A Geoinformatics Approach to Water Erosion: Soil Loss and Beyond

1. Soil erosion: The general problem Chapter 1 will describe the extent of the phenomenon, its impact on nations and communities, and the role that agricultural catchments play in the process. The first section of Chapter 1 will present the basic terms and definitions, as well as the extent of soil loss as an earth surface phenomenon at the global scale. It will describe the effects of climatic, geological, and human-induced parameters on the spatial distribution of soil loss due to water erosion. It will explain why certain areas around the world are hotspots of soil loss due to water erosion, and the ramifications for the societies living in those areas (developed vs. developing countries, for example). The second section of Chapter 1 will describe the historical aspect of soil loss – namely, the effect of ancient civilizations (especially empires) on soil loss due to water erosion throughout the Paleo-Anthropocene. In the third and final section of Chapter 1, the important role of the soil layer as a finite resource will be discussed through an analysis of rates of soil loss versus rates of soil gain in various parts of the world under different land-use regimes. This last section will show how the human race irreversibly destroys – in an intense process over a few decades – a resource that takes nature thousands of years to develop. Chapter 1 will be summarized by describing the phenomena of soil loss due to water erosion at regional and global scales. It will also explain and demonstrate why aeolian erosion is omitted from this book, and the consequences of that omission. 2. The case of agricultural catchments Chapter 2’s first section will describe the phenomenon of soil loss at the catchment scale – with particular focus on agricultural catchments, their distinctive character compared with catchments under other land uses, and the economic implications that soil loss due to water erosion has in agriculturally important areas. The second section of Chapter 2 will focus on the differences between (1) the on-site consequences of soil loss for the health of the remaining soil and (2) the risk of off-site consequences of soil erosion for farmers and infrastructure – for built areas, for other fields in the entire catchment, and for life itself – due to the intersection of soil mass with roads. Also in this section, the reader will learn that soil loss from agricultural fields and the diminishing quality of fields are not the only threats that may arise: soil movement and deposition in other areas can also yield substantial damage. The third and last section of this chapter will survey the direct damage to farmers and will describe the barriers – psychological, as well as physical – that prevent farmers from joining in the battle against soil erosion in his fields, and review and propose tools for enlisting farmers in the effort. Part I will provide a summary from the existing literature of the damage wrought by water erosion to individuals and communities at both the regional (national) and catchment (settlements and farmers) scales of observation. It will demonstrate why soil loss is a major environmental problem that justifies the deployment of every technology we have at our disposal. Part II: Water erosion processes 3. The physical process This chapter will begin with a short description of the soil as a medium, and the physical forces that integrate the three soil phases (solid, fluid, and gas). Special attention will be given to soil-water characteristics – such as initial soil-water conditions and soil moisture level; erosivity; erodibility; and soil movement and sedimentation. The effects of rainfall and surface characteristics on runoff and runon processes will also be discussed and illustrated through diagrams and equations. Emphasis will be placed on runoff processes in agricultural fields as a unique case – no rockiness, and roughness increases after soil plowing and crop growth in the early winter. Chapter 3 will also quantify soil movement by water to provide a relatively simple case study in synthetic computerized environments. Special attention will be given to the description of channeled flow (rill, gully, and piping), as well as to sheet flow in inter-rill erosion processes. Splash erosion will also be described – albeit but less rigorously, as its importance in soil loss from agricultural fields is relatively minor compared with channeled water flow. These water erosion processes will be demonstrated using graphics, diagrams, flow charts, and quantitative simulations. Finally, this chapter will review both theoretical and empirical models. 4. Spatial variation in the catchment This Chapter will provide with the reader the necessary data models to quantify environmental variation in the catchment scale. While Chapter 3 will focus on governing forces, Chapter 4 will describe and review the parameter space that affects water erosion processes in various environments – in particular, the dry environments that serve as the case study in Losing Ground. The parameters will be studied and described based on their weighting and importance for the process as a whole, and in terms of their quantifiability in the field using geoinformatics. The first part of this chapter will review the current literature on the impacts of climatic, environmental, and human-induced factors on water erosion processes. These studies will cover various scales of observation; various methods (laboratory experiments, as well as field measurements, and the use of geoinformatics); and various parts of the world. The second part will describe the diverse tools that may be used to extract governing parameters and to interpolate them in the field, using spatial analysis tools and catchment analysis tools. It will review and provide the reader with the two main approaches for spatial data modeling with geoinformatics – namely, discrete (the first part of the Chapter) and continuous-data models (the second). Once again, concrete equations and algorithms will be provided. Part II will rigorously quantify the forces that cause soil to move from one location to another, through equations and diagrams that incorporate data on, and simulations of, water erosion processes and the parameter space. This part will constitute the background for the geoinformatics and data analytics methods that may be used to battle soil loss. Part III: Using geoinformatics This section of the book (and also its longest) will contain its principal contribution – namely, a description of the relevant geoinformatics and data-science methods. It will provide a step-by-step, practical guide to the technologies, techniques, and procedures involved, as described below. Every chapter in this part will include theoretical background, as well as a detailed description of procedures, and examples of applications. 5. Earth-based observations Chapter 5 will review the remote sensing methodologies available in the literature to quantify and monitor water erosion phenomena. It will review the use of multi-spectral and hyper-spectral indices for the study of soil attributes, as well as soil movement due to land soil loss and land degradation. Next, it will describe classification tools for mapping catchment surface coverage, by means of machine-learning techniques as well as unmixing techniques to map surface components in the sub-pixel domain. It will demonstrate the use of remotely sensed data and DEMs for catchment-scale analysis of contributing areas and channel network extraction (Svoray 2004), and will demonstrate the use of 3D data from aerial sources to simulate erosion processes. It will explain the methodology available in the field, and provide an analysis of the results obtained from this methodology, through maps and cross-section graphs. The data will include a temporal analysis of gully and piping volumes, changes in gully-head location (Svoray & Markovitch 2009), and volume differences over time. The chapter will also discuss changes in gully volumes, and the limitations of the methodology. It will review the use of the methods under various conditions of the parameter space, and provide examples of maps under various conditions (e.g., temporal), to show how context affects conclusions. We will demonstrate the data analysis using drone data and images from the new French-Israeli Vens satellite. 6. Predicting erosion risk: from expert knowledge to data mining The aim of Chapter 6 will be to present the reader with the core methods that are available for catchment scale analysis of soil water erosion – including spatially explicit analysis of topographic indices, expert-based systems, data-mining approaches, and fuzzy rule-based models. The first section will describe the principle of spatial interpolation of the topographic threshold (TT) (Svoray & Markovitch 2009). This will include the use of TT as a proxy for the critical threshold needed to initiate a concentrated flow, when water shear stress is greater than surface resistance. The effect of topography on soil loss will be compared with the effect of other factors (in particular, various types of human intervention, such tillage direction, and paved roads), to show that a more comprehensive approach is needed to predict water erosion at the catchment scale. The second section will present methods for identifying areas under threat of water erosion, using expert knowledge with a formalized methodology known as AHP (analytic hierarchical process). These predictions will be compared with actual measurements. The third section will present ways of predicting water erosion processes in the same catchment, using algorithms based on training data (i.e., data-mining approaches). As an example, expert-based and data-mining systems will be compared, and the pros and cons of each method (data- versus theory-driven) will be discussed. The fourth section will deal with process-based modeling, and surface analysis – starting with the use of fuzzy rule-based methodology as a language for simulating process-based models that are dynamic in space and time. It will discuss how the model is programmed, how the parameters are tuned, and how sensitivity analysis is applied, in a bid to determine the validity of weights. This section will also discuss how more advanced and physically based models, such as soilscape evolution models (Cohen et al. 2015) can replace fuzzy logic for process-based modeling. Finally, it will show how simpler models – such as calculation of water-flow directions, runoff accumulation, descriptions of extrusion systems, and sub-basin delimitation – can be effective in representing the plots of continuous phenomena that are represented by categorical variables, such as focal and zonal majority (Cohen et al. 2008). 7. Health of the remaining soil This chapter will discuss the formalization of spatial information on soil quality on-site, and its implications for the quality of the remaining soil in the wake of erosion processes. The chapter will describe the differences between distinct forms of representation (continuous versus discrete), and will quantify spatial autocorrelation through various methods – such as Moran’s I, and variogram envelope analysis. It will also discuss spatial sampling approaches – such including random, stratified random, and matrix – as well as the sample size question, and the effect of noise. These aspects will be demonstrated by using GIS layers and procedures to show how to actually produce the layers for the stratified random approach. Spatial interpolation techniques – such as ordinary kriging, kriging with external drift, and co-kriging – will be discussed as tools for predicting spatial variation in soil quality. This chapter will extend the interpolation of soil characteristics in soil quality mapping, based on a soil health index developed in the literature – which, until recently (Svoray et al. 2015), had only been applied to the local scale of point measurements. The limited ability to upscale soil health mapping from point measurements to wide agricultural areas is a major gap in the research (Poesen 2018), and will also be discussed in this chapter. Another important direction to be developed in this chapter is the idea of using point measurements data to adjust for bias in remotely sensed data. In this approach, data extracted from satellite images (including data on attributes such as organic matter) may be used to bias predictions of soil quality from geostatistics. The idea to use such remotely sensed data as ancillary information for catchment scale analysis was developed in Svoray (2004), and will be further extended here. The outcome of this chapter will therefore be a procedure for sampling soil characteristics in the field, analyzing their spatial auto-correlation characteristics, interpolating them in the field, and creating soil-quality maps over wide regions. The advantages and the limitations of this process will be further discussed in this section, and clearly presented to the reader, along with the necessary information for applying the methodology. In this regard, Chapter 7 is of major importance from a scientific point of view, as it will provide the community with a methodology for studying the effect of water erosion on remaining soil. From an applied standpoint, it will also provide farmers and professionals with a tool for estimating the state and dynamics of their soil, in a spatially explicit fashion. 8. Decision-making Chapter 8 will be the most application-oriented section of the book, as it will provide the reader with up-to-date study materials on SDSS methodologies that may be used to determine land management practices and to allocate agricultural machinery to combat land management issues. It will begin with a description of the factors to be taken into account when developing SDSSs for catchment analysis, in an effort to control water erosion and reduce soil loss. To this end, it will review existing SDSSs, and show how these systems may provide recommendations for soil conservation treatments that can reduce erosion risks under various scenarios, while enhancing the ecosystem services provided in the catchment. This will be followed by an example that will demonstrate the development of just such an SDSS for an at-risk agricultural catchment. The demonstration will be introduced by means of the German decision-making system GISCAME (Geographic Information System, Cellular Automaton, Multi-Criteria Evaluation; http://www.giscame.com/giscame/index.html), which can easily be applied by the reader in at-risk catchments. The sample system will introduce principles gathered from the literature and expert knowledge, as a tool. In terms of evaluating the system’s performance, this section will present various measures, including the confusion matrix and the kappa coefficient. It will demonstrate their use for gauging the reliability of input data from remote sensing, and show the SDSS’s performance under various rainfall intensity scenarios. Finally, the pros and cons of the system will be discussed, to show the reader the risks and benefits that the automatic application of such a system can provide.