Digital soil mapping

Sustainable development needs soil information at appropriate scales. However, conventional methods of soil survey are slow and expensive, and limit the production of this information in short time, especially in mountain environments. The evolution of geospatial technologies such
as geographic information systems (GIS) and remote sensing, the development of geomorphometric techniques, and the progress of the multivariate statistical methods and
artificial intelligence technologies, provide new opportunities to produce soil information more
efficiently. In order to propose a digital mapping approach as an alternative to conventional soil survey at semidetailed level for mountainous landscapes, we selected a reference area in the Caramacate river basin (Guárico river tributary) on the mountains of North-Central in Aragua
state, Venezuela. To obtain the landscape units underwent quantitative unsupervised classification based on fuzzy Kohonen clustering network (FKCN). Digital elevation models
(DEM) with different spatial resolutions (8, 10, 15 and 20 m) were evaluated to select the 15m
model as the most representative. From this model, 10 digital classes of land surface were
obtained, whose geomorphological meaning was interpreted, base on the spatial distribution,
the class centers and the membership values generated by the neuro-fuzzy classification. A soil sampling comprising 64 sites, was carried out to analyze the structure of spatial variation in soil attributes at various scales of variation. The sampling points were arranged in five nested hierarchical levels including: lithological units, landscape type, land-surface classes, a window 3x3 and in a cell (15 m) spatial resolution,. It was determined that the level that contributed
most to the soil variance was the one corresponding to digital classes of land surface. This confirmed that such classes are appropriate for subsequent soil sampling, and for interpolation and generalization of soil attributes. Four different approaches of spatial inference were applied
for mapping of soil properties from sampling sites. The first is a model of soil-landscape spatial inference combined with theories of fuzzy sets (FCM), the second is based on a multiple linear regression (MLR), the third combines regression analysis multiple linear with interpolation techniques (RK, Regression Kriging), and the fourth combines neuro-fuzzy techniques (FKCN) with residuals interpolation (RKF). Assessing the reliability of the prediction models
determined that the accuracy of predictions is high when most of soil variation occurs at landsurface classes or broader landscape classes, and is low when a short-range high variation predominates. The maps of individual soil properties were integrated using two spatial
inference techniques: the first based on the algorithm of neuro-fuzzy clustering FKCN, and the second based on the application of a soil-landscape model combined with theories of fuzzy sets for a group of pre-established taxonomic categories. Finally, the results obtained are
synthesized as a proposal for digital mapping of soils and landscapes in mountainous areas. The proposal provides a conceptual and methodological framework for generating
information of soils and landscapes in upper watersheds.

Tropical peatlands have an important role in the global carbon cycle. In order to quantify carbon stock for peatland management and conservation, the knowledge of the spatial distribution of peat and its depth is essential. This paper proposed a cost-effective and accurate methodology for mapping peat depth and carbon stocks in Indonesia. The method, based on the scorpan spatial soil prediction function framework, was tested in Ogan Komering Ilir, South Sumatra and Katingan, Central Kalimantan. A peat hydrological unit, where a peatland is bounded by at least two rivers, is defined as the mapping area or extent. Peat depth is modelled as a function of topography and spatial position. Four machine learning models were evaluated to model and map peat depth: Cubist regression tree, Random Forests (RF), Quantile Regression Forests (QRF) and Artificial Neural Network (ANN). Covariates representing topography and spatial position were derived from the 1 arc-second digital elevation model (DEM) of the Shuttle Radar Topography Mission (SRTM) (resolution of 30.7 m). The spatial models were calibrated from field observations. For model calibration and uncertainty analysis, the k-fold cross validation approach was used. Three models: Cubist, Random Forests, and Quantile Regression Forests models showed excellent accuracies of peat depth prediction for both areas where the coefficient of determination values range from 0.67 to 0.92 and root mean squared error (RMSE) values range from 0.6 to 1.1 m. ANN showed inferior results. In addition, QRF and Cubist showed the best account of the uncertainty of prediction, in terms of percentage of observations that fall within the defined 90% confidence interval. In terms of the best predictor, elevation comes first. Using the spatial prediction functions, peat depth maps along with their 90% confidence interval were generated. The estimated mean carbon stock for Ogan Komering Ilir is 0.474 Gt and for Katingan is 0.123 Gt. Our estimate for Ogan Komering Ilir is twice larger than a previous study because we mapped the peatland hydrological unit, while the previous study only delineated peat domes. Finally, we recommend a sampling method for peat depth mapping using numerical stratification of elevation to cover both the geographical and covariate space. We expect that the combination of an improved sampling strategy, machine learning models, and kriging will increase the accuracy of peat depth mapping.

Digital soil mapping (DSM) is a successful sub discipline of soil science with an active research output. The success of digital soil mapping is a confluence of several factors in the beginning of 2000 including the increased availability of spatial data (digital elevation model, satellite imagery), the availability of computing power for processing data, the development of data-mining tools and GIS, and numerous applications beyond geostatistics. In addition, there was an increased global demand for spatial data including uncertainty assessments, and a rejuvenation of many soil survey and university centres which helped in the spreading of digital soil mapping technologies and knowledge. The theoretical framework for digital soil mapping was formalised in a 2003 paper in Geoderma. In this paper, we define what constitutes digital soil mapping, sketch a brief history of it, and discuss some lessons. Digital soil mapping requires three components: the input in the form of field and laboratory observational methods, the process used in terms of spatial and non-spatial soil inference systems, and the output in the form of spatial soil information systems, which includes outputs in the form of rasters of prediction along with the uncertainty of prediction. We also illustrate the history with a number of sleeping beauty papers that seem too precocious and consequently the ideas were not taken up by contemporaries and largely forgotten. It took another 30 to 40 years before the ideas were rediscovered and then flourished. Examples include proximal soil sensing that was developed in the 1920s, soil spectroscopy in 1970s, and soil mapping based on similarity of environmental factors in 1979. In summary, the coming together of emerging topics and timeliness greatly assists in the development of paradigm. We learned that research and ideas that are too precocious are largely ignored — such work warrants (re)discovery.

There is a growing need for spatially continuous and quantitative soil information for environmental modeling and management, especially at the national scale. This study was aimed at predicting soil particle-size fractions (PSF) for Nigeria using random forest model (RFM). Equal-area quadratic splines were fitted to Nigerian legacy soil profile data to estimate PSFs at six standard soil depths (0–5, 5–15, 15–30, 30–60, 60–100, and 100–200 cm) using the GlobalSoilMap project specification. We applied an additive log-ratio (ALR) transformation of the PSFs. There was a better prediction performance (based on 33% model validation) in the upper depth intervals than the lower depth intervals (e.g., R2 of 0.53; RMSE of 13.59 g kg−1 for clay at 0–5 cm and R2 of 0.16; RMSE of 15.60 g kg−1 at 100–200 cm). Overall, the PSFs show marked variations across the entire Nigeria region with a higher sand content compared with silt and clay contents and increasing clay content with soil depth. The variation in soil texture (ST) shows a progressive transition from a coarse texture (sand) along the fringes of northern Nigeria (e.g., upper part of Maiduguri and Sokoto), to finer texture (loam to clay loam) toward the western part of the Niger Delta region in the south. The inclusion of depth as a predictor variable significantly improved the prediction accuracy of RFM especially at lower depth intervals. These results could be used for producing soil function maps for national agricultural planning and in assessments of environmental sustainability.

After decades of mining and industrialization in Qatar, it is important to estimate their impact on soil pollution with toxic metals. The study utilized 300 topsoil (0–30 cm) samples, multi-spectral images (Landsat 8), spectral indices and environmental variables to model and map the spatial distribution of arsenic (As), chromium (Cr), nickel (Ni), copper (Cu), lead (Pb) and zinc (Zn) in Qatari soils. The prediction model used condition-based rules generated in the Cubist tool. In terms of R 2 and the ratio of performance to interquartile distance (RPIQ), the models showed good predictive capabilities for all elements. Of all of the prediction results, Cu had the highest R 2 = 0.74, followed by As > Pb > Cr > Zn > Ni. This study found that all of the models only chose images from January and February as predictors, which indicates that images from these two months are important for soil toxic metals' monitoring in arid soils, due to the climate and the vegetation cover during this season. Topsoil maps of the six toxic metals were generated. The maps can be used to prioritize the choice of remediation measures and can be applied to other arid areas of similar environmental/socio-economic conditions and pollution causes.

An assessment of the field-scale variation of apparent electrical conductivity (ECa) survey and the characterization of correlated soil property was investigated for mapping of a terrace and a floodplain soil of Bangladesh. The electromagnetic induction (EMI) technique was applied by a soil sensor, EM38 which provide ancillary ECa data sets accurately. The survey was supported by soil sampling to assure the reliability and to make evident the potential of ECa measurements for soil mapping. The soils of the study sites greatly differed, the terrace site consisted of a shallow depth clay substratum and the floodplain site bears a sandy substratum. The ECa readings in mS m-1 ranged from 40 to 64 and 32 to 53 in the vertical (ECa-V) and horizontal (ECa-H) orientation of measurement respectively in the terrace site. In the floodplain site, ECa readings in mS m-1 ranged from 35.1 to 50.8 and 33.8 to 47.4 for the ECa-H and ECa-V respectively. In the terrace site, the ECa readings correlated best with soil property such as top-, sub-, and deepsoil texture (clay and sand), and topsoil chemical property, i. e., pH, CEC, Ca2+ and Mg2+. For the floodplain site, the ECa readings correlated best with top- and deepsoil texture, and topsoil organic C. A modest correlation was found between ECa-V and the subsoil clay (r = 0.78), and ECa-V and the subsoil sand (r = - 0.84) in the terrace site. In the floodplain site, the correlation between ECa-H and topsoil clay was the highest (r = 0.66). The variogram analysis revealed that a large portion of the total variation of soil property (about 70 %) was accounted by the spatially structured component of the variogram. The study findings have brought an expectation that soil mapping through ECa measurement is possible in Bangladesh. For mapping, the ECa-V measurement in the terrace soil is more predictive than ECa-H while the ECa-H measurement in the floodplain soil is more predictive than ECa-V. The maps of ECa can fairly represent the spatial variation of soil properties. Thus provide useful information on soil texture, chemical fertility and organic matter content. The ECa map also provides a means of monitoring the spatial variation of soil properties that potentially influence the crop production. The ECa maps can also guide directed soil sampling with the purpose of updating the existing soil maps of Bangladesh.

... decline in soil mineralogy, soil morphology and soil genesis research in comparison to a ... part of an ecosystem interacting with environmental factors generating complex patterns and processes ... consider broader issues such as soil and water quality, food security, carbon cycling ...

Lateritic soils of  Mathamangalam, Kannur
District, located in midlands of Kerala, were morphologically
studied, characterized, classified and mapped
at 1:50,000 scale using remote sensing techniques. The
terrain of the study area being hilly and covered with
perennial vegetation, soil-landscape model was applied.
For this purpose physiographic information was inferred
from SRTM DEM, Resourcesat-1 LISS-III
satellite image and topographical maps. The interpreted
units were validated in the field and characterized
through soil-site examination, soil profile study and soil
analysis. The study indicated that the lateritic soils of
midlands of Kerala vary in physical, chemical and
morphological properties in relation to micro-relief.
Soils developed on moderately steeply sloping side
slopes (15–30% slope) are deep, moderately well
drained with gravelly clay textured, where as the soils
developed on moderately slopping side slope (10–15%
slope) are very deep and well drained. The soils of
valleys are very deep, moderately well drained with fine
texture. Very gently sloping (1–3%) laterite plateau tops
have extremely shallow soils associated with rock
outcrops. These soils mainly belong to Order Ultisols
followed by Inceptisols and Entisols. These were further
grouped up to Family and Series level by tentatively
establishing seven soil series. This study helps in
understanding the behaviour of lateritic soils of midlands
of Kerala, which can be useful in generation of
interpretative maps and in optimizing the land use.

A B S T R A C T Soil organic carbon (SOC) plays a crucial role in maintaining fertility and productivity in sandy soils. This study mapped the spatial variability of SOC concentration, A-horizon thickness, and SOC stocks from the Central Sands in Wisconsin. Soil samples were collected from three different areas (area A, B, and C) that were sampled through grid sampling (GS, n = 100), conditioned latin hypercube sampling (cLHS, n = 100), and random sampling (RS, n = 150) schemes. Average SOC concentration of the A-horizon from soil sampling area A, B, and C were 6.1, 7.1, 8.3 g kg − 1 , respectively. The mean A-horizon thickness for agricultural soils was 28 cm compared to 15 cm under adjacent grassland. Regression kriging was selected as prediction model where EC a and local topographic information (i.e., slope gradient, slope aspect, elevation, wetness index, altitude above channel network etc.) were used as predictors. We observed an increased SOC content, SOC stock, and A-horizon thickness with EC a and wetness index. SOC from area B had the strongest spatial dependency (NSR = 0.64) followed by area A (NSR = 0.72), whereas that from area C was the weakest (NSR = 0.78). Compared to SOC content and A-horizon thickness prediction, SOC stocks prediction had the maximum uncertainty. Predicted SOC stock (t ha − 1) ranged from 28 for sampling area A to 40 for B, and 59 for area C. These high SOC stocks are the result of decade long intensive agriculture with high amount of nitrogen input and irrigation. It has resulted in deep A-horizon and high SOC stocks. This study found that SOC stocks in the Central Sands could be estimated from A-horizon thickness (R 2 ~0.5).

An assessment of the field-scale variation of apparent electrical conductivity (ECa) survey and the characterization of correlated soil property was investigated for mapping of a terrace and a floodplain soil of Bangladesh. The electromagnetic induction (EMI) technique was applied by a soil sensor, EM38 which provide ancillary ECa data sets accurately. The survey was supported by soil sampling to assure the reliability and to make evident the potential of ECa measurements for soil mapping. The soils of the study sites greatly differed, the terrace site consisted of a shallow depth clay substratum and the floodplain site bears a sandy substratum. The ECa readings in mS m-1 ranged from 40 to 64 and 32 to 53 in the vertical (ECa-V) and horizontal (ECa-H) orientation of measurement respectively in the terrace site. In the floodplain site, ECa readings in mS m-1 ranged from 35.1 to 50.8 and 33.8 to 47.4 for the ECa-H and ECa-V respectively. In the terrace site, the ECa readings correlated best with soil property such as top-, sub-, and deepsoil texture (clay and sand), and topsoil chemical property, i. e., pH, CEC, Ca2+ and Mg2+. For the floodplain site, the ECa readings correlated best with top- and deepsoil texture, and topsoil organic C. A modest correlation was found between ECa-V and the subsoil clay (r = 0.78), and ECa-V and the subsoil sand (r = - 0.84) in the terrace site. In the floodplain site, the correlation between ECa-H and topsoil clay was the highest (r = 0.66). The variogram analysis revealed that a large portion of the total variation of soil property (about 70 %) was accounted by the spatially structured component of the variogram. The study findings have brought an expectation that soil mapping through ECa measurement is possible in Bangladesh. For mapping, the ECa-V measurement in the terrace soil is more predictive than ECa-H while the ECa-H measurement in the floodplain soil is more predictive than ECa-V. The maps of ECa can fairly represent the spatial variation of soil properties. Thus provide useful information on soil texture, chemical fertility and organic matter content. The ECa map also provides a means of monitoring the spatial variation of soil properties that potentially influence the crop production. The ECa maps can also guide directed soil sampling with the purpose of updating the existing soil maps of Bangladesh.

In the context of sustainable development of African natural resources, the preservation and sustainable management of soil resources are more important than anything else. It is not possible to deal with soil degradation and fertility problems without having the fundamental knowledge base about the variation and distributions of African soils. The modern research in the field of soil resources sustainability need, not only an available hard copy maps but also, an up-to-date digital soil maps. Consequently, the main purpose of this paper is to review and assessment the most recent efforts of surveying and producing a digital map for African soils. The second goal of this study is to facilitate that in fulfillment one of the missions of the Department of Natural Resources, Institute of African Research and Studies to build and own a comprehensive up-to-date Geo-database for African soil resources and their potential capabilities for sustainable development.
Two major project for introducing digital maps for African soils; AfSIS (Africa Soil Information Service) dataset and maps (2013) & Soil Atlas of Africa (Jones et al., 2013) by The Joint Research Centre (JRC) of the European Commission; were investigated, criticized and compared to each other as well as to some previous studies of some African soils related to the author. Then, the most appropriate data were selected and processed by using ESRI ArcGIS 10.x to produce a set of ESRI shapefiles for Soil orders, soil fertility, soil physical chemical properties, and soil degradation at 1 km resolution using an automated mapping framework (3D regression kriging). The created soil degradation map was used to highlight the most affected area that need to get more concentration in the upcoming conservation plans.

This paper presents national approaches to design soil databases in four countries of Southeast Europe. Ways of data collecting are exposed and the basic sources of the soil information are listed. Numerous references of publications provide evidence of the opportunities to use the soil databases in each country for research, as well as their applicability to data harmonization and integration. (http://agrosoil.yolasite.com/resources/2015_AiG_83_pp_5-17_En.pdf)

This article presents the steps taken to ensure the consistency and usability of the geographic information system of soil resources of Romania SIGSTAR-200 within INSPIRE, the Infrastructure for Spatial Information of the European Community. At present, SIGSTAR-200 contains tens of thousands of polygons characterized by three attributes collected from legacy paper maps, namely (i) soil mapping unit (SMU), including soil association, (ii) topsoil texture class and (iii) skeleton class, all defined as stated by the national methodologies. In addition, each soil is characterized concerning the risk of land degradation (by water erosion, wind erosion, salinization, alkalization, gleyzation and waterlogging) through attributes inferred by expert system rules built on pedogenesis. To achieve the compatibility and usability within INSPIRE, SIGSTAR-200 has been transformed in accordance with the common Implementing Rules in force. To this end, first, the SMUs have been correlated at the dominant soil type level with the international soil classification system World Reference Base of Soil Resources (WRB) to ensure the semantic interoperability. Second, the SMUs, modeled as INSPIRE feature type SoilBody, have been populated with new attributes. This step was performed for SIGSTAR-200 dataset in three coordinate reference systems (CRSs): EPSG:3844 (CRS of Romania), EPSG:3035 and EPSG: 4258 (the last two CRSs being required or recommended by INSPIRE). Finally, the data transformed in INSPIRE-compliant GML have been checked by the INSPIRE Validator (Executable Test Framework), passing all the tests currently available for soil datasets, i.e., regarding (i) data consistency, (ii) INSPIRE GML application schemas, (iii) information accessibility, and (iv) reference systems. The next steps of the work, synchronized with the availability of the Executable Test Suites for the themes defined in Annex III of the Directive, aim to fully validate SIGSTAR-200 GML and to conclude its first integration into INSPIRE (Accession Number: WOS: 000526176700012).

It is critical to produce more crop per drop in an environment where water availability is decreasing and competition for water is increasing. In order to build such agricultural production systems, well parameterized crop growth models are essential. While in most crop growth modeling research, focus is on gathering model inputs such as climate data, less emphasis is paid to collecting the critical soil hydraulic properties (SHPs) data needed to operate crop growth models. Collection of SHPs data for the Zambezi River Basin (ZRB) is extremely labor-intensive and expensive, thus alternate technologies such as digital soil mapping (DSM) must be explored. We evaluated five types of DSM models to establish the best spatially explicit estimates of the soil water content at pF0.0 (saturation), pF2.0 (field capacity), and pF4.2 (wilting point), and of the saturated hydraulic conductivity (Ksat) across the ZRB by using estimates of locally calibrated pedotransfer functions of 1481 locations for training and testing the DSM models, as well as a reference dataset of measurements from 174 locations for validating the DSM models. We produced coverages of environmental covariates from various source datasets, including climate variables, soil and land use maps, parent materials and lithologic units, derivatives of a digital elevation model (DEM), and Landsat imagery with a spatial resolution of 90 m. The five types of models included multiple linear regression and four machine learning techniques: artificial neural network, gradient boosted regression trees, random forest, and support vector machine. Where the residuals of the initial DSM models were spatially autocorrelated, the models were extended/complemented with residual kriging (RK). Spatial autocorrelation in the model residuals was observed for all five models of each of the three water contents, but not for Ksat. On average for the water content, the R 2 ranged from 0.40 to 0.80 in training and test datasets before adding kriged model residuals and ranged from 0.80 to 0.95 after adding model residuals. Overall, the best prediction method consisted of random forest as the deterministic model, complemented with RK, whereby soil texture followed by climate and topographic elevation variables were the most important covariates. The resulting maps are a ready-to-use resource for hydrologists and crop modelers to aliment and calibrate their hydrological and crop growth models.

O uso cada vez mais intensivo do solo e sua ocupação
desordenada resultam em sua degradação. Existe uma
grande carência de informações mais detalhadas de solos
e seus levantamentos. Por apresentarem informações
básicas sobre esse recurso natural e sua distribuição na
paisagem, os mapeamentos são fundamentais para o
manejo sustentável das terras. A produção de mapas de
classes e atributos, com maior nível de detalhamento, só
será possível com a utilização de geotecnologias
combinadas com dados coletados no campo. Assim, o
objetivo desta revisão foi ressaltar a importância dos
levantamentos de solos dentro de um novo paradigma,
definido como Mapeamento Digital de Solos (MDS). A
prática de MDS requer que: i) os dados utilizados tenham
sua origem em coletas de campo e resultados analíticos de
laboratório, incluindo dados legados e novas coletas; ii) o
processo de inferência e espacialização inclui a
proposição de modelos matemáticos e estatísticos entre as
observações do solo, as covariáveis ambientais e os
demais fatores scorpan; e, iii) os resultados estejam na
forma de um sistema espacial em solos, com dados
contínuos e/ou discretos, agregados à incerteza dessas
informações, podendo ser atualizados sempre que novas
observações estiverem disponíveis. A aplicação do MDS
irá permitir que informações espaciais em solos, de
atributos e classes, sejam geradas e prontamente
disponibilizadas em formatos digitais e com a incerteza
associada. Entre os desafios para uma maior aplicação das
várias estratégias disponíveis em MDS está a capacitação
e formação das futuras gerações de pedólogos.

About National Centre for Soil Mapping, Italy
Guidelines of the methods of soil survey and data informatization
SISI Compliancy with INSPIRE and IUSS model
Sampling and analytical procedures
Monitored environments
Main Parameters
Geographical distribution of observations
Soil monitoring in decades
The soil samples archive
The spectral library
Collected metadata
Soil properties prediction of the main Reference Soil Groups
Open questions
CREA Open Data policy
Aknowledgments

An increasing demand for full spatio-temporal coverage of soil information drives the growing use of soil spectroscopy. Soil spectroscopy application performed under laboratory conditions or in-field studies in semi-arid areas have shown promising results. However, when acquiring data in temperate zones, limitations by vegetation-free coverage, variation in soil moisture and management are driving coherent spatio-temporal data collection. This study explores the use of multi-temporal imaging spectroscopy data to increase the total mapping area of bare soils in a heterogeneous agricultural landscape. Spectrally and spatially high-resolution data from the Airborne Prism Experiment (APEX) were collected in September 2013, April 2014 and April 2015. Bare soils in all acquisitions were identified. To eliminate short-term differences in soil moisture and soil surface roughness, the empirical line method was used to calibrate the reflectance values of the singular images (2013 and 2015) towards the singular image with most bare soil pixels (2014). Difference indicators show that the calibration was successful (decrease in root mean square difference and angle difference, increase in R 2 and gain and offset close to one and zero). Finally, the multi-temporal composite image contained more than double the amount of bare soil pixels as compared to a singular acquisition. Summary statistics show that reflectance values of the multi-temporal composite approximate the single image data of 2014 (mean and standard deviation of 2014: 24.2 ± 8.9 vs. 24.0 ± 9.5 for the multi-temporal composite of 2013, 2014 and 2015). This indicates that global differences in soil moisture and land management have been corrected for. As a result, an improved spatial representation of soil parameters can be retrieved from the composite data. Spatial distribution of the correction factors and analysis of the spatial variability of all images, however, indicate that non-linear, short-term differences like variation in soil moisture and land management largely influence the result of the multi-temporal composite. Quantification and attribution of those factors will be required in the future to allow correcting for them.