Search

The advent of affordable, ground-based, global positioning information (GPS)–enabled sensor technologies provides a new method to rapidly acquire georeferenced soil datasets in situ for high-resolution soil attribute mapping. Our research deployed vehicle-mounted electromagnetic sensor survey equipment to map and quantify soil variability (˜50 ha per day) using apparent electrical conductivity as an indirect measure of soil texture and moisture differences. A portable visible–near infrared (VNIR) spectrometer (350–2500 nm) was then used in the field to acquire hyperspectral data from the side of soil cores to a specified depth at optimized sampling locations. The sampling locations were derived by statistical analysis of the electromagnetic survey dataset, to proportionally sample the full range of spatial variability. The VNIR spectra were used to predict soil organic carbon (prediction model using field-moist spectra: R2 = 0.39; RPD = 1.28; and air-dry spectra: R2 = 0.80; RPD = 2.25). These point values were combined with the electromagnetic survey data to produce a soil organic carbon map, using a random forest data mining approach (validation model: R2 = 0.52; RMSE = 3.21 Mg C/ha to 30 cm soil depth; prediction model: R2 = 0.92; RMSE = 1.53 Mg C/ha to 30 cm soil depth). This spatial modeling method, using high-resolution sensor data, enables prediction of soil carbon stocks, and their spatial variability, at a resolution previously impractical using a solely laboratory-based approach. [ABSTRACT FROM PUBLISHER]

• Gradient Forest was used to study soil-type turnover along environmental gradients. • Soil-forming factors were ranked and transformed to amplify their critical parts. • The transformed variables were segmented into regions, at three levels of scale. • The resulting segments were clustered to produce discrete soilscapes. • Validation showed the soilscapes delineation yielded sensible results at all scales. Soilscapes are a regionalisation of the landscape into a suite of contiguous segments with an homogeneous set of soil forming factors. The concept has been present in the soil science literature since the late 1950s, proving to be a useful framework for generalising traditional soil survey maps. This concept is inherently multi-scale, and current approaches generally build either on existing soil maps, or on cluster analysis of environmental proxies of soil forming factors. In this study, in parallel with the concept of beta-diversity in numerical ecology, we built a soilscape regionalisation for New Zealand based on detecting the turnover in soil types along different environmental gradients. It was implemented using Gradient Forest, a method that analyses the frequency of species (in our case, soil type) turnover alongside such gradients, which are then ranked relative to their ecological (in our case, pedological) relevance. Based on this analysis, the environmental covariates were transformed to amplify their most important parts in terms of soil type turnover. The transformed variables were then segmented into contiguous regions, at three different levels of scale. Finally, hierarchical clustering was used to produce discrete soilscapes, at different scales, and for the entire country. This approach was developed using pseudo-observations derived from a national scale coverage of soil types, and validated using an independent set of actual soil profile observations. Validation results reveal strong agreement between the dissimilarity quantified between soilscapes and the taxonomic distance between actual soil profile observations (R = 0.61, P < 0.001). We expect the resulting layers from this soilscape delineation exercise to be used for a diverse suite of pedometrics applications, from stratifying soil sampling schemes, to optimising the soil survey boundaries, and for digital soil mapping. [ABSTRACT FROM AUTHOR]

Understanding the spatial variation of soil pH is critical for many different stakeholders across different fields of science, because it is a master variable that plays a central role in many soil processes. This study documents the first attempt to map soil pH (1:5 H 2 O) at high resolution (100 m) in New Zealand. The regression framework used follows the paradigm of digital soil mapping, and a limited number of environmental covariates were selected using variable selection, before calibration of a quantile regression forest model. In order to adapt the outcomes of this work to a wide range of different depth supports, a new approach, which includes depth of sampling as a covariate, is proposed. It relies on data augmentation, a process where virtual observations are drawn from statistical populations constructed using the observed data, based on the top and bottom depth of sampling, and including the uncertainty surrounding the soil pH measurement. A single model can then be calibrated and deployed to estimate pH a various depths. Results showed that the data augmentation routine had a beneficial effect on prediction uncertainties, in particular when reference measurement uncertainties are taken into account. Further testing found that the optimal rate of augmentation for this dataset was 3-fold. Inspection of the final model revealed that the most important variables for predicting soil pH distribution in New Zealand were related to land cover and climate, in particular to soil water balance. The evaluation of this approach on those validation sites set aside before modelling showed very good results ( R 2 = 0.65 , C C C = 0.79 , R M S E = 0.54 ), that significantly out-performed existing soil pH information for the country. [ABSTRACT FROM AUTHOR]

• Hand-feel soil texture and particle-size distribution are compared using a large database. • The overall accuracy of hand-feel soil texture class allocation was 73% • Most discrepancies were explained by very fine and coarse sand content. • Predicting soil water retention at pF2 using hand-feel texture gave satisfactory results. Due to cost constraints, field texture classes estimated by hand-feel by soil surveyors are more abundant than laboratory measurements of particle-size distribution. Thus, there is a considerable potential to use field-estimated soil textures for mapping on the condition that they are reliable and can be characterized by a probability distribution function similar to values obtained by laboratory measurements. This study aimed to investigate and elucidate the differences between the field texture classes estimated by hand-feel and soil texture determined from particle-size analysis under laboratory conditions in a region of Central France. We tested several hypotheses to explain the discrepancies between field estimates and laboratory measurements (organic C content, pH, more detailed particle-size analyses, and CEC). Finally, we simulated the consequences of using particle-size distribution estimated from field texture on a pedotransfer function (PTF) for water retention. Laboratory measurements of clay, silt, and sand content for each field texture class were available for about 17,400 samples. Considering laboratory measurements and the French texture triangle as the reference, the overall accuracy of field texture class allocation was 73%, which was better than most of the results previously reported in the literature. When looking at each field texture class, most predictions were consistent; however, there were noticeable differences between a few field texture classes and particle-size classes. The extreme texture classes located at the corners of the texture triangle were better predicted than those located at the centre of the triangle. We found the discrepancy of field texture classes can be explained by the very fine sand (50–100 µm) and very coarse sand (1000–2000 µm) contents. Based on the particle-size distribution from each field texture class, we calculated their joint probability distribution function of their corresponding laboratory measurements of clay, silt, and sand content. Results showed that PTF values predicted using hand-feel texture were consistent with those obtained with the measured particle-size distribution. Overall, we demonstrated the value of hand-feel texture in expanding the soil texture database and supporting the expansion of the national database to inform soil water retention properties. [ABSTRACT FROM AUTHOR]

• Digital soil mapping studies at broad scale strongly increased from 2003 to 2019. • Most studies focused on mapping SOC and soil particle size fractions. • Model performance usually decreased in deeper depth intervals. • New studies should explore other covariates, functional soil properties and deeper layers. • Challenges remain for assessing performance, uncertainty and changes with time. Soils are essential for supporting food production and providing ecosystem services but are under pressure due to population growth, higher food demand, and land use competition. Because of the effort to ensure the sustainable use of soil resources, demand for current, updatable soil information capable of supporting decisions across scales is increasing. Digital soil mapping (DSM) addresses the drawbacks of conventional soil mapping and has been increasingly used for delivering soil information in a time- and cost-efficient manner with higher spatial resolution, better map accuracy, and quantified uncertainty estimates. We reviewed 244 articles published between January 2003 and July 2021 and then summarised the progress in broad-scale (spatial extent >10,000 km2) DSM, focusing on the 12 mandatory soil properties for GlobalSoilMap. We observed that DSM publications continued to increase exponentially; however, the majority (74.6%) focused on applications rather than methodology development. China, France, Australia, and the United States were the most active countries, and Africa and South America lacked country-based DSM products. Approximately 78% of articles focused on mapping soil organic matter/carbon content and soil organic carbon stocks because of their significant role in food security and climate regulation. Half the articles focused on soil information in topsoil only (<30 cm), and studies on deep soil (100–200 cm) were less represented (21.7%). Relief, organisms, and climate were the three most frequently used environmental covariates in DSM. Nonlinear models (i.e. machine learning) have been increasingly used in DSM for their capacity to manage complex interactions between soil information and environmental covariates. Soil pH was the best predicted soil property (average R2 of 0.60, 0.63, and 0.56 at 0–30, 30–100, and 100–200 cm). Other relatively well-predicted soil properties were clay, silt, sand, soil organic carbon (SOC), soil organic matter (SOM), SOC stocks, and bulk density, and coarse fragments and soil depth were poorly predicted (R2 < 0.28). In addition, decreasing model performance with deeper depth intervals was found for most soil properties. Further research should pursue rescuing legacy data, sampling new data guided by well-designed sampling schemas, collecting representative environmental covariates, improving the performance and interpretability of advanced spatial predictive models, relating performance indicators such as accuracy and precision to cost-benefit and risk assessment analysis for improving decision support; moving from static DSM to dynamic DSM; and providing high-quality, fine-resolution digital soil maps to address global challenges related to soil resources. [ABSTRACT FROM AUTHOR]