Vote for the Best Paper in Pedometrics 2016

D G Rossiter, Chairman Pedometrics Awards Committee

Pedometrics commission of the International Union of Soil Sciences

e-mail: dgr2@cornell.edu                  04-May-2017

 

Dear fellow Pedometricians,

The Pedometrics Awards committee for the best paper award (Grunwald, McBratney, Oliver, Rossiter, Yang) received a strong response to our call for nominations: 26 papers spread over sixteen journals. These were scored by the committee. Because of the large number of excellent submissions we’ve decided to present the top eight papers for your reading pleasure and evaluation. These are a good mix of pedometrics: novel and low-cost proximal sensing, a global spectral library, landscape complexity using spatial adjacency graphs, Bayesian spatial modelling, boundary-line analysis, structural equation modelling for digital soil mapping, and sampling optimization. Reading these papers will bring you up-to-date on some of the most exciting developments in pedometrics published in 2016.

Both the 2015 and this 2016 awards will be presented at Pedometrics 2017 (25th anniversary of the first Pedometrics conference) in Wageningen (NL) 26 June through 1 July 2017 (see information at http://www.pedometrics2017.org).Please send in your votes for the best paper 2016 by 15-June-2017.

Please rank the papers in the “instant runoff” system: first choice, second choice, etc. up till the last paper you are willing to vote for, i.e., the last paper that you think would deserve the award. Votes should then be sent to me from a traceable e-mail address (to prevent over-voting). I will apply the instant runoff system to determine the winner. A co-author may not vote for her/his own paper(s). What defines “best”? It’s up to you to decide, but I think the best paper should be the one that most advances pedometrics.

 

The papers and their abstracts are listed here in order of DOI.

  • Lark, R. M., & Milne, A. E. (2016). Boundary line analysis of the effect of water-filled pore space on nitrous oxide emission from cores of arable soil. European Journal of Soil Science, 67(2), 148–159. https://doi.org/10.1111/ejss.12318

The boundary line has been proposed as a model of the effects of a variable on a biological response, when this variable might limit the response in only some of a set of observations. It is proposed that the upper boundary (in some circumstances the lower boundary) represents the response function of interest. Boundary-line analysis is a method for estimating this response function from data. The approach has been used to model the emission of N2O from soil in response to various soil properties. However, the methods that have been used to identify the boundary are based on somewhat ad hoc partitions of the data. A statistical model that we have presented previously has not been applied to this problem in soil science, and we do so here to represent how the water-filled pore space (WFPS) of the soil affects the rate of N2O emission. We derive a boundary-line response that can be shown to be a better model for the data than an unbounded alternative by statistical criteria. Furthermore, the fitted boundary-response model is consistent with past empirical observations and modelling studies with respect to both the WFPS at which the potential emission rate is largest and the measurement error for the emission rates themselves. We show how the fitted model might be used to interpret data on soil volumetric water content with respect to seasonal changes in potential emissions, and to compare potential emissions between soil series that have contrasting physical properties.

  • Lobsey, C. R., & Viscarra Rossel, R. A. (2016). Sensing of soil bulk density for more accurate carbon accounting. European Journal of Soil Science, 67(4), 504–513. https://doi.org/10.1111/ejss.12355

Measurements of soil bulk density can aid our understanding of soil functions and the effects of land use and climate change on soil organic carbon (C) stocks. Current methods for measuring bulk density are laborious and expensive, subject to errors and complicated by the need to measure below the soil surface. These shortcomings are emphasized when there is need to characterize the spatial (lateral and vertical) and temporal variation of soil bulk density and related properties. We developed a technique that combines gamma-ray attenuation and visible–near infrared (vis–NIR) spectroscopy to measure ex situ the bulk density of 1-m soil cores that are sampled freshly, wet and under field conditions. We found that the accuracy of the sensor measurements was similar to that of the conventional single-ring method, but sensing is rapid, inexpensive, non-destructive and practical. Sensing can be used to measure many soil cores efficiently at fine depth resolutions (e.g. every 2 cm along the core), thereby allowing effective characterization of spatial variation in both lateral and vertical directions. The measurements can be made in the field, on wet soil cores, which reduces the costs and errors associated with transport, handling, oven-drying and laboratory measurements. We show that sensing of bulk density can be used to measure organic C stocks on either a fixed-depth (FD) or cumulative soil mass (CSM) basis. Our sensing approach to measure bulk density meets all the requirements for inclusion in a well-designed soil organic C accounting system; it provides accurate and verifiable data on the spatial variation of soil bulk density so that changes in C stocks might be attributed more accurately to changes in either bulk density or in C content.

Soil provides ecosystem services, supports human health and habitation, stores carbon and regulates emissions of greenhouse gases. Unprecedented pressures on soil from degradation and urbanization are threatening agro-ecological balances and food security. It is important that we learn more about soil to sustainably manage and preserve it for future generations. To this end, we developed and analyzed a global soil visible–near infrared (vis–NIR) spectral library. It is currently the largest and most diverse database of its kind. We show that the information encoded in the spectra can describe soil composition and be associated to land cover and its global geographic distribution, which acts as a surrogate for global climate variability. We also show the usefulness of the global spectra for predicting soil attributes such as soil organic and inorganic carbon, clay, silt, sand and iron contents, cation exchange capacity, and pH. Using wavelets to treat the spectra, which were recorded in different laboratories using different spectrometers and methods, helped to improve the spectroscopic modelling. We found that modelling a diverse set of spectra with a machine learning algorithm can find the local relationships in the data to produce accurate predictions of soil properties. The spectroscopic models that we derived are parsimonious and robust, and using them we derived a harmonized global soil attribute dataset, which might serve to facilitate research on soil at the global scale. This spectroscopic approach should help to deal with the shortage of data on soil to better understand it and to meet the growing demand for information to assess and monitor soil at scales ranging from regional to global. New contributions to the library are encouraged so that this work and our collaboration might progress to develop a dynamic and easily updatable database with better global coverage. We hope that this work will reinvigorate our community’s discussion towards larger, more coordinated collaborations. We also hope that use of the database will deepen our understanding of soil so that we might sustainably manage it and extend the research outcomes of the soil, earth and environmental sciences towards applications that we have not yet dreamed of.

Soil landscapes often exhibit complex spatial patterns, with some aspects of soil variation apparently unrelated to measurable variations in environmental controls. However, these local, contingent complexities are not truly random or intrinsically unknowable. The purpose of this work is to develop and apply a method for identifying or teasing out causes of soil landscape complexity. Soil spatial adjacency graphs (SAG) represent the geography of soil landscapes as a network that can be analyzed using algebraic graph theory. These SAGs include linear sequential subgraphs that represent sequences of soil forming factors. The number and length of these soil factor sequences (SFS), and their associated spectral radius values, determine whether the SFS are sufficient to explain the spatial pattern of soil adjacency. SAGs and associated graph theory methods provide useful tools for guiding pedological investigations and identifying gaps in knowledge. The methods also allow sources of soil landscape complexity and variability to be determined in a way that can help assess the underlying deterministic sources of chaos and dynamical instability in pedology. The approach is applied to a soil landscape in central Kentucky, producing a SAG with 13 nodes (soil types) and 36 links indicating whether the soils occur contiguously. Five SFS were identified, the sum of whose spectral radius values is 6.35. The spectral radius of the SAG is 6.56, indicating that the SFS can explain most, but not all, of the complexity of the soil relationships. The analysis also points to potential environmental controls that could potentially enable full explanation.

A spatial sampling design that uses pair-copulas is presented that aims to reduce prediction uncertainty by selecting additional sampling locations based on both the spatial configuration of existing locations and the values of the observations at those locations. The novelty of the approach arises in the use of pair-copulas to estimate uncertainty at unsampled locations. Spatial pair-copulas are able to more accurately capture spatial dependence compared to other types of spatial copula models. Additionally, unlike traditional kriging variance, uncertainty estimates from the pair-copula account for influence from measurement values and not just the configuration of observations. This feature is beneficial, for example, for more accurate identification of soil contamination zones where high contamination measurements are located near measurements of varying contamination. The proposed design methodology is applied to a soil contamination example from the Swiss Jura region. A partial redesign of the original sampling configuration demonstrates the potential of the proposed methodology.

  • Poggio, L., Gimona, A., Spezia, L., & Brewer, M. J. (2016). Bayesian spatial modelling of soil properties and their uncertainty: The example of soil organic matter in Scotland using R-INLA. Geoderma, 277, 69–82. https://doi.org/10.1016/j.geoderma.2016.04.026

As any model for digital soil mapping suffers from different types of errors, including interpolation errors, it is important to quantify the uncertainty associated with the maps produced. The most common approach is some form of regression kriging or variation involving geostatistical simulation. Another way of assessing the spatial uncertainty lies in the Bayesian approach where the uncertainty is described by the posterior density. Typically Markov Chain Monte Carlo is used to compute the posterior density; however, this process is computationally intensive. The aim of this paper is to present an example of Bayesian uncertainty evaluation using (Bayesian) latent Gaussian models fitted using INLA (Integrated Nested Laplace Approximation) and with the SPDE (Stochastic Partial Differential Equation) approach for modelling the spatial correlation. For illustration, soil organic matter content in the Grampian region of Scotland (UK, about 12,100 km2) was modelled for topsoil (2D) and whole-profile data (3D). Results were assessed using in-sample and out-of-sample measures and compared for distribution similarity, variogram and spatial structure reproduction, computational load and uncertainty ranges. The results were also compared with outputs from an extension of scorpan-kriging. The Bayesian framework using INLA offers a viable alternative to existing methods for digital soil mapping, with comparable validation results, important computational gains, good assessment of uncertainty and potential for integrated modelling uncertainty propagation.

  • Angelini, M. E., Heuvelink, G. B. M., Kempen, B., & Morrás, H. J. M. (2016). Mapping the soils of an Argentine Pampas region using structural equation modelling. Geoderma, 281, 102–118. https://doi.org/10.1016/j.geoderma.2016.06.031

Current digital soil mapping (DSM) methods have limitations. For instance, it is difficult to predict a large number of soil properties simultaneously, while preserving the relationships between them. Another problem is that prevalent prediction models use pedological knowledge in a very crude way only. To tackle these problems, we investigated the use of structural equation modelling (SEM). SEM has its roots in the social sciences and is recently also being used in other scientific disciplines, such as ecology. SEM integrates empirical information with mechanistic knowledge by deriving the model equations from known causal relationships, while estimating the model parameters using the available data. It distinguishes between endogenous and exogenous variables, where, in our application, the first are soil properties and the latter are external soil forming factors (i.e. climate, relief, organisms). We introduce SEM theory and present a case study in which we applied SEM to a 22,900 km2 region in the Argentinian Pampas to map seven key soil properties. In this case study, we started with identifying the main soil forming processes in the study area and assigned for each process the main soil properties affected. Based on this analysis we defined a conceptual soil-landscape model, which was subsequently converted to a SEM graphical model. Finally, we derived the SEM equations and implemented these in the statistical software R using the latent variable analysis (lavaan) package. The model was calibrated using a soil dataset of 320 soil profile data and 12 environmental covariate layers. The outcomes of the model were maps of seven soil properties and a SEM graph that shows the strength of the relationships. Although the accuracy of the maps, based on cross-validation and independent validation, was poor, this paper demonstrates that SEM can be used to explicitly include pedological knowledge in prediction of soil properties and modelling of their interrelationships. It bridges the gap between empirical and mechanistic methods for soil-landscape modelling, and is a tool that can help produce pedologically sound soil maps.

The optimization of spatial sampling by simulated annealing has been demonstrated and applied for a range of objective functions. In practice more than one objective function may be important for sampling, and there may be complex trade-offs between them. In this paper it is shown how a multi-objective optimization algorithm can be applied to a spatial sampling problem. This generates a set of solutions which is non-dominated (no one solution does better than any other on all objective functions). These solutions represent different feasible trade-offs between the objective functions, and a subset might be practically acceptable. The algorithm is applied to a hypothetical example of sampling for a regional mean with the variance of the mean and the total distance travelled between sample points as the two objective functions. The solutions represent a transition of sample arrays from a loose grid to a tight loop. The potential to develop this approach and apply it to other spatial sampling problems is discussed.

You may also like...