in arid and semiarid inland regions. While this pattern broadly follows continental vegetation, soil moisture, and temperature patterns, it is governed by a spatially variable hierarchy of different climatic and other variables across bioregions of Australia. At the continental scale, soil moisture level, rather than temperature, seems most important in controlling SOC.
Soil is often described as mantling the land more or less continuously with the exception being where there is bare rock and ice (Webster and Oliver 2006). Our understanding of soil variation in any region is usually based on only a small number of observations made in the field. Across the spatial domain of the region of interest, predictions of the spatial distribution of soil properties are made at unobserved locations based on the properties of the small number of soil observations. There are two principal approaches for making predictions of soil at unobserved locations. The first approach subdivides the soil coverage into discrete spatial units within which the soils conform to the characteristics of a class in some soil classification (Heuvelink and Webster 2001). The second approach treats soils as a suite of continuous variables and attempts to describe the way these variables vary across the landscape (Heuvelink and Webster 2001). The second approach is necessarily quantitative, as it requires numerical methods for interpolation between the locations of actual soil observations.
Efficient and effective modelling methods to assess soil organic carbon (SOC) stock are central in understanding the global carbon cycle and informing related land management decisions. However, mapping SOC stocks in semi-arid rangelands is challenging due to the lack of data and poor spatial coverage. The use of remote sensing data to provide an indirect measurement of SOC to inform digital soil mapping has the potential to provide more reliable and cost-effective estimates of SOC compared with field-based, direct measurement. Despite this potential, the role of remote sensing data in improving the knowledge of soil information in semi-arid rangelands has not been fully explored. This study firstly investigated the use of high spatial resolution satellite data (seasonal fractional cover data; SFC) together with elevation, lithology, climatic data and observed soil data to map the spatial distribution of SOC at two soil depths (0–5 cm and 0–30 cm) in semi-arid rangelands of eastern Australia. Overall, model performance statistics showed that random forest (RF) and boosted regression trees (BRT) models performed better than support vector machine (SVM). The models obtained moderate results with R2 of 0.32 for SOC stock at 0–5 cm and 0.44 at 0–30 cm, RMSE of 3.51 Mg C ha−1 at 0–5 cm and 9.16 Mg C ha−1 at 0–30 cm without considering SFC covariates. In contrast, by including SFC, the model accuracy for predicting SOC stock improved by 7.4–12.7% at 0–5 cm, and by 2.8–5.9% at 0–30 cm, highlighting the importance of including SFC to enhance the performance of the three modelling techniques. Furthermore, our models produced a more accurate and higher resolution digital SOC stock map compared with other available mapping products for the region. The data and high-resolution maps from this study can be used for future soil carbon assessment and monitoring.
Tropical peatland holds a large amount of carbon in the terrestrial ecosystem. Indonesia, responding to the global climate issues, has legislation on the protection and management of the peat ecosystem. However, this effort is hampered by the lack of fine-scale, accurate maps of peat distribution and its thickness. This paper presents an open digital mapping methodology, which utilises open data in an open-source computing environment , as a cost-effective method for mapping peat thickness and estimating carbon stock in Indonesian peatlands. The digital mapping methodology combines field observations with factors that are known to influence peat thickness distribution. These factors are represented by multi-source remotely-sensed data derived from open and freely available raster data: digital elevation models (DEM) from SRTM, geographical information , and radar images (Sentinel and ALOS PALSAR). Utilising machine-learning models from an open-source software, we derived spatial prediction functions and mapped peat thickness and its uncertainty at a grid resolution of 30 m. Peat volume can be calculated from the thickness map, and based on measurements of bulk density and carbon content, carbon stock for the area was estimated. The uncertainty of the estimates was calculated using error propagation rules. We demonstrated this approach in the eastern part of Bengkalis Island in Riau Province, covering an area around 50,000 ha. Results showed that digital mapping method can accurately predict the thickness of peat, explaining up to 98% of the variation of the data with a median relative error of 5% or an average error of 0.3 m. The accuracy of this method depends on the number of field observations. We provided an estimate of the cost and time required for map production, i.e. 2 to 4 months with a cost between $0.3 and $0.5/ha for an area of 50,000 ha. Obviously, there is a tradeoff between cost and accuracy. The advantages and limitations of the method were further discussed. The methodology provides a blueprint for a national-scale peat mapping.
Core Ideas
Regression kriging with elevation, topographic wetness index, field (west vs. east), and irrigation (yes vs. no) accurately predicted soil organic C (SOC) in the 0 to 15‐ and 15 to 30‐cm layers.
Lack of spatial structure and a lack of relationships between SOC and auxiliary variables precluded the use of regression kriging for the 30 to 60‐ and 60 to 90‐cm layers.
From 0 to 15 cm, SOC in the west field increased by 7% because of gains in irrigated portions of the field, but no changes were found in the east field or from 15 to 30 cm in either field.
Simple means indicated SOC gains of 13% in the 30 to 60‐cm layer and 24% in the 60 to 90‐cm layer across both fields.
Typical field management practices associated with large, modern dairies can sequester SOC.
Accurate measurement of soil organic C (SOC) stock changes over time is essential to verify management effects on C sequestration. This study quantified spatial and temporal changes in SOC stocks on adjacent 65‐ha corn ( Zea mays L.) silage–alfalfa ( Medicago sativa L.) fields receiving liquid dairy manure in west central Minnesota. We used regression kriging to interpolate SOC in four soil layers in 2006 and 2015, and calculated stock changes over time. Regression kriging with elevation, topographic wetness index, field (west vs. east), and irrigation (yes vs. no) accurately predicted SOC in the 0 to 15‐cm ( R 2 = 0.89) and 15 to 30‐cm layers ( R 2 = 0.51–0.95), where variogram analysis indicated moderate to strong spatial correlation. From 0 to 15 cm, SOC in the west field increased by 7% (+4.5 Mg C ha –1 ) over the study period caused by gains in irrigated portions of the field. No changes were found in the east field or from 15 to 30 cm in either field. Below 30 cm, a lack of spatial structure and a lack of relationships between SOC and auxiliary variables was found, but simple means indicated SOC gains of 13% (+4.7 Mg C ha –1 ) in the 30 to 60‐cm layer and 24% (+3.9 Mg C ha –1 ) in the 60 to 90‐cm layer across both fields. Regression kriging with easily acquired auxiliary variables offers a highly accurate method of monitoring SOC stock changes over time to 30 cm depth. Current management practices maintain or increase SOC in these fields.
As soil oxidizable carbon (Cox) has several different absorptions in the visible– and near infrared (vis–NIR) region due to its complex composition, multivariate calibration techniques such as multiple linear regression (MLR), partial least squares regression (PLSR), support vector machines (SVM) or random forest (RF) can be advantageously used to obtain a good prediction. Besides that, the content of Cox is often affected by the character of the terrain, mainly due to prevailing water regime with associated transport and sedimentation processes. Therefore, the question arises; if predictive models calibrated by combining vis–NIR diffuse reflectance spectroscopy (vis–NIR DRS, 350–2500 nm) and the digital elevation model (DEM) derivatives will provide a more accurate estimate of Cox. Focused on a sloping arable land (100 ha) affected by distinct water erosion, we tested for this purpose two conceptually different predictive approaches that differ in the nature of the spectroscopic predictor variables. In Approach A that relied on absorption feature (AF) parameters, the inclusion of DEM derivatives resulted in improved Cox prediction using all the tested calibration techniques, i.e. MLR, RF, PLSR and SVM. The MLR prediction that was the most accurate among all others improved from R²cv = 0.81 (vis–NIR DRS dataset) or 0.50 (DEM derivatives dataset) to 0.84 (combination of both). For that prediction especially AF centered at 500, 700, 900, 1800, 1900, 2200 and 2400 nm, as well as elevation, LS factor and plan curvature were important. In contrast, in the Approach B that relied on reflectance (RF and PLSR) or normalized reflectance (SVM) values at each wavelength, no positive effect of inclusion of DEM derivatives was observed.
Thesupport-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
This work presents the first high-resolution map of soil organic carbon (SOC) in mainland France, including soils below 30 cm. The research was performed within the framework of GlobalSoilMap (GSM). SOC predictions for different depth layers (0–5 cm, 5–15 cm, 15–30 cm, 30–60 cm, 60–100 cm and > 100 cm) were made at 90 and 500 m resolution for mainland France, along with their upper and lower confidence intervals. The maps were developed using data mining and an elaborate cross-validation scheme. The 90 m maps were compared to 500 m resolution GlobalSoilMap maps and the SoilGrids1km (SG1km) product. The latter is a global model for predicting soil properties for the same depth layers, at 1 km resolution.
Function estimation/approximation is viewed from the perspective of numerical optimization iti function space, rather than parameter space. A connection is made between stagewise additive expansions and steepest-descent minimization. A general gradient descent "boosting" paradigm is developed for additive expansions based on any fitting criterion. Specific algorithms are presented for least-squares, least absolute deviation, and Huber-M loss functions for regression, and multiclass logistic likelihood for classification. Special enhancements are derived for the particular case where the individual additive components are regression trees, and tools for interpreting such "TreeBoost" models are presented. Gradient boosting of regression trees produces competitives highly robust, interpretable procedures for both regression and classification, especially appropriate for mining less than clean data. Connections between this approach and the boosting methods of Freund and Shapire and Friedman, Hastie and Tibshirani are discussed.
When predicting a categorical outcome, some measure of classification accuracy is typically used to evaluate the model’s effectiveness. However, there are different ways to measure classification accuracy, depending of the modeler’s primary objectives. Most classification models can produce both a continuous and categorical prediction output. In Section 11.1, we review these outputs, demonstrate how to adjust probabilities based on calibration plots, recommend ways for displaying class predictions, and define equivocal or indeterminate zones of prediction. In Section 11.2, we review common metrics for assessing classification predictions such as accuracy, kappa, sensitivity, specificity, and positive and negative predicted values. This section also addresses model evaluation when costs are applied to making false positive or false negative mistakes. Classification models may also produce predicted classification probabilities. Evaluating this type of output is addressed in Section 11.3, and includes a discussion of receiver operating characteristic curves as well as lift charts. In Section 11.4, we demonstrate how measures of classification performance can be generated in R.
Mulching effect on carbon (C) sequestration depends on soil properties, mulch material, and the rate and duration of application. Thus, rate of soil C sequestration was assessed on a 15 year field study involving three levels of wheat straw at 0 (M0), 8 (M8) and 16 (M16) Mg ha−1 yr−1, at two levels (244 kg N ha−1 yr−1, F1 and without, F0) of supplemental N. Soil C concentration was assessed in relation to aggregation and occlusion in aggregates of a silt loam Alfisol under a no-till (NT) and crop-free system in central Ohio. In comparison to control, soil organic carbon (SOC) concentration in the 0–10 cm depth of bulk soil increased by 32% and 90% with M8 and M16 treatments with a corresponding increase in the SOC stock by 21–25% and 50–60%, respectively. With increase in rate of residue mulch, proportion of water stable aggregates (small macroaggregates, >250 μm size) increased by 1.4–1.8 times and of microaggregates (53–250 μm) by 1.4 times. Fertilizer N significantly increased the SOC concentration of small macroaggregates under M16 treatments only. Ultra-sonication showed that 12–20% of SOC occluded in the inter-microaggregate space of small macroaggergates, was a function of both mulch and fertilizer rates. Significantly higher and positive correlation of greenhouse gases (GHGs), CO2, CH4 and N2O flux was observed with C and N concentrations of small macroaggregates and also of the occluded fraction of small macroaggregates. The higher correlation coefficient indicated the latter to be prone to microbial attack. On the contrary, non-significant relationship with C and N concentrations of microaggregates indicate a possible protection of microaggregate C. The diurnal fluxes of CO2, CH4 and N2O were the lowest under bare soil and the highest under high mulch rate with added N, with values ranging from 1.51 to 2.31 g m−2 d−1, −2.79 to 3.15 mg m−2 d−1 and 0.46 to 1.02 mg m−2 d−1, respectively. Mulch rate affected the GHGs flux more than did the fertilizer rates. The net global warming potential (GWP) was higher for high mulch (M16) than low mulch (M8) rates, with values ranging from 0.46 to 0.57 Mg CO2 equivalent – C ha−1 yr−1 (M8) and 1.98 to 3.05 Mg CO2 equivalent – C ha−1 yr−1 (M16). In general, mulch rate determined the effect of fertilizers. The study indicated that over long-term, a mulch rate between 8 and 16 Mg ha−1 yr−1 may be optimal for Alfisols in Central Ohio.
Contents 1. Introduction 94 2. Information for Soil Assessment 95 2.1 Mapping, modeling, and monitoring 96 2.2 The new global imperative for soil science 96 2.3 Current map coverage 98 2.4 Conventional concepts of soil survey and classification 99 2.5 Grids or polygons or both? 100 3. Digital Soil Mapping and Origins of GlobalSoilMap 101 4. Technical Specifications of GlobalSoilMap 102 4.1 Defining the soil individual 103 4.2 Definition of the grid 107 5. Minimum Data Set 109 5.1 Concept 109 5.2 Time 113 Advances in Agronomy, Volume 125 # 2014 Elsevier Inc. ISSN 0065-2113 All rights reserved.
As the largest pool of terrestrial organic carbon, soils interact strongly with atmospheric composition, climate, and land cover change. Our capacity to predict and ameliorate the consequences of global change depends in part on a better understanding of the distributions and controls of soil organic carbon (SOC) and how vegetation change may affect SOC distributions with depth. The goals of this paper are (1) to examine the association of SOC content with climate and soil texture at different soil depths; (2) to test the hypothesis that vegetation type, through patterns of allocation, is a dominant control on the vertical distribution of SOC; and (3) to estimate global SOC storage to 3 m, including an analysis of the potential effects of vegetation change on soil carbon storage. We based our analysis on >2700 soil profiles in three global databases supplemented with data for climate, vegetation, and land use. The analysis focused on mineral soil layers. Plant functional types significantly affected the v...
In order to construct prediction intervals without the cumbersome–and typically unjustifiable–assumption of Gaussianity, some form of resampling is necessary. The regression set-up has been well-studied in the literature but time series prediction faces additional difficulties. The paper at hand focuses on time series that can be modeled as linear, nonlinear or nonparametric autoregressions, and develops a coherent methodology for the construction of bootstrap prediction intervals. Forward and backward bootstrap methods using predictive and fitted residuals are introduced and compared. We present detailed algorithms for these different models and show that the bootstrap intervals manage to capture both sources of variability, namely the innovation error as well as estimation error. In simulations, we compare the prediction intervals associated with different methods in terms of their achieved coverage level and length of interval.
Brazil contains the world's largest expanse of tropical forest, but its forests are experiencing high levels of conversion to other uses. There is concern that releases of CO2 and other greenhouse gases resulting from deforestation will contribute to global climate change. The total amount of C that could be released by deforestation depends upon the amount currently contained in the terrestrial biota and soils. Knowledge of the areas of Brazil's major ecosystems and land use types and their C densities was used to estimate the total amount of C stored in vegetation, litter and coarse woody debris, and soils. The total estimated C pools were (58–81) × 109 Mg C in vegetation, (6–9) × 109 Mg C in litter and coarse woody debris, and about 72 × 109 Mg C in soil. Over 80% of the vegetation pool was contained in the closed tropical moist forests of Brazil.
Soil organic carbon (SOC) is one of the most important parameters affecting the hydraulic characteristics of natural soils. Despite being rather easy to measure, SOC is known to be highly variable in space. In this study, vegetation, climate, and morphology factors were used to reproduce the spatial distribution of SOC in the mineral horizons of forest and grassland areas in north-western Italy and the feasibility of the approach was evaluated. When the overall sample (114 samples) was analyzed, average annual rainfall and elevation were significant descriptors of the SOC variability. However, a large part of the variability remains unexplained. Two stratification criteria were then adopted, based on vegetation and topographic properties. We obtained an improvement of the quality of the estimates, particularly for grasslands and forests in the absence of local curvatures. These results indicate that the spatial variability of soil organic matter is scarcely reproducible at the regional scale, unless an a-priori reduction of the heterogeneity is applied. A discussion on the feasibility of applying stratification criteria to deal with heterogeneous samples closes the paper.
Forest management and associated litter inputs and decomposition rates are thought to affect the carbon storage in mineral soils. Here, we studied the effects of forest management on soil organic carbon (OC) stocks in density fractions of Ah-horizons in soils that developed on loess. We used 82 beech (Fagus sylvatica L.) dominated forest plots in Thuringia, Germany that differed in their management (unmanaged forest, forests under age-class management and forests under selection cutting forest). After density fractionation of the mineral soil with a 1.6 g cm−3 polytungstate solution we determined OC concentrations and stocks as well as CN-ratios in the free (f-LF) light fraction, the occluded (o-LF) light fraction and in the mineral associated organic matter (MOM) fraction. In our study, Ah-horizons of beech forests stored on average 2.6 ± 0.2 kg m−2 (38.7 ± 1.3 kg m−3) OC. The results showed that 37% of the bulk soil OC was stored in the light fractions. We could show that OC stocks in the light fraction were significantly affected by the amount of C stored in organic layers (p = 0.011). The OC stocks in the organic layers, in turn, were higher in unmanaged forests and in forests under selection cutting. This suggests a sensitivity of unprotected OC in the f-LF of beech forests against forest management. In contrast to the f-LF, the OC stocks in the MOM fraction are mainly controlled by pedogenic properties such as clay and iron oxide content. Even after several decades of forest management and with large sample size, an effect of forest management on the stable MOM fraction could not be detected.
Soil organic matter (SOM) content and texture are important factors affecting carbon (C) and nitrogen (N) mineralisation under constant soil moisture but their effects on organic matter mineralisation and associated biogenic gas (carbon dioxide (CO2) and nitrous oxide (N2O)) production during dry/wet cycles is poorly understood. A laboratory incubation study was conducted to quantify CO2 and N2O production during sequential dry/wet cycles and under constant soil moisture conditions along a gradient of SOM contents in two soil types representing different texture classes (silt loam vs. clay loam). Three soil moisture treatments were established: wet (WW; field capacity), moderately dry (MD; 120% of soil moisture content (SMC) at wilting point (WP)) and very dry (VD; 80% of SMC at WP). To each of the two ‘dry’ treatments two different dry/wet treatments were applied where the soils were either maintained continuously dry (MD & VD) or subjected to three sequential 20-day long dry/wet cycles (MDW & VDW) during the treatment phase of the experiment. At field capacity soil moisture content, the rate of C mineralisation increased with increases in SOC content and the increase per unit of C was twice as high in silt loam (0.30 mg CO2-C g−1 SOC d−1) as in clay loam (0.13 mg CO2-C g−1 SOC d−1) soils. N2O-N emissions also increased with increasing in SOC content. However, in contrast to C mineralisation, the effect was four-fold greater for clay loam (1.38 μg N2O-N g−1 SOC d−1) than silt loam (0.32 μg N2O-N g−1 SOC d−1) soils. Following rewetting, the VDW and MDW soils produced a short-term C mineralisation flush that was, on average, 30% and 15% greater, respectively, than in WW soils. However, the flush of C mineralisation was not sufficient to compensate for the reduction in mineralisation during the drying phase of each cycle, resulting in a lower total C mineralisation from MDW and VDW soils, on average, compared with WW soils over the three sequential dry/wet cycles. The C mineralisation flush also remained a relatively constant proportion of the total C mineralised from both silt loam (23%) and clay loam soils (22%), irrespective of their SOC content. In contrast, the short-term flush of N2O that followed rewetting of dry soil accounted for 62% and 68% of the total N2O emissions from silt loam and clay loam soils, respectively. On average, the total N2O emissions from dry/wet treatments imposed on silt loam and clay loam soils were 33% and 270% greater, respectively, than from the WW treatments, though the effect varied greatly and depended on SOC content. Overall, N2O emissions were highest where we had a combination of fine texture, an adequate supply of available C (i.e. high SOM content), and a water-filled pore space (WFPS) > 0.60 cm cm−3 at field capacity. Prediction of C mineralisation over dry/wet cycles using mineralisation data from soils at constant moisture content is possible, but knowledge of the stress history for the soil would be required to improve accuracy. The prediction of N2O-N emissions during dry/wet cycles using emission data from soils at constant moisture was very inaccurate, due to the inherent spatial variability of N2O emissions.
We describe a model to predict regolith thickness in a 128,000 ha study area in the central Mt Lofty Ranges in South Australia. The term regolith encompasses soil (A and B horizons) and highly weathered bedrock (C horizon). The thickness of the regolith has a major control on water holding capacity for plant growth and movement of water through the landscape, and as such is important in hydropedological modelling and in evaluating land suitability, e.g. for forestry and agriculture. Thickness estimates also have direct application in mineral exploration and seismic risk assessment.Geology and landscape evolution within the study area is complex, reflecting the variable nature of bedrock materials, and the partial preservation of deeply weathered profiles as a consequence of weathering processes dating to the Cenozoic, or possibly older. These characteristics, together with strong climatic gradients, make the study area an ideal location to understand environmental and landscape evolution controls on weathering depth. The area also features weathered landscape analogues to many parts of southern Australia. We use a digital soil mapping rules based approach to develop a model to predict regolith thickness. This model uses statistically-based relationships established between 714 regolith thickness measurements and 29 geographic environmental covariates including geology, geochemistry, terrain and climate themes, and with a ground resolution of 10 m. Accuracy testing based on a 75:25% training:test data split on the resulting map established a correlation R2 of 0.64. This result is encouraging and is a significant advance over regolith depth mapping by traditionally-based regolith-landscape mapping methods. Finally, it leads the way towards a nationally consistent regolith thickness map for landscape scale environmental simulation modelling and decision support.
The objective of this study was to predict and map SOC stocks at different depth intervals within the upper I-in depth using profile depth distribution functions and ordinary kriging. These approaches were tested for the state of Indiana as a case study. A total of 464 pedons representing 204 soil series was obtained from the National Soil Survey Center database. Another 48 soil profile samples were collected to better represent the heterogeneity of the environmental variables. Two methods were used to model the depth distribution of the SOC stocks using negative exponential profile depth functions. In Procedure A, the functions to describe the depth distribution of volumetric C content for each soil profile were fitted using nonlinear least squares. In Procedure B, the exponential functions were fitted to describe the depth distribution of the cumulative SOC stocks. The parameters of the functions were interpolated for the entire study area using ordinary kriging on 81% of the data points (n = 414). The integral of the exponential function up to the desired depth was used to predict SOC stocks within the 0- to 1-, 0- to 0.5-, and 0.5- to 1-m depth intervals. These estimates were validated using the remaining 19% (n = 98) of the data. Procedure B showed a higher prediction accuracy for all depths, with higher rand lower RMSE values. The highest prediction accuracy (r = 0.75, RMSE = 2.89 kg m(-2)) was obtained for SOC stocks in the 0- to 0.5-m depth interval. Using Procedure B, SOC stocks within the top 1 m of Indiana soils were estimated to be 0.90 Pg C.
The results of this study provide the first baseline for predicting D(b) from soil properties for soils across the Amazon basin. Bulk density values are needed to convert nutrient content and organic carbon (OC) content to weight of nutrient and OC per unit area; unfortunately, common field methods to measure D(b) are limited with regard to reliable, complete, and uniform soil data. Much effort has been made in finding alternative solutions to predict D(b) from soil properties. We hypothesized that D(b) could be reliably estimated by multiple regression of OC, soil textural properties, and some chemical properties. Using the data of 323 soil horizons from the Brazilian Amazon basin, a stepwise multiple regression (SMR) procedure was developed to predict D(b) from other soil properties. Multiple regression relationships were obtained for all the data, which were also partitioned by layer and then by main soil order: Latossolos (Oxisols, 62 horizons) and Podzolicos (Alfisols and Ultisols; 212 horizons). The SMR on all the data showed that clay content is the best predictor of D(b), accounting for 37% of the variation. Adding OC content increased the explained variance up to nearly 50%. Predictions of the models were improved when the data were partitioned by order and by horizon type. In the case of Latossolos (Oxisols), the use of OC and clay content as predictors increased the percentage of explained variation, reaching 71% using all layers and 79% for A horizons. The results of this study will provide a basis for estimating OC stocks in the Amazon basin.
We determined stocks of C and N for soils under undisturbed vegetation across the Brazilian Amazon Basin based on 1162 soil profiles of the RADAMBRASIL survey and a digitized Brazilian soil survey map. Mean basin soil C density was 10.3 kg C m-2. Forty-seven petagrams C and 4.4 Pg N were contained in the top 1 m of soil. Forty-five percent of total basin soil C (21 Pg C) and 41% of total soil N (1.8 Pg N) were contained in the top 20 cm across a ≈ 5 000 000-km2 area. Mean C/N ratio for the basin to a depth of 1 m was 10.7. Because these data represent sites with forest vegetation in the absence of significant disturbances, they represent a valuable baseline for evaluating the effects of land-use changes on soil C stocks in the Amazon Basin.
Belgium's soil survey data collected between 1950 and 1970 (pre-Kyoto Protocol) contain more than 13 000 geo-referenced soil profile descriptions, which allow the computation of a spatially distributed baseline carbon content for incremental soil depths, based on soil/land-use combinations (landscape units) and multiple matching soil profile observations. The results show that the soil organic carbon (SOC) and soil inorganic carbon (SIOC) contents of many landscape units do not differ significantly. However, landscape units under forest and grassland tend to contain more carbon. The same is true for landscape units on poorly drained and/or clayey soils, podzols or anthropogenic soils. The change of the SOC in the upper 100 cm of mineral soil follows a logarithmic decline with increasing depth, useful for the extrapolation of SOC of surface layers to deeper layers. SIOC values are strongly related to the geological soil characteristics and increase linearly with depth. Integrating the mean SOC and SIOC content of landscape units over the Belgian territory results in a total soil carbon stock of 303 Mt C in the upper 100 cm layer. Ectorganic horizons contain 35 Mt C and mineral soil contains 245 Mt C in organic form and 23 Mt C in inorganic form. These results are shown to be consistent with an independent set of SOC measurements on 3134 surface samples.