In Memoriam: Dr. Ali Keshavarzi. An Inspiring Voice in Digital Soil Mapping and Scientific Generosity

Since different pedologists will draw different soil maps of a same area, it is important to compare the differences between mapping by specialists and mapping techniques, as for example currently intensively discussed Digital Soil Mapping. Four detailed soil maps (scale 1:10.000) of a 182-ha sugarcane farm in the county of Rafard, São Paulo State, Brazil, were compared. The area has a large variation of soil formation factors. The maps were drawn independently by four soil scientists and compared with a fifth map obtained by a digital soil mapping technique. All pedologists were given the same set of information. As many field expeditions and soil pits as required by each surveyor were provided to define the mapping units (MUs). For the Digital Soil Map (DSM), spectral data were extracted from Landsat 5 Thematic Mapper (TM) imagery as well as six terrain attributes from the topographic map of the area. These data were summarized by principal component analysis to generate the map designs of groups through Fuzzy K-means clustering. Field observations were made to identify the soils in the MUs and classify them according to the Brazilian Soil Classification System (BSCS). To compare the conventional and digital (DSM) soil maps, they were crossed pairwise to generate confusion matrices that were mapped. The categorical analysis at each classification level of the BSCS showed that the agreement between the maps decreased towards the lower levels of classification and the great influence of the surveyor on both the mapping and definition of MUs in the soil map. The average correspondence between the conventional and DSM maps was similar. Therefore, the method used to obtain the DSM yielded similar results to those obtained by the conventional technique, while providing additional information about the landscape of each soil, useful for applications in future surveys of similar areas.

Soils are often at the heart of the services that ecosystems deliver, not only in terms of food production, but also in filtering, the cycling of nutrients, the storage and regulation of water and in providing habitats for soil biota. As a result of overuse, soils and their functions are under increasing pressure. Degradation of soils in the form of erosion, dust storms, salinisation, pollution, compaction, depletion, decomposition of organic matter and destruction of soil aggregates, is the result. The mapping of soil properties and functions and the monitoring of changes over time are important to secure soil functions in the future. Especially, spatially distributed soil information has become more important with the use of global and regional models, which often require full coverage soil information. The use of remote sensing can offer spatial and temporal quantitative soil information of extended areas, which can be acquired with limited fieldwork. Remote sensing under laboratory conditions or infield studies in semi-arid areas has shown promising results for soil purposes. However, when acquiring data with airborne or satellite sensors of extended areas in temperate zones, limitations by vegetation coverage and soil surface variations are driving coherent spatiotemporal data collection. The use of multi-temporal data in agricultural areas can be used to increase the bare soil area captured by remote sensing data. The alternation of crops results in bare fields at the moment of seeding and harvesting. We used spectrally and spatially high-resolution data from an airborne imaging spectrometer of three consecutive years to create a multi-temporal composite. This composite contained more than double the amount of bare soil pixels as compared to a singular acquisition. Global linear soil surface variations could be compensated based on the spectral differences only. In order to compensate for local non-linear soil surface variations, however, quantitative information was needed. Based on independent datasets of soil moisture and soil surface roughness, we were able to correct per wavelength for these local non-linear variations with a relative simple algorithm. The advantage of the independent datasets is that the used algorithm can be applied to all imaging spectroscopy data with known soil moisture and/or soil surface roughness. A better soil surface roughness dataset is needed in order to improve the results. Additionally, the concept of the multi-temporal composite was also applied to Landsat time series from 1985 to 2017. Although spectrally less detailed, these sensors provide denser time series and larger extents than the high-resolution airborne data. About 5 years of Landsat data were needed for a full-coverage composite (90% of the maximum bare soil area). We show the substantial potential of Landsat time series for digital soil mapping, as well as for land management applications and policy making. This thesis shows that optical remote sensing for soil purposes offers valuable spatial soil information. Spectrally and spatially high-resolution data are able to show in-field variations. These are not covered by more standard soil mapping approaches like digital and conventional soil mapping. The need to correct for soil surface variations is, however, necessary in order to give a realistic picture of these in-field variations. The temporally high-resolution, but spectrally less detailed, data provides soil maps with very similar patterns compared to the available digital soil map. In areas with limited alternating crops, the use of remote sensing for soil purposes is limited. Finally, we discuss to which extent the focus of soil studies with remote sensing data should be on the standardisation of protocols and the necessary pre-processing steps. A further development of the scientific community in this context is desirable, especially since there is a need to map, monitor and model soil changes. The present work has shown the added value of remote sensing data and products for these purposes.

The Natural Resources Conservation Service (NRCS), within the context of the U.S. National Cooperative Soil Survey (NCSS), is working to integrate digital soil mapping methods with existing soil survey procedures. As this effort moves forward, it must address technological, managerial, and political challenges. To better understand these challenges and potential solutions, NRCS is establishing Digital Soil Mapping Operational Initiatives. These projects aim to demonstrate the utility of digital soil mapping in a production setting, provide training to soil scientists in digital soil mapping methods, contribute to completion of the initial soil survey or update of existing surveys, develop detailed instructions for implementing digital soil mapping methods, provide useful soil information products to complement existing soil survey data, and document methods and results. The first Operational Initiative was initiated at the Victorville, California Major Land Resource Area (MLRA) Soil Survey Office (SSO), which is responsible for the soil survey of Mojave Desert region. The immediate focus of this office is completing the initial soil survey for Joshua Tree National Park and adjacent private lands. Under the operational initiative umbrella, detailed digital data sets including IFSAR digital elevation models and an ASTER mosaic have been compiled. Derivatives from these and other data sets are being used to stratify the project area for sampling and modeling, and as inputs into continuous soil property predictive models. Model outputs will be used to develop Soil Survey Geographic (SSURGO) data products. Technical support for this project is provided by digital soil mapping soil scientists at the MLRA SSO the California State Office, and the National Geospatial Development Center, as well as other NRCS staff and NCSS cooperators.

Digital soil mapping utilizes computer analysis and digital data to create a soil map. This exciting field provides the opportunity for employing new data and developing new techniques to create better soil maps, and to do so more efficiently. Although digital soil mapping often relies on concepts developed from traditional soil mapping, it is distinguished by the quantitative nature of the data and the analysis processes employed to generate the map. Digital soil maps are critical resources for precision agriculture and land‐use and conservation planning, as well as for environmental modeling. Pedometrics is a quantitative and statistical form of pedology. Both the quantitative basis of digital soil mapping and the importance of soil geography to pedology make digital soil mapping closely related to pedometrics. However, in the dynamic and diverse field of soil science, these terms are not necessarily synonymous, because not all techniques for digital soil mapping are considered to be a part of the field of pedometrics.

Pedology focuses on understanding soil genesis in the field and includes soil classification and mapping. Digital soil mapping (DSM) has evolved from traditional soil classification and mapping to the creation and population of spatial soil information systems by using field and laboratory observations coupled with environmental covariates. Pedological knowledge of soil distribution and processes can be useful for digital soil mapping. Conversely, digital soil mapping can bring new insights to pedogenesis, detailed information on vertical and lateral soil variation, and can generate research questions that were not considered in traditional pedology. This review highlights the relevance and synergy of pedology in soil spatial prediction through the expansion of pedological knowledge. We also discuss how DSM can support further advances in pedology through improved representation of spatial soil information. Some major findings of this review are as follows: (a) soil classes can be mapped accurately using DSM, (b) the occurrence and thickness of soil horizons, whole soil profiles and soil parent material can be predicted successfully with DSM techniques, (c) DSM can provide valuable information on pedogenic processes (e.g. addition, removal, transformation and translocation), (d) pedological knowledge can be incorporated into DSM, but DSM can also lead to the discovery of knowledge, and (e) there is the potential to use process‐based soil–landscape evolution modelling in DSM. Based on these findings, the combination of data‐driven and knowledge‐based methods promotes even greater interactions between pedology and DSM. Highlights Demonstrates relevance and synergy of pedology in soil spatial prediction, and links pedology and DSM. Indicates the successful application of DSM in mapping soil classes, profiles, pedological features and processes. Shows how DSM can help in forming new hypotheses and gaining new insights about soil and soil processes. Combination of data‐driven and knowledge‐based methods recommended to promote greater interactions between DSM and pedology.

Using a fuzzy logic approach to reveal soil-landscape relationships produced by digital soil maps in the humid tropical region of East Java, Indonesia

Bibliometric analyses provide a clear understanding of the scientific performance and relate them with standards of the global scientific production. Soil science is an outstanding and developing field among environmental sciences. Knowledge about soil characteristics and their distribution in the environment has been enriched by the use of new geotechnologies, resulting in what is known as digital soil mapping. Thus, the objective of this work was to characterize the scientific production in digital soil mapping in Brazil and in the world, in the period from 1996 to 2017, in databases such as Scopus and Web of Science. In the general context of increasing numbers of papers, the journal Geoderma published the highest number of related papers. Among the 10 with most published papers, the Revista Brasileira de Ciência do Solo is the only open access journal. Although there are countries at the cutting edge of digital soil mapping such as the United States and Australia, the position of Brazil in the number of papers and authors cannot be overlooked, showing the importance of the nation's participation in digital soil mapping, as a field of science that can provide guidelines for public policies for the development of agriculture in the country.

The uptake of machine learning (ML) algorithms in digital soil mapping (DSM) is transforming the way soil scientists produce their maps. Within the past two decades, soil scientists have applied ML to a wide range of scenarios, by mapping soil properties or classes with various ML algorithms, on spatial scale from the local to the global, and with depth. The wide adoption of ML for soil mapping was made possible by the increase in data availability, the ease of accessing environmental spatial data, and the development of software solutions aided by computational tools to analyse them. In this article, we review the current use of ML in DSM, identify the key challenges and suggest solutions from the existing literature. There is a growing interest in the use of ML in DSM. Most studies emphasize prediction and accuracy of the predicted maps for applications, such as baseline production of quantitative soil information. Few studies account for existing soil knowledge in the modelling process or quantify the uncertainty of the predicted maps. Further, we discuss the challenges related to the application of ML for soil mapping and suggest solutions from existing studies in the natural sciences. The challenges are: sampling, resampling, accounting for the spatial information, multivariate mapping, uncertainty analysis, validation, integration of pedological knowledge and interpretation of the models. Overall, the current literature shows few attempts in understanding the underlying soil structure or process using the predicted maps and the ML model, for example by generating hypotheses on mechanistic relationships among variables. In this regard, several additional challenging aspects need to be considered, such as the inclusion of pedological knowledge in the ML algorithm or the interpretability of the calibrated ML model. Tackling these challenges is critical for ML to gain credibility and scientific consistency in soil science. We conclude that for future developments, ML could incorporate three core elements: plausibility, interpretability, and explainability, which will trigger soil scientists to couple model prediction with pedological explanation and understanding of the underlying soil processes.

Selected highlights in American soil science history from the 1980s to the mid-2010s

Digital soil mapping: A brief history and some lessons

On digital soil assessment with models and the Pedometrics agenda

The United Nations pledged to achieve the Sustainable Development Goals by 2030. Regional land use analyses (RLUA) have an essential contribution to achieving these goals. To better meet the needs for achieving sustainable development, RLUA became more quantitative and more interdisciplinary over recent decades. This change resulted in an increased use of quantitative simulation models, which changed the type and nature of input data as well. Soil data are one of the input data RLUA require. Available soil data often do not meet the soil data requirements anymore, due to the change in RLUA. Therefore, a gap exists between the available and required soil data. This thesis aims to find possible solutions to bridge this gap. In Chapter 2, different soil datasets are compared to identify the gap and to analyse the effect of using different soil datasets as input for a regional land use analysis (RLUA). Main challenges with soil data in RLUA are: i) understanding the assumptions in soil datasets, ii) creating soil datasets that meet the requirements for regional land use analysis, iii) not only rely on available soil data but also collect new soil data and iv) validate soil datasets. Chapter 2 demonstrated differences between soil datasets, which had significant effect on the results of RLUA. Three potential solutions on bridging the gap between the available and required soil data are given in Chapter 3, 4 and 5. A literature study showed that RLUA hardly combine available and newly collected soil data. Chapter 3 analyses what complementary data RLUA require by combining available soil data and newly collected soil data. Two case studies were carried out to illustrate how a combination can enrich the soil data for RLUA. Predicting soil properties, in particular soil organic matter, using newly collected soil data often result in soil maps of poor quality. The digital soil mapping techniques that are currently being used for predicting soil properties make dominantly use of statistical models, while much knowledge on the mechanistic processes that influence a soil property are available. To improve the prediction of soil organic matter, a mechanistic model for digital soil mapping (DSM) is developed and the potential of mechanistic digital soil mapping is explored in Chapter 4. Mechanistic digital soil mapping predicts soil properties by values that typically stay within realistic boundaries. Complex soil mapping techniques are increasingly being used to better meet the data requirements, because the use of quantitative simulation models in RLUA increased over recent decades. Chapter 5 analyses whether the required soil data can be obtained more targeted to RLUA. Three case studies were carried out to illustrate that the complexity of quantitative simulation models can differ from the complexity required by the RLUA. Therefore, the spatial variation at which the soil properties are provided need to be in line with the spatial variation at which the RLUA operate. In the synthesis (Chapter 6), the research findings, the implementation of the research findings, the hypothesis and future perspectives are discussed and recommendations towards the soil science community and the people involved in RLUA are provided. In this chapter, the ARDAIG approach is introduced, which aims to be an approach that helps obtaining the required soil data for RLUA more targeted. If the soil science community and the people involved in RLUA will implement the presented recommendations, there are opportunities to make soil science contribute more efficiently in RLUA.

The Thematic Strategy for Soil Protection, prepared by the European Commission in 2006, concluded that soil degradation is a significant problem in Europe. Degradation is driven or exacerbated by human activity and has a direct impact on water and air quality, biodiversity, climate, and the quality of (human) life. High-resolution soil property maps are a major prerequisite for the specific protection of soil functions and the restoration of degraded soils, as well as for sustainable land use and water and environmental management. To generate such maps, a combination of digital soil mapping approaches and remote and proximal soil sensing techniques is most promising. However, a feasible and reliable combination of these technologies for the investigation of large areas (e.g. catchments and landscapes) and the assessment of soil degradation threats is still missing. There is insufficient dissemination – to relevant authorities as well as prospective users – of knowledge on digital soil mapping and proximal soil sensing from the scientific community. As a consequence, there is inadequate standardisation of the techniques. In this chapter we present the EU project iSOIL, which is funded within the 7th Framework Program of the European Commission. iSOIL focuses on improving and developing fast and reliable mapping of soil properties, soil functions, and soil degradation threats. This requires the improvement and integration of advanced soil sampling approaches, geophysical and spectroscopic measurement techniques, as well as pedometric and pedophysical approaches. Another important aspect of the project is the sustainable dissemination of the technologies and the concepts developed. For this purpose, guidelines for soil mapping on different scales, and using various methods for field measurements, will be written. Outcomes of the project’s measurements will be implemented in national and European soil databases. The present state of knowledge and future perspectives will be communicated to authorities, providers of technologies (e.g. small and medium enterprises), and end-users.

Numerous technological advancements have assisted to secure the vital key for scientific predictions in various fields, and soil science is no exception. Soil has always been chosen as an indispensable component by scientists, environmentalists, and policy makers for shaping a sustainable present and a secure future. Evidently, a huge amount of soil spatial data is required in the process to attain the desired outcomes. However, the challenges posed by the paucity of time and resources pose the hurdle in the collection and analysis of soil information. State-of-art technologies like Machine Learning (ML) and Big Data come as saviors to address those challenges. ML is helping to quantify, predict, identify, and classify the soil resources. The advanced algorithms, and models helped to gain better insights into soil mapping along with widening the perspective for its better management. Digital Soil Mapping (DSM), ML integrated with spectroscopic soil studies is gaining momentum among the scientific communities. These innovative approaches have the capabilities to solve global issues like desertification, ecological stability, carbon pool management and climate mitigation in a holistic and integrated way by keeping the soil as one of the key parameters. In this chapter, an attempt has been made to present a comprehensive overview of ML algorithms, which have been adopted by many researchers across the globe in prediction of various soil properties and presented a case study on digital soil mapping by using ML algorithms.KeywordsMachine learningDigital soil mappingSoil propertiesSpectroscopic soil studiesPredictionRandom forest modelSupport vector machine

Precision agriculture has been proposed to improve the sustainability of agriculture and solve the environmental pollution of soil. In precision agriculture process, the management of water and fertilizer is carried out on agricultural operation units. Therefore, acquisition of accurate soil nutrient distribution information is a key step for precision agriculture application and digital soil mapping is an effective technology. Significant progress has been made in digital soil mapping over the past 20 years. However, the current digital soil mapping framework was implemented based on grids, which was not consistent with the operation units of precision agriculture. This paper proposed a geo-parcel based digital soil mapping framework on the support of artificial intelligence technology for precision agriculture application. Two key technologies were studied for the implementation of this framework. Geo-parcels automatic extraction was the basis of this method, and a modified VGG 16 network was used for geo-parcels' accurate boundary extraction from high resolution images. Different machine learning methods were attempted to construct the relationship between soil available phosphorus and environment on geo-parcels. We chose an agricultural region in Zhongning County, Ningxia Province as the study area, and the new digital soil mapping framework was applied for soil available phosphorus mapping. This research showed that geo-parcel based digital mapping method could reduce the number of prediction units more than 50% for fine soil mapping, and effectively improve the prediction and application efficiency. This study was an attempt to realize soil mapping based on agricultural operation units for precision agriculture application. The high resolution remote sensing images provide basic data for the realization of this idea and the development of AI technology provides technical support for it. In the future, we will carry out experiments in larger areas to further optimize this method and key technologies for the applications in more complex environments.