Best Paper 2018

The Awards Committee of the Pedometrics Commission has received nominations from an open call for the Best Paper 2018 competition.  With nominations from the panel members a total of 27 papers were considered.  Each committee member selected their top 15 papers in rank order (excluding papers on which they were coauthors) and the top five papers in terms of support across the committee as a whole are to be put to an open vote.  The result will be announced at the Pedometrics 2019 Meeting at Guelph in June 2019.

Votes should be received by Murray Lark at murray.lark( at ) before midday, BST on 31st May.  As before:

  • Please rank papers in order, with the first paper the one you regard as most deserving. You need not provide a rank for every paper nominated.
  • Votes must be received from a traceable email address, and if I cannot verify their origin they will be discarded.
  • Authors/co-authors should not vote for their own papers.


Murray Lark on behalf of the Awards Committee (Sabine Grunwald, Gerard Heuvelink, Lin Yang, Uta Stockman, Alessandro Samuel-Rosa).



  1. Angelini, M.E., Gerard B.M.Heuvelink. 2018. Including spatial correlation in structural equation modelling of soil properties Spatial Statistics 25, 35–51.

Digital soil mapping techniques usually take an entirely data-driven approach and model soil properties individually and layer by layer, without consideration of interactions. In previous studies we implemented a structural equation modelling (SEM) approach to include pedological knowledge and between-properties and between-layer interactions in the mapping process. However, it typically does not consider spatial correlation. Our goal was to extend SEM by accounting for residual spatial correlation using a geostatistical approach. We assumed second-order stationary and estimated the semivariogram parameters, together with the usual SEM parameters, using maximum likelihood estimation. Spatial prediction was done using regression kriging. The methodology is applied to mapping cation exchange capacity, clay content and soil organic carbon for three soil horizons in a 150100-km study area in the Great Plains of the United States. The calibration process included all parameters used in lavaan, a SEM software, plus two extra parameters to model residual spatial correlation. The residuals showed substantial spatial correlation, which indicates that including spatial correlation yields more accurate predictions. We also compared the standard SEM and the spatial SEM approaches in terms of SEM model coefficients. Differences were substantial but none of the coefficients changed sign. Presence of residual spatial correlation suggests that some of the causal factors that explain soil variation were not captured by the set of covariates. Thus, it is worthwhile to search for additional covariates leaving only unstructured residual noise, but provided that this is not achieved, it is beneficial to include residual spatial correlation in mapping using SEM.

  • Behrens, Thorsten; Schmidt, Karsten; MacMillan, Robert A.; et al. 2018. Multi-scale digital soil mapping with deep learning  Scientific Reports  8   Article Number: 15244.

We compared different methods of multi-scale terrain feature construction and their relative effectiveness for digital soil mapping with a Deep Learning algorithm. The most common approach for multi-scale feature construction in DSM is to filter terrain attributes based on different neighborhood sizes, however results can be difficult to interpret because the approach is affected by outliers. Alternatively, one can derive the terrain attributes on decomposed elevation data, but the resulting maps can have artefacts rendering the approach undesirable. Here, we introduce ‘mixed scaling’ a new method that overcomes these issues and preserves the landscape features that are identifiable at different scales. The new method also extends the Gaussian pyramid by introducing additional intermediate scales. This minimizes the risk that the scales that are important for soil formation are not available in the model. In our extended implementation of the Gaussian pyramid, we tested four intermediate scales between any two consecutive octaves of the Gaussian pyramid and modelled the data with Deep Learning and Random Forests. We performed the experiments using three different datasets and show that mixed scaling with the extended Gaussian pyramid produced the best performing set of covariates and that modelling with Deep Learning produced the most accurate predictions, which on average were 4–7% more accurate compared to modelling with Random Forests.

  • Finke, Peter A.; Yin, Qiuzhen; Bernardini, Nicholas J.; et al. 2018. Climate-soil model reveals causes of differences between Marine Isotope Stage 5e and 13 paleosols.  Geology   46,  99–102

Over the last decades, numerous studies have used the loess-paleosol sequences in China to reconstruct the East Asian climate and to investigate their linkage with global climate change. The paleosols embedded in the loess developed during warm periods and contain valuable information on climate and vegetation under warm conditions. However, because soil formation is controlled by multiple factors, it is not straightforward to obtain a pure climate signal based on soil property analyses. This leads often to debates and questions. Here, for the first time, we use a soil formation model together with a climate model to identify the main factors that control the paleosol formation. A case study has been performed on paleosols in the Chinese Loess Plateau of the last interglacial (Marine Isotope Stage [MIS] 5e) and the interglacial at ca. 500 kyr B.P. (MIS 13). Our results show that although the peak warmth and peak summer monsoon precipitation are stronger during MIS 5e, the soil formation is stronger during MIS 13, which is supported by field evidence. This is mainly due to larger accumulative precipitation surplus, weaker dust deposition, and longer interglacial duration during MIS 13. Our results provide a new interpretation of the climate signal recorded by the paleosols, and an explanation for the seeming paradox that strongly developed soils formed during relatively weak interglacials. They also highlight the necessity to include proxy modeling in paleoclimate studies.

  • Guenet, B., Camino‐Serrano, M., Ciais, P., Tifafi, M., Maignan, F., Soong, J. L., & Janssens, I. A. Impact of priming on global soil carbon stocks. Global Change Biology, 24, 1873–1883.

Fresh carbon input (above and belowground) contributes to soil carbon sequestration, but also accelerates decomposition of soil organic matter through biological priming mechanisms. Currently, poor understanding precludes the incorporation of these priming mechanisms into the global carbon models used for future projections. Here, we show that priming can be incorporated based on a simple equation calibrated from incubation and verified against independent litter manipulation experiments in the global land surface model, ORCHIDEE. When incorporated into ORCHIDEE, priming improved the model’s representation of global soil carbon stocks and decreased soil carbon sequestration by 51% (12 ± 3 Pg C) during the period 1901–2010. Future projections with the same model across the range of CO2 and climate changes defined by the IPCC‐RCP scenarios reveal that priming buffers the projected changes in soil carbon stocks — both the increases due to enhanced productivity and new input to the soil, and the decreases due to warming‐induced accelerated decomposition. Including priming in Earth system models leads to different projections of soil carbon changes, which are challenging to verify at large spatial scales.

  • Paterson, S.; Minasny, B. & McBratney, A. 2018. Spatial variability of Australian soil texture: A multiscale analysis. Geoderma, 309, 60–74.

Understanding how soil variability changes with spatial scale is critical to our ability to understand and model soil processes at scales relevant to decision makers. The compilation of large legacy data sets has opened up new possibilities to model spatial variability at the continental or even global scale. Using the National Soil Site Collation (NSSC) dataset of Australia we created empirical variograms for sand and clay fraction at extents from 1 km to continental. The NSSC dataset is highly spatially clustered; a typical feature of legacy datasets. This leads to lumpy artefacts in the variograms. To reduce this lumpiness we employed grid based declustering. We used the declustered empirical variograms to calculate the Hausdorff Besicovitch Dimension – a unitless measure of spatial roughness. We first fit a power model to each declustered variogram and calculated the Hausdorff Besicovitch dimension at each modelled scale. This allowed us to assess the roughness or variability at each modelled extent, however this assessment was somewhat arbitrary and showed that roughness depends on the extent. We have proposed a new model that allows us to calculate the Hausdorff Besicovitch dimension continuously across all extents. The conceptual basis of this model moves away from a multi-fractal framework typically used by soil scientists. It allows us to describe spatial variability or stochasticity as a continuous function of spatial separation. Both our new model and the continental scale variograms of texture emphasise the high degree of short range variability in soil texture. Empirical variograms indicate that around 50% of spatial variability occurs at < 10 km, and 30% at < 1 km. Spatial variability of soil texture increases with depth consistently across all modelled extents. Beyond extents of around 100 km, the Hausdorff Besicovitch Dimension remains relatively stable. Soil spatial variability is highly stochastic at fine scales however it changes gradually with extent and scale rather than abruptly.

You may also like...