Digital soil mapping: strategy for data pre-processing

The region of greatest variability on soil maps is along the edge of their polygons, causing disagreement among pedologists about the appropriate description of soil classes at these locations. The objective of this work was to propose a strategy for data pre-processing applied to digital soil mapping (DSM). Soil polygons on a training map were shrunk by 100 and 160 m. This strategy prevented the use of covariates located near the edge of the soil classes for the Decision Tree (DT) models. Three DT models derived from eight predictive covariates, related to relief and organism factors sampled on the original polygons of a soil map and on polygons shrunk by 100 and 160 m were used to predict soil classes. The DT model derived from observations 160 m away from the edge of the polygons on the original map is less complex and has a better predictive performance.

Comparison between artificial neural networks and maximum likelihood classification in digital soil mapping

Soil surveys are the main source of spatial information on soils and have a range of different applications, mainly in agriculture. The continuity of this activity has however been severely compromised, mainly due to a lack of governmental funding. The purpose of this study was to evaluate the feasibility of two different classifiers (artificial neural networks and a maximum likelihood algorithm) in the prediction of soil classes in the northwest of the state of Rio de Janeiro. Terrain attributes such as elevation, slope, aspect, plan curvature and compound topographic index (CTI) and indices of clay minerals, iron oxide and Normalized Difference Vegetation Index (NDVI), derived from Landsat 7 ETM+ sensor imagery, were used as discriminating variables. The two classifiers were trained and validated for each soil class using 300 and 150 samples respectively, representing the characteristics of these classes in terms of the discriminating variables. According to the statistical tests, the accuracy of the classifier based on artificial neural networks (ANNs) was greater than of the classic Maximum Likelihood Classifier (MLC). Comparing the results with 126 points of reference showed that the resulting ANN map (73.81 %) was superior to the MLC map (57.94 %). The main errors when using the two classifiers were caused by: a) the geological heterogeneity of the area coupled with problems related to the geological map; b) the depth of lithic contact and/or rock exposure, and c) problems with the environmental correlation model used due to the polygenetic nature of the soils. This study confirms that the use of terrain attributes together with remote sensing data by an ANN approach can be a tool to facilitate soil mapping in Brazil, primarily due to the availability of low-cost remote sensing data and the ease by which terrain attributes can be obtained.

An appropriate data set size for digital soil mapping in Erechim, Rio Grande do Sul, Brazil

Digital information generates the possibility of a high degree of redundancy in the data available for fitting predictive models used for Digital Soil Mapping (DSM). Among these models, the Decision Tree (DT) technique has been increasingly applied due to its capacity of dealing with large datasets. The purpose of this study was to evaluate the impact of the data volume used to generate the DT models on the quality of soil maps. An area of 889.33 km² was chosen in the Northern region of the State of Rio Grande do Sul. The soil-landscape relationship was obtained from reambulation of the studied area and the alignment of the units in the 1:50,000 scale topographic mapping. Six predictive covariates linked to the factors soil formation, relief and organisms, together with data sets of 1, 3, 5, 10, 15, 20 and 25 % of the total data volume, were used to generate the predictive DT models in the data mining program Waikato Environment for Knowledge Analysis (WEKA). In this study, sample densities below 5 % resulted in models with lower power of capturing the complexity of the spatial distribution of the soil in the study area. The relation between the data volume to be handled and the predictive capacity of the models was best for samples between 5 and 15 %. For the models based on these sample densities, the collected field data indicated an accuracy of predictive mapping close to 70 %.

Digital soil mapping: evaluation of sampling systems for soil surveys and refinement of soil maps at lower cost using legacy data

In soil surveys, several sampling systems can be used to define the most representative sites for sample collection and description of soil profiles. In recent years, the conditioned Latin hypercube sampling system has gained prominence for soil surveys. In Brazil, most of the soil maps are at small scales and in paper format, which hinders their refinement. The objectives of this work include: (i) to compare two sampling systems by conditioned Latin hypercube to map soil classes and soil properties; (II) to retrieve information from a detailed scale soil map of a pilot watershed for its refinement, comparing two data mining tools, and validation of the new soil map; and (III) to create and validate a soil map of a much larger and similar area from the extrapolation of information extracted from the existing soil map. Two sampling systems were created by conditioned Latin hypercube and by the cost-constrained conditioned Latin hypercube. At each prospection place, soil classification and measurement of the A horizon thickness were performed. Maps were generated and validated for each sampling system, comparing the efficiency of these methods. The conditioned Latin hypercube captured greater variability of soils and properties than the cost-constrained conditioned Latin hypercube, despite the former provided greater difficulty in field work. The conditioned Latin hypercube can capture greater soil variability and the cost-constrained conditioned Latin hypercube presents great potential for use in soil surveys, especially in areas of difficult access. From an existing detailed scale soil map of a pilot watershed, topographical information for each soil class was extracted from a Digital Elevation Model and its derivatives, by two data mining tools. Maps were generated using each tool. The more accurate of these tools was used for extrapolation of soil information for a much larger and similar area and the generated map was validated. It was possible to retrieve the existing soil map information and apply it on a larger area containing similar soil forming factors, at much low financial cost. The KnowledgeMiner tool for data mining, and ArcSIE, used to create the soil map, presented better results and enabled the use of existing soil map to extract soil information and its application in similar larger areas at reduced costs, which is especially important in development countries with limited financial resources for such activities, such as Brazil.

The use of agricultural soil maps and the status of agricultural soil mapping in the United States. Presented at the Twenty-Eighth annual meeting, 1948.

Cover title: Engineering use of agricultural soil maps.

Digital soil mapping using reference area and artificial neural networks.

Digital soil mapping is an alternative for the recognition of soil classes in areas where pedological surveys are not available. The main aim of this study was to obtain a digital soil map using artificial neural networks (ANN) and environmental variables that express soillandscape relationships. This study was carried out in an area of 11,072 ha located in the Barra Bonita municipality, state of São Paulo, Brazil. A soil survey was obtained from a reference area of approximately 500 ha located in the center of the area studied. With the mapping units identified together with the environmental variables elevation, slope, slope plan, slope profile, convergence index, geology and geomorphic surfaces, a supervised classification by ANN was implemented. The neural network simulator used was the Java NNS with the learning algorithm "back propagation." Reference points were collected for evaluating the performance of the digital map produced. The occurrence of soils in the landscape obtained in the reference area was observed in the following digital classification: medium-textured soils at the highest positions of the landscape, originating from sandstone, and clayey loam soils in the end thirds of the hillsides due to the greater presence of basalt. The variables elevation and slope were the most important factors for discriminating soil class through the ANN. An accuracy level of 82% between the reference points and the digital classification was observed. The methodology proposed allowed for a preliminary soil classification of an area not previously mapped using mapping units obtained in a reference area

Evaluation of statistical and geostatistical models of digital soil properties mapping in tropical mountain regions

Soil properties have an enormous impact on economic and environmental aspects of agricultural production. Quantitative relationships between soil properties and the factors that influence their variability are the basis of digital soil mapping. The predictive models of soil properties evaluated in this work are statistical (multiple linear regression-MLR) and geostatistical (ordinary kriging and co-kriging). The study was conducted in the municipality of Bom Jardim, RJ, using a soil database with 208 sampling points. Predictive models were evaluated for sand, silt and clay fractions, pH in water and organic carbon at six depths according to the specifications of the consortium of digital soil mapping at the global level (GlobalSoilMap). Continuous covariates and categorical predictors were used and their contributions to the model assessed. Only the environmental covariates elevation, aspect, stream power index (SPI), soil wetness index (SWI), normalized difference vegetation index (NDVI), and b3/b2 band ratio were significantly correlated with soil properties. The predictive models had a mean coefficient of determination of 0.21. Best results were obtained with the geostatistical predictive models, where the highest coefficient of determination 0.43 was associated with sand properties between 60 to 100 cm deep. The use of a sparse data set of soil properties for digital mapping can explain only part of the spatial variation of these properties. The results may be related to the sampling density and the quantity and quality of the environmental covariates and predictive models used.

A new approach to soil classification mapping based on the spatial distribution of soil properties

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Using Imaging Spectroscopy to study soil properties

Imaging Spectroscopy (IS) is a promising tool for studying soil properties in large spatial domains. Going from point to image spectrometry is not only a journey from micro to macro scales, but also a long stage where problems such as dealing with data having a low signal-to-noise level, contamination of the atmosphere, large data sets, the BRDF effect and more are often encountered. In this paper we provide an up-to-date overview of some of the case studies that have used IS technology for soil science applications. Besides a brief discussion on the advantages and disadvantages of IS for studying soils, the following cases are comprehensively discussed: soil degradation (salinity, erosion, and deposition), soil mapping and classification, soil genesis and formation, soil contamination, soil water content, and soil swelling. We review these case studies and suggest that the 15 data be provided to the end-users as real reflectance and not as raw data and with better signal-to-noise ratios than presently exist. This is because converting the raw data into reflectance is a complicated stage that requires experience, knowledge, and specific infrastructures not available to many users, whereas quantitative spectral models require good quality data. These limitations serve as a barrier that impedes potential end-users, inhibiting researchers from trying this technique for their needs. The paper ends with a general call to the soil science audience to extend the utilization of the IS technique, and it provides some ideas on how to propel this technology forward to enable its widespread adoption in order to achieve a breakthrough in the field of soil science and remote sensing. (C) 2009 Elsevier Inc. All rights reserved.

Australia-wide predictions of soil properties using decision trees

This paper describes the construction of Australia-wide soil property predictions from a compiled national soils point database. Those properties considered include pH, organic carbon, total phosphorus, total nitrogen, thickness. texture, and clay content. Many of these soil properties are used directly in environmental process modelling including global climate change models. Models are constructed at the 250-m resolution using decision trees. These relate the soil property to the environment through a suite of environmental predictors at the locations where measurements are observed. These models are then used to extend predictions to the continental extent by applying the rules derived to the exhaustively available environmental predictors. The methodology and performance is described in detail for pH and summarized for other properties. Environmental variables are found to be important predictors, even at the 250-m resolution at which they are available here as they can describe the broad changes in soil property.

Variability of apparently homogeneous soilscapes in São Paulo state, Brazil: II. quality of soil maps

The quality of semi-detailed (scale 1:100.000) soil maps and the utility of a taxonomically based legend were assessed by studying 33 apparently homogeneous fields with strongly weathered soils in two regions in São Paulo State: Araras and Assis. An independent data set of 395 auger sites was used to determine purity of soil mapping units and analysis of variance within and between mapping units and soil classification units. Twenty three soil profiles were studied in detail. The studied soil maps have a high purity for some legend criteria, such as B horizon type (> 90%) and soil texture class (> 80%). The purity for the "trophic character" (eutrophic, dystrophic, allic) was only 55% in Assis. It was 88% in Araras, where many soil units had been mapped as associations. In both regions, the base status of clay-textured soils was generally better than suggested by the maps. Analysis of variance showed that mapping was successful for "durable" soil characteristics such as clay content (> 80% of variance explained) and cation exchange capacity (≥ 50% of variance explained) of 0-20 and 60-80 cm layers. For soil characteristics that are easily modified by management, such as base saturation of the 0-20 cm layer, the maps had explained very little (< 15%) of the total variance in the study areas. Intermediate results were obtained for base saturation of the 60-80 cm layer (56% in Assis; 42% in Araras). Variance explained by taxonomic groupings that formed the basis for the legend of the soil maps was similar to, often even smaller than, variance explained by mapping units. The conclusion is that map boundaries have been very carefully located, but descriptions of mapping units could be improved. In future mappings, this could possibly be done at low cost by (a) bulk sampling to remove short range variation and enhance visualization of spatial patterns at distances > 100 m; (b) taking advantage of correlations between easily measured soil characteristics and chemical soil properties and, (c) unbending the link between legend criteria and a taxonomic system. The maps are well suited to obtain an impression of land suitability for high-input farming. Additional field work and data on former land use/management are necessary for the evaluation of chemical properties of surface horizons.

Building predictive models of soil particle-size distribution

Is it possible to build predictive models (PMs) of soil particle-size distribution (psd) in a region with complex geology and a young and unstable land-surface? The main objective of this study was to answer this question. A set of 339 soil samples from a small slope catchment in Southern Brazil was used to build PMs of psd in the surface soil layer. Multiple linear regression models were constructed using terrain attributes (elevation, slope, catchment area, convergence index, and topographic wetness index). The PMs explained more than half of the data variance. This performance is similar to (or even better than) that of the conventional soil mapping approach. For some size fractions, the PM performance can reach 70 %. Largest uncertainties were observed in geologically more complex areas. Therefore, significant improvements in the predictions can only be achieved if accurate geological data is made available. Meanwhile, PMs built on terrain attributes are efficient in predicting the particle-size distribution (psd) of soils in regions of complex geology.

Comparison between detailed digital and conventional soil maps of an area with complex geology

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.

Artificial neural networks applied for soil class prediction in mountainous landscape of the Serra do Mar¹

Soil information is needed for managing the agricultural environment. The aim of this study was to apply artificial neural networks (ANNs) for the prediction of soil classes using orbital remote sensing products, terrain attributes derived from a digital elevation model and local geology information as data sources. This approach to digital soil mapping was evaluated in an area with a high degree of lithologic diversity in the Serra do Mar. The neural network simulator used in this study was JavaNNS and the backpropagation learning algorithm. For soil class prediction, different combinations of the selected discriminant variables were tested: elevation, declivity, aspect, curvature, curvature plan, curvature profile, topographic index, solar radiation, LS topographic factor, local geology information, and clay mineral indices, iron oxides and the normalized difference vegetation index (NDVI) derived from an image of a Landsat-7 Enhanced Thematic Mapper Plus (ETM+) sensor. With the tested sets, best results were obtained when all discriminant variables were associated with geological information (overall accuracy 93.2 - 95.6 %, Kappa index 0.924 - 0.951, for set 13). Excluding the variable profile curvature (set 12), overall accuracy ranged from 93.9 to 95.4 % and the Kappa index from 0.932 to 0.948. The maps based on the neural network classifier were consistent and similar to conventional soil maps drawn for the study area, although with more spatial details. The results show the potential of ANNs for soil class prediction in mountainous areas with lithological diversity.

Prediction of soil orders with high spatial resolution: response of different classifiers to sampling density

The objective of this work was to evaluate sampling density on the prediction accuracy of soil orders, with high spatial resolution, in a viticultural zone of Serra Gaúcha, Southern Brazil. A digital elevation model (DEM), a cartographic base, a conventional soil map, and the Idrisi software were used. Seven predictor variables were calculated and read along with soil classes in randomly distributed points, with sampling densities of 0.5, 1, 1.5, 2, and 4 points per hectare. Data were used to train a decision tree (Gini) and three artificial neural networks: adaptive resonance theory, fuzzy ARTMap; self‑organizing map, SOM; and multi‑layer perceptron, MLP. Estimated maps were compared with the conventional soil map to calculate omission and commission errors, overall accuracy, and quantity and allocation disagreement. The decision tree was less sensitive to sampling density and had the highest accuracy and consistence. The SOM was the less sensitive and most consistent network. The MLP had a critical minimum and showed high inconsistency, whereas fuzzy ARTMap was more sensitive and less accurate. Results indicate that sampling densities used in conventional soil surveys can serve as a reference to predict soil orders in Serra Gaúcha.