Sampling Designs for Validating Digital Soil Maps: A Review

doi:10.1016/S1002-0160(18)60001-3

Pedosphere

Volume 28, Issue 1, February 2018, Pages 1-15

https://doi.org/10.1016/S1002-0160(18)60001-3 Get rights and content

Abstract

Sampling design (SD) plays a crucial role in providing reliable input for digital soil mapping (DSM) and increasing its efficiency. Sampling design, with a predetermined sample size and consideration of budget and spatial variability, is a selection procedure for identifying a set of sample locations spread over a geographical space or with a good feature space coverage. A good feature space coverage ensures accurate estimation of regression parameters, while spatial coverage contributes to effective spatial interpolation. First, we review several statistical and geometric SDs that mainly optimize the sampling pattern in a geographical space and illustrate the strengths and weaknesses of these SDs by considering spatial coverage, simplicity, accuracy, and efficiency. Furthermore, Latin hypercube sampling, which obtains a full representation of multivariate distribution in geographical space, is described in detail for its development, improvement, and application. In addition, we discuss the fuzzy k-means sampling, response surface sampling, and Kennard-Stone sampling, which optimize sampling patterns in a feature space. We then discuss some practical applications that are mainly addressed by the conditioned Latin hypercube sampling with the flexibility and feasibility of adding multiple optimization criteria. We also discuss different methods of validation, an important stage of DSM, and conclude that an independent dataset selected from the probability sampling is superior for its free model assumptions. For future work, we recommend: 1) exploring SDs with both good spatial coverage and feature space coverage; 2) uncovering the real impacts of an SD on the integral DSM procedure; and 3) testing the feasibility and contribution of SDs in three-dimensional (3D) DSM with variability for multiple layers.

References (127)

T F A Bishop et al.
Validation of digital soil maps at different spatial supports
Geoderma
(2015)
A Biswas
Landscape characteristics influence the spatial pattern of soil water storage: Similarity over times and at depths
Catena
(2014)
A Biswas
Season- and depth-dependent time stability for characterising representative monitoring locations of soil water storage in a hummocky landscape
Catena
(2014)
A Biswas et al.
Scales and locations of time stability of soil water storage in a hummocky landscape
J Hydrol
(2011)
C W Brungard et al.
Machine learning for predicting soil classes in three semi-arid landscapes
Geoderma
(2015)
D J Brus et al.
Random sampling or geostatistical modelling? Choosing between design-based and model-based sampling strategies for soil (with discussion)
Geoderma
(1997)
D J Brus et al.
Designing spatial coverage samples using the fc-means clustering algorithm
D J Brus et al.
Optimization of sample patterns for universal kriging of environmental variables
Geoderma
(2007)
P A Burrough et al.
High-resolution landform classification using fuzzy k-means
Fuzzy Sets Syst
(2000)
A H Cambule et al.
A methodology for digital soil mapping in poorly-accessible areas
Geoderma
(2013)

V Chaplot et al.

Digital mapping of A-horizon thickness using the correlation between various soil properties and soil apparent electrical resistivity

Geoderma

(2010)

D Clifford et al.

Pragmatic soil survey design using flexible Latin hypercube sampling

Comput Geosci

(2014)

D L Corwin et al.

Characterizing soil spatial variability with apparent soil electrical conductivity: I. Survey protocols

Comput Electron Agric

(2005)

P De Zorzi et al.

A soil sampling reference site: The challenge in defining reference material for sampling

Appl Radiat Isot

(2008)

M Duffera et al.

Spatial variability of Southeastern US Coastal Plain soil physical properties: Implications for site-specific management

Geoderma

(2007)

D M Evans et al.

Digital soil mapping of a red clay subsoil covered by loess

Geoderma

(2014)

G J Fitzgerald et al.

Directed sampling using remote sensing with a response surface sampling design for site-specific agriculture

Comput Electron Agri

(2006)

J E Holland et al.

Scoping for scale-dependent relationships between proximal gamma radiometrics and soil properties

Catena

(2017)

F Jonard et al.

Characterization of tillage effects on the spatial variation of soil properties using ground-penetrating radar and electromagnetic induction

Geoderma

(2013)

S B Karunaratne et al.

Catchment scale mapping of measureable soil organic carbon fractions

Geoderma

(2014)

B Kempen et al.

Operationalizing digital soil mapping for nationwide updating of the 1:50 000 soil map of the Netherlands

Geoderma

(2015)

R Kerry et al.

Average variograms to guide soil sampling

Int J Appl Earth Obs Geoinf

(2004)

R Kerry et al.

Disaggregation of legacy soil data using area to point kriging for mapping soil organic carbon at the regional scale

Geoderma

(2012)

D Kidd et al.

Operational sampling challenges to digital soil mapping in Tasmania, Australia

Geoderma Reg

(2015)

M Lacoste et al.

High resolution 3D mapping of soil organic carbon in a heterogeneous agricultural landscape

Geoderma

(2014)

R M Lark

Optimized spatial sampling of soil for estimation of the variogram by maximum likelihood

Geoderma

(2002)

S M Lesch

Sensor-directed response surface sampling designs for characterizing spatial variation in soil properties

Comput Electron Agri

(2005)

M R Levi et al.

Covariate selection with iterative principal component analysis for predicting physical soil properties

Geoderma

(2014)

M Ließ et al.

Uncertainty in the spatial prediction of soil texture: Comparison of regression tree and random forest models

Geoderma

(2012)

B P Louis et al.

Statistical sampling design impact on predictive quality of harmonization functions between soil monitoring networks

Geoderma

(2014)

B P Malone et al.

Empirical estimates of uncertainty for mapping continuous depth functions of soil attributes

Geoderma

(2011)

A B McBratney et al.

An overview of pedometric techniques for use in soil survey

Geoderma

(2000)

A B McBratney et al.

On digital soil mapping

Geoderma

(2003)

D Michot et al.

Digital assessment of soil-salinity dynamics after a major flood in the Niger River valley

Geoderma

(2013)

B Minasny et al.

A conditioned Latin hypercube method for sampling in the presence of ancillary information

Comput Geosci

(2006)

V L Mulder et al.

Representing major soil variability at regional scale by constrained Latin Hypercube Sampling of remote sensing data

Int J Appl Earth Obs Geoinf

(2013)

N P Odgers et al.

Generation of kth-order random toposequences

Comput Geosci

(2008)

N P Odgers et al.

Bottom-up digital soil mapping. I. Soil layer classes

Geoderma

(2011)

N P Odgers et al.

Bottom-up digital soil mapping. II. Soil series classes

Geoderma

(2011)

N P Odgers et al.

Equal-area spline functions applied to a legacy soil database to create weighted-means maps of soil organic carbon at a continental scale

Geoderma

(2012)

D J Pennock et al.

Development and application of landform segmentation procedures

Soil Till Res

(2001)

K Piikki et al.

Sensor data fusion for topsoil clay mapping

Geoderma

(2013)

L Poggio et al.

Regional scale mapping of soil properties and their uncertainty with a large number of satellite-derived covariates

Geoderma

(2013)

C Z Qin et al.

Mapping soil organic matter in small low-relief catchments using fuzzy slope position information

Geoderma

(2012)

L Ramirez-Lopez et al.

Sampling optimal calibration sets in soil infrared spectroscopy

Geoderma

(2014)

J A Royle et al.

An algorithm for the construction of spatial coverage designs with implementation in SPLUS

Comput Geosci

(1998)

K Samyn et al.

Assessment of vulnerability to erosion: Digital mapping of a loess cover thickness and stiffness using spectral analysis of seismic surface-waves

Geoderma

(2012)

K Adhikari et al.

Digital mapping of soil organic carbon contents and stocks in Denmark

PLoS ONE

(2014)

K Adhikari et al.

Digital mapping of topsoil carbon content and changes in the Driftless Area of Wisconsin, USA

Soil Sci Soc Am J

(2015)

S I C Akpa et al.

Digital mapping of soil particle-size fractions for Nigeria

Soil Sci Soc Am J

(2014)

Cited by (66)

The roles of sampling depth, lateral profile density and vertical sampling density in 3D digital soil mapping of soil organic carbon
2024, Geoderma
Sampling density and depth play crucial roles in three-dimensional (3D) soil modeling and prediction, particularly in digital soil mapping (DSM). However, previous studies have yielded inconsistent and even contradictory results regarding impacts of sampling density and depth on the accuracy of 3D DSM. Hence, this study aimed to evaluate the impacts of sampling depth, vertical sampling density and lateral profile density on 3D soil mapping accuracy based on a case study of soil organic carbon (SOC). To achieve this, a comprehensive analysis was conducted based on 511 soil samples collected from 111 profiles in a local hilly area spanning 5.52 km² in China. A 3D regression geostatistical approach was employed for the analysis. The results revealed that samples taken from different depth intervals exhibited varying degrees of importance in relation to prediction accuracy. Notably, a reduced number of surface soil samples (0–0.3 m) led to significant fluctuations in the accuracy of predictions for the entire soil profile. Furthermore, a lower number of subsurface soil samples (0.3–0.6 m) also diminished the overall prediction accuracy, while the influence of deeper soil samples (0.6–1.2 m) on the overall accuracy was relatively less pronounced. Reducing the number of profiles for calibration led to significantly worse and more variable prediction accuracy, compared to reducing vertical samples. For 3D mapping, this study recommends prioritizing the collection of additional lateral profiles, specifically focusing on surface and subsurface soils, especially for properties that display vertical variation similar to SOC.
Preserving soil data privacy with SoilPrint: A unique soil identification system for soil data sharing
2024, Geoderma
Soil is an indispensable resource with critical implications in various fields such as agriculture, environmental science, climate change, hydrology, ecology, and geoscience. Accuracy and accessibility of soil data are crucial for informed decision making. However, the sharing and harmonization of soil data present significant challenges, particularly owing to the lack of a comprehensive identification system that ensures privacy and stewardship in a federated data sharing framework. Moreover, the inherent heterogeneity of soil properties across space and time complicates the establishment of connections between soil profiles and their corresponding properties. To address these challenges, a novel and persistent soil-data identifier, called SoilPrint, akin to a fingerprint, was proposed. SoilPrint utilizes a mathematical algorithm to effectively integrate the properties of soil profile layers (SPLP) with Geohashes, providing an efficient solution. The incorporation of SoilPrint streamlines the data federation process within a secure and distributed ledger, eliminating the need for complex data mapping or alignment. This approach ensures data privacy throughout the sharing process and addresses concerns associated with data management. To demonstrate the practical applications of SoilPrint, a case study using soil data from Ontario, Canada was presented. The results underscored the unique identification capabilities of SoilPrint for soil profiles and their associated properties, establishing it a promising tool for soil data management. SoilPrint facilitates data tracking, reuse, and analysis, thereby enhancing the efficiency and effectiveness of soil-related research and decision-making processes.
An efficient soil moisture sampling scheme for the improvement of remotely sensed soil moisture validation over an agricultural field
2024, Geoderma
An efficient and robust soil moisture (SM) sampling scheme that can capture the spatial variability of SM is required for the accurate calibration and validation of satellite-based SM retrievals. Often, this process requires numerous sampling points, consuming a significant amount of time. Therefore, it is crucial to develop efficient sampling methods for the improvement of satellite-based SM estimations. The objectives of this study were to define an efficient sampling strategy that could be beneficial for the validation of satellite SM estimations; investigate the role of RS covariates in developing such a strategy; and evaluate the performance of the new sampling scheme over various spatial and temporal domains. In this study, we used the conditioned Latin hypercube sampling (cLHS) algorithm to define an efficient sampling strategy. To this end, remote sensing (RS) raster and digital elevation models (DEM) were used to identify numerous environmental covariates to locate sampling points for characterizing spatial variability of SM at the agricultural field scale. A random forest-based technique, the Boruta algorithm, was also applied to select the most important covariates for utilization into the cLHS algorithm. We used the statistical moments (mean and standard deviation, SD) of the field to select the efficient sample size that can best represent SM status in the field. To evaluate the new sampling scheme, a second data set obtained during a different month for the same agricultural field was used. However, because of the potential for high spatial and temporal correlations between training and test covariates when obtained for the same region, we also used different test datasets in New Zealand to evaluate the sampling scheme. Results showed that the RS covariates obtained from SAR and optical imagery were among the most significant covariates for capturing the spatial variability of SM even if they were not acquired on the day of collection. Also, the new sampling scheme could capture the SM spatial pattern of the field for both test datasets with RMSE less than 4% volumetric SM, which is within the range of the expected performance for most satellite SM products. The evaluation of the new sampling scheme on the New Zealand datasets confirmed the functionality of the proposed sampling scheme for a different temporal and spatial domain.
Multi-property digital soil mapping at 30-m spatial resolution down to 1 m using extreme gradient boosting tree model and environmental covariates
2024, Remote Sensing Applications: Society and Environment
Information on the variability of soil properties including clay, calcium carbonate equivalent (CCE) and acidity (pH) in space and depth play major role in sustainable agricultural development but scarce in many regions in the world, specifically in developing countries. Digital soil mapping technique has shown strong success in the prediction of soil properties both in space and depth from their relationship with environmental covariates. The aim of the study was to digitally map soil properties including clay, CCE and pH at multiple depths and identify the dominant factors controlling their variability using Minadoab region of Iran as a case study area due to its significance for agricultural production and economic development in the country. A total of 104 soil profiles were randomly selected and described in field during 15–30 July 2017, and 386 samples were collected from each genetic horizon with variable depths. A continuous spline function was fitted to the horizon data and properties were estimated at five standardized depths following GlobalSoilMap project. The soil layers were then grouped into three classes: i) H1: top (0–5 cm); ii) H2–H4: middle (5–15, 15–30 and 30–60 cm); and iii) H5: bottom (60–100 cm) for easy interpretation. A total of 17 static environmental covariates were derived from terrain-related attributes. Additionally, matching with the soil sampling date, a total of eight band ratios in addition to six individual bands were derived from Landsat-8 OLI (Product ID: LC08_L1TP_168034_20170721_20170728_01_T1) for the study area. Finally, all covariates were categorized in nine groups representing various factors controlling soil variations. The digital maps associated with uncertainties were prepared using extreme gradient boosting (XGBoost) Tree model and bootstrapping method, respectively. Overall, the highest accuracy using XGBoost Tree model was found for CCE (R² = 0.73, on average), followed by pH (R² = 0.72, on average) and clay (R² = 0.70, on average). The highest amount of clay was observed for the bottom layer, followed by middle and top layers across the study area. H1 to H3 layers were subjected to tillage operations over the time and may have contributed to the redistribution of clays. A similar trend was also observed for CCE and pH as the location of the study area is underlain by massive limestone. The uncertainties helped quantify over- and under-estimation of the predictions and the factors with the highest variable importance was identified as the dominant factor. The information generated for the study area provides soil variability information that is critical for developing management strategies for sustainable production in the area.
Airborne laser scanning to optimize the sampling efficiency of a forest management inventory in South-Eastern Germany
2023, Ecological Indicators
Effective forest stewardship relies on comprehensive field-inventories describing forest resources. Increasing demands for data and indicators that improve understanding of climate change impacts, timber production, and ecosystem processes make access to robust field inventories crucial. A trade-off between cost and statistical efficacy exists however, necessitating that practitioners be familiar with the spatial and structural composition and variability of their management areas. Remotely sensed data, like airborne laser scanning (ALS), can improve data availability and sampling efficiency. In this study, we simulate sampling approaches and provide an indication of the benefits of incorporating ALS-derived auxiliary data. We evaluate the ability of sub-samples from an existing field-inventory to accurately estimate ALS structural metrics. Additionally, we explore data-driven approaches to allocate new field plots, reducing bias and improving accuracy. The Monte Carlo simulation compared the local pivotal method (LPM), Latin hypercube sampling (LHS), and simple random sampling (SRS) at a variety of sample sizes. Precision and variability measures were used to comparatively assess the efficacy of sampling method and sample size. Results demonstrate the value of ALS as an auxiliary dataset, with LPM and LHS achieving sampling efficiencies over SRS of up to 88.6% and 94.3%, respectively. By applying the adapted Latin hypercube evaluation of a legacy sample (AHELS) algorithm, we reduced the mean average percent deviation (MAPD) by over 20% between sample measures and wall-to-wall ALS metrics. These methods can aid practitioners in planning cost-effective and statistically rigorous forest inventory campaigns, particularly in determining where to re-sample within an existing plot network.
Spatial prediction and uncertainty estimation of crucial GlobalSoilMap properties - A contextual study in the semi-arid area of western Iran
2023, Geoderma Regional
Information on the spatial distribution of organic carbon (OC), salinity (EC), and soil pH in the semi-arid region of the Chahardowli Plain in western Iran is limited. Though soil carbon is the most common property mapped in GlobalSoilMap, EC, and pH affect OC content and other soil properties and functions. To study variations in these properties and map their distribution, soil samples were collected at depths of 0–5, 5–15, 15–30, 30–60, and 60–100 cm. A total of 145 soil samples were collected from 30 profiles. The relationships between soil characteristics and environmental covariates were modeled using random forest (RF), decision tree (DT), and multiple linear regression (MLR) models. We used a k-fold cross-validation to assess the quality of the predictions. The RF model demonstrated the highest prediction accuracy for all three soil properties. The OC validation results for the RF model show that the R² value was between 0.80 and 0.98, the R² value for EC was between 0.74 and 0.98, and the R² value for pH was between 0.80 and 0.93. The channel network base level (CNBL) was found to be the most crucial covariate in predicting EC, while CNBL and vegetation indices were the most significant covariates in predicting OC. The covariates found to be crucial in predicting pH were the difference in vegetation index (DVI) and slope (S). The bootstrap method was used to compute the prediction uncertainty. The bootstrap method provided reliable estimates of uncertainties associated with these predictions. For all layers and all points, the coverage percentage for OC, EC, and pH was between 80 and 95% at the 95% confidence level. This shows the reliability of the estimated confidence limits. This study indicated that high pH and EC levels are associated with a reduction in soil OC percentage. To provide an accurate representation of OC distribution in any region, it is necessary to report not only an OC map but also maps of EC and pH, as the spatial interrelationship between soil properties highlights the need for estimating pH and EC for better understanding OC variability and soil functioning.

View all citing articles on Scopus

View full text

Sampling Designs for Validating Digital Soil Maps: A Review

Abstract

Geoderma

Catena

Catena

J Hydrol

Geoderma

Geoderma

Geoderma

Fuzzy Sets Syst

Geoderma

Geoderma

Comput Geosci

Comput Electron Agric

Appl Radiat Isot

Geoderma

Geoderma

Comput Electron Agri

Catena

Geoderma

Geoderma

Geoderma

Int J Appl Earth Obs Geoinf

Geoderma

Geoderma Reg

Geoderma

Geoderma

Comput Electron Agri

Geoderma

Geoderma

Geoderma

Geoderma

Geoderma

Geoderma

Geoderma

Comput Geosci

Int J Appl Earth Obs Geoinf

Comput Geosci

Geoderma

Geoderma

Geoderma

Soil Till Res

Geoderma

Geoderma

Geoderma

Geoderma

Comput Geosci

Geoderma

Digital mapping of soil organic carbon contents and stocks in Denmark

PLoS ONE

Digital mapping of topsoil carbon content and changes in the Driftless Area of Wisconsin, USA

Soil Sci Soc Am J

Digital mapping of soil particle-size fractions for Nigeria

Soil Sci Soc Am J