Is Horizon Sampling More Powerful Than Depth Sampling

Advances in Agronomy, 2020

Soil horizons reflect soil processes and convey information about past and present soil conditions. The identification and delineation of soil horizons are affected by lateral and vertical variation in soil properties. Early studies focused on the variation of horizon thickness and the waviness of horizon boundaries, but did not consider within-horizon lateral and vertical variation. Here we review studies that investigated variation in the master horizons O, A, E, B, and C. We summarize what is known about soil horizon variation, quantify the variation in different horizons, and investigate whether the variation increases or decreases with depth. The variation within horizons differs among soils, and the magnitude of the variation varies for different soil properties. Variation within soil horizons or laterally within a few square meters may be considerable, and the within-horizon variation changes with depth. Horizon thickness does not seem to be related to the variation of soil chemical and physical properties within the horizon, i.e., thicker horizons do not necessarily have higher variation in their soil properties. Three case studies are presented: Spodosols and Histosols (Russia), Alfisol and Mollisol (USA), and Oxisol (Brazil). Factors that affect the within-horizon variations include landscape position, parent material, vegetation, fertilization, tillage, drainage, and time. The vertical distribution of soil properties can be quantified using soil depth functions. Digital soil morphometrics techniques can assist in the quantification of two-dimensional soil profile properties and variations.

Digital Soil Assessment and Beyond

While many studies have mapped the thickness of individual soils horizons, few have mapped the horizons as a whole soil profile. We developed soil-landscape regression models to describe and predict the occurrence and the thickness of several soil horizons to 1 m depth. The study was carried out at a 75 km 2 area in the Hunter Valley, New South Wales in Australia, using 1050 soil profiles observations. The landscape factors that were used to make the models included terrain attributes, landuse, and geology. We derived regression models to predict the thickness of the individual soil horizons. Because not all horizons can be present in the whole area, we cannot develop a regression model which has the assumption of constant error variance. Therefore, we used a combination of a logistic regression with an ordinary regression to first model the occurrence of each horizon and then their thickness, respectively. This model revealed significant relations between the soil attributes and the prediction of the occurrence and thickness of each of the soil horizons.

Geoderma, 2017

Soil depth has played a key role in the development of soil survey, implementation of soil-specific management and validation of hydrological models. Generally, soil depth at field scale is difficult to map due to complex interactions of factors of soil formation at field scale. As a result, the conventional sampling schemes to map soil depth are generally laborious, time consuming and expensive. In this study, we presented, tested and evaluated a method to optimize the sampling scheme to map soil depth to petrocalcic horizon at field scale. The method was tested with real data at four agricultural fields localized in the southeast Pampas plain of Argentina. The purpose of the method was to minimize the sample dataset size to map soil depth to petrocalcic horizon based on ordinary cokriging, five calibration sample sizes (returned by Conditioned Latin hypercube-cLHS-), and apparent electrical conductivity (ECa) or elevation as variables of auxiliary information. The results suggest that (i) only 30% of samples collected on a 30-m grid are required to provide high prediction accuracy (R 2 N 0.95) to map soil depth to petrocalcic horizon; (ii) an independent validation dataset based on 50% of the samples on a 30-m grid is adequate to validate the most realistic accuracy estimate; and (iii) ECa and elevation, as variables of auxiliary information, are sufficient to map soil depth to petrocalcic horizon. The method proposed provides a significant improvement over conventional to map soil depth and allows reducing cost, time and field labour. Extrapolation of the results to other areas needs to be tested.

Journal of Environment and Earth Science, 2014

This study attempts to evaluate some interpolation techniques for mapping spatial distribution of A horizon depth and OM in Shahrekord, Iran. 15000 hectares of South West Shahrekord soils were studied in which totally 92 soil profiles were excavated and classified according to USDA. The performance of methods was evaluated by RMSE, ME and R 2. Calculated RMSE for depth of A horizon were 0.01074, 0.19670 and 0.19858, respectively by IDW and OK (with Spherical and Exponential models). The RMSE for surface horizon OM were obtained 0.05593, 0.12121 and 0.05078, respectively by IDW and OK (with Spherical and Exponential models). The results showed that IDW could estimate the variability of A horizon depth and Ok (with Exponential semivariogram) could estimate the variability of depth of A horizon more better than other methods. The weakness of kriging in prediction of spatial continuity of depth of A horizon is due to effects of variability of soil forming factors in evolution of soils evolved in different landforms of study area which could take out the stationary assumptions.

Geoderma, 2010

Conventional soil science methods for the estimation of the spatial variations of soil properties within landscapes are destructive, time consuming, and do not allow the estimation of the short range variability. Recent advances in geomatic global positioning systems and sensors offer new possibilities for the mapping of spatially varying soil patterns. Although geophysical techniques offer an alternative to traditional soil sampling methods, the resulting data are still often misinterpreted, especially in terms of the interrelationships between, for example, soil apparent electrical resistivity (ρ) and several soil physical or chemical properties. Our main objective was to test the suitability of using ρ measurements for mapping the thickness of the organo-mineral A-horizon (ThickA). ρ was mapped (0-0.5 m), using an RM15, in a highly water eroded field plot of 300 m² showing large variations in ThickA, but no significant variations in soil clay and nutrient content. The correlations between ρ and ThickA, the top-soil water content (θ 0.1 ), and the bulk density (BD 0.1 ) were determined at the nodes of a regular grid of 2 m (n = 96). Principal Component Analysis (PCA) was used to identify the impact of the different soil properties on the ρ data. ThickA varied between 0.17 and 0.3 m, and ρ between 1046 and 3864 Ω m. There was a significant correlation between ρ and ThickA (r = 0.56) and also between θ 0.1 (r = 0.23) and BD 0.1 (r = 0.22). If it was assumed that θ 0.1 and BD 0.1 were constant over the study plot then there was a correlation between ρ and ThickA with an r of 0.88, thus making possible the accurate mapping of ThickA (MAE = 0.017 m; i.e., 7% of the average value of ThickA). Prediction (MAE = 0.015 m) was only slightly improved through the use of regression kriging of ThickA with ρ recalculated as the covariate. This result showed that apparent soil electrical resistivity measurements could be very successful for estimating A-horizon thickness provided that limited information on top-soil characteristics is included in the mapping process. Geoderma j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / g e o d e r m a 157 V. Chaplot et al. / Geoderma 157 (2010) 154-164

Geoderma, 2010

Soil variability across landscapes is well known and results from the combination of geomorphologic and pedogenetic processes. Despite its importance, little quantitative information exists on the horizontal and vertical variation of soil profiles, especially in natural landscapes that are unaffected by soil erosion. This study aims at measuring the variation in soil horizon depth in loess-derived soils and relating it to surface characteristics. Several terrain variables and the variable soil type, as derived from existing soil maps, are considered in the prediction model.In total 399 augerings of up to 8.7 m deep were made in natural forest areas in Central Belgium, where soil profiles were not affected by anthropogenic soil erosion. The variability of 5 different soils horizons was evaluated: eluvial (E) and illuvial clay (Bt) horizons, transition horizon (BC), decalcified loess material (C1) and undisturbed calcareous loess material (C2). All horizons exhibited significant variability. The top two soil horizons could not be linked significantly to any of the predictor variable. For the lower three soil horizons some weak, yet significant relations were found with the predictor variables slope gradient, plan curvature, wetness index, landform and soil type. Geostatistical analysis indicated a lack of spatial dependence for all horizons, except for the upper eluvial E horizon. This high spatial randomness resulted in poor predictions models for the depth of these soil horizons, with model efficiencies ranging between − 0.14 and 0.08, while for the E horizon, a simple ordinary kriging model provided a model efficiency of 0.43.

Geoderma, 2005

Soils and underlying parent materials form a continuous system we must understand and manage in total. Numerous concerns (e.g., water quality, on-site waste disposal, landfill placement, and nutrient or pesticide movement) require an integrated knowledge and understanding of soil, the soil-to-substratum transition, and the deeper substratum. Soil C-horizons can exceed the thickness of the overlying A and B-horizons and contain unique morphological properties. The subsolum including C-horizons receives less descriptive emphasis than upper soil horizons. Soil scientists map and classify soils mainly on A and B-horizon properties. Soil forming and hydrologic processes that impart morphological features, however, extend considerably below these horizons. Precise adherence to Soil Taxonomy places an arbitrary constraint on field observations at 2 m. Soil scientists routinely observe C and R horizons and deeper underlying substrata in gravel pits, road cuts, barrow pits, foundation excavations, and drill cores, but provide less documentation than for upper horizons. Parent material and stratigraphy need more consideration in soil map unit design and delineation. Field observations by soil scientists below 2 m are crucial for understanding the subsolum (i.e., the morphology of, and relationships of solum to substratum). Soil surveys can convey concise and more descriptive soil-to-substrata information with little added effort or resources. Soil surveys can accomplish this end by use of block diagrams, parent material maps, and geomorphic maps that include both pedostratigraphic and lithostratigraphic detail. Soil surveys must develop soil and map unit descriptions linked to measured sections and named stratigraphic units, and describe and analyze soils and parent materials to greater depths (N2 m). We use case examples to demonstrate these concepts. Soil-to-substrata documentation and presentation conveys crucial information to soil survey users. Soil-to-substrata relationships identified and recorded during a soil survey create a knowledge window to the subsurface.

Geoderma, 2008

This paper aims to investigate the potential of using soil-landscape pattern extracted from a soil map to predict soil distribution at unvisited location. Recent machine learning advances used in previous studies showed that the knowledge embedded within soil units delineated by experts can be retrieved and explicitly formulated from environmental data layers However, the extent to which the models can yield valid prediction has been little studied. Our approach is based on a classification tree analysis which has underwent a recent statistics advance, namely, stochastic gradient boosting. We used an existing soil-landscape map to test our methodology. Explanatory variables included classical terrain factors (elevation, slope, curvature plan and profile, wetness index, etc.), various channels and combinations of channels from LANDSAT ETM imagery, land cover and lithology maps. Overall classification accuracy indexes were calculated under two validation schemes, either taken within the training area or from a separated validation area. We focused our study on the accuracy assessment and testing of two modelling parameters: sampling intensity and spatial context integration. First, we observed strong differences in accuracy between the training area and the extrapolated area. Second, sampling intensity, in proportion to the class extent, did not largely influence the classification accuracy. Spatial context integration by the use of a mean filtering algorithm on explanatory variables increased the Kappa index on the extrapolated area by more than ten points. The best accuracy measurements were obtained for a combination of the raw explanatory dataset with the filtered dataset representing regional trend. However, the predictive capacity of models remained quite low when extrapolated to an independent validation area. Nevertheless, this study offers encouragement for the success of extrapolating soil patterns from existing soil maps to fill the gaps in present soil map coverage and to increase efficiency of ongoing soil survey.

A B S T R A C T Three soil profiles in Wisconsin, USA, were sampled using a 10 × 10 cm raster: a Mollisol (1 × 1 m), Alfisol (1 × 1 m), and Entisol (1 × 0.5 m). The soils were described in the field, and samples were taken from the center of each cell. Soil organic carbon concentration, texture, and color were measured and used to revise field-delineated horizons and their boundaries. Using soil texture, an Eb horizon was identified on the raster maps in the upper part of the field-delineated Btb horizon of the Mollisol. Soil color, soil texture, and Ti showed little lateral variation. The pH tended to vary the most laterally. The raster method characterizes soil profiles in two dimensions and can be used to quantify lateral variation and improve field delineation of soil horizons.