Soil-landscape modelling and spatial prediction of soil attributes

Gessler, Paul E.; Moore, Ian D.; McKenzie, N. J.; Ryan, Philip

doi:10.1080/02693799508902047

Cited by 552 publications

(335 citation statements)

References 14 publications

Supporting

Mentioning

317

Contrasting

Unclassified

Order By: Relevance

“…Generally, geostatistical and statistical methods are used in predicting soil attributes (S a ), where the predicted outputs are continuous values (Scull et al, 2003). Logistic regression has been used to predict the presence or absence of soil features, such as an E horizon, and fuzzy logic has been used to predict soil classes (Odeh et al, 1992;Gessler et al, 1995).…”

Section: Soil Spatial Prediction Functionsmentioning

confidence: 99%

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Stum

2010

Digital Soil Mapping

View full text Add to dashboard Cite

Initial soil surveys are incomplete for large tracts of public land in the western USA. Digital soil mapping offers a quantitative approach as an alternative to traditional soil mapping. I sought to predict soil classes across an arid to semiarid watershed of western Utah by applying random forests (RF) and using environmental covariates derived from Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and digital elevation models (DEM). Random forests are similar to classification and regression trees (CART).However, RF is doubly random. Many (e.g., 500) weak trees are grown (trained) independently because each tree is trained with a new randomly selected bootstrap sample, and a random subset of variables is used to split each node. To train and validate the RF trees, 561 soil descriptions were made in the field. An additional 111 points were added by case-based reasoning using aerial photo interpretation. As RF makes classification decisions from the mode of many independently grown trees, model iii uncertainty can be derived. The overall out of the bag (OOB) error was lower without weighting of classes; weighting increased the overall OOB error and the resulting output did not reflect soil-landscape relationships observed in the field. The final RF model had an OOB error of 55.2% and predicted soils on landforms consistent with soil-landscape relationships. The OOB error for individual classes typically decreased with increasing class size. In addition to the final classification, I determined the second and third most likely classification, model confidence, and the hypothetical extent of individual classes.Pixels that had high possibility of belonging to multiple soil classes were aggregated using a minimum confidence value based on limiting soil features, which is an effective and objective method of determining membership in soil map unit associations and complexes mapped at the 1:24,000 scale. Variables derived from both DEM and Landsat 7 ETM+ sources were important for predicting soil classes based on Gini and standard measures of variable importance and OOB errors from groves grown with exclusively DEM-or Landsat-derived data. Random forests was a powerful predictor of soil classes and produced outputs that facilitated further understanding of soil-landscape relationships.(147 pages) iv ACKNOWLEDGMENTS

show abstract

Section: Soil Spatial Prediction Functionsmentioning

confidence: 99%

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Stum

2010

Digital Soil Mapping

View full text Add to dashboard Cite

show abstract

“…In the 1990s, with the emerging of GIS and remote sensing technologies, soil surveyors became interested to use exhaustively mapped secondary variables to directly map soil variables. The first applications were based on the use of simple linear regression models between terrain attribute maps and soil parameters (Gessler et al, 1995;Moore et al, 1993). In the next phase, the predictors were extended to a set of environmental variables and remote sensing images.…”

Section: Introductionmentioning

confidence: 99%

A generic framework for spatial prediction of soil variables based on regression-kriging

Hengl¹,

Heuvelink

Stein³

2004

Geoderma

881

558

View full text Add to dashboard Cite

A methodological framework for spatial prediction based on regression-kriging is described and compared with ordinary kriging and plain regression. The data are first transformed using logit transformation for target variables and factor analysis for continuous predictors (auxiliary maps). The target variables are then fitted using step-wise regression and residuals interpolated using kriging. A generic visualisation method is used to simultaneously display predictions and associated uncertainty. The framework was tested using 135 profile observations from the national survey in Croatia, divided into interpolation (100) and validation sets (35). Three target variables: organic matter, pH in topsoil and topsoil thickness were predicted from six relief parameters and nine soil mapping units. Prediction efficiency was evaluated using the mean error and root mean square error (RMSE) of prediction at validation points. The results show that the proposed framework improves efficiency of predictions. Moreover, it ensured normality of residuals and enforced prediction values to be within the physical range of a variable. For organic matter, it achieved lower relative RMSE than ordinary kriging (53.3% versus 66.5%). For topsoil thickness, it achieved a lower relative RMSE (66.5% versus 83.3%) and a lower bias than ordinary kriging (0.15 versus 0.69 cm). The prediction of pH in topsoil was difficult with all three methods. This framework can adopt both continuous and categorical soil variables in a semi-automated or automated manner. It opens a possibility to develop a bundle algorithm that can be implemented in a GIS to interpolate soil profile data from existing datasets. D

show abstract

“…Whereas relief was demonstrated as a useful and dominant predictive variable on the spatial distribution of soils and associated thicknesses (e.g., [Huggett, 1975] , Bourennane, 1997, [Heimsath et al, 1999] and [King et al, 1999] ), few studies linked mathematically the morphologies of the anthropogenic features to their associated soil thicknesses. The easy acquisition of elevation data for large-scale areas makes its use very common for soil mapping ( [Odeh et al, 1994] , [Gessler et al, 1995] , [Isambert et al, 1997] and [Grinand et al, 2008] ).…”

Section: Introductionmentioning

confidence: 99%

Classification and mapping of anthropogenic landforms on cultivated hillslopes using DEMs and soil thickness data — Example from the SW Parisian Basin, France

et al. 2011

View full text Add to dashboard Cite

Macaire. Classification and mapping of anthropogenic landforms on cultivated hillslopes using DEMs and soil thickness data -Example from the SW Parisian Basin, France. Geomorphology, Elsevier, 2011Elsevier, , 135 (1-2), pp.8-20. 10.1016Elsevier, /j.geomorph.2011 Classification and mapping of anthropogenic landforms on cultivated hillslopes using DEMs and soil thickness data -Example from the SW Parisian Basin, France AbstractThis study focuses on linear anthropogenic landforms of decametric width on cultivated hillslopes and their relations to soil thickness variability. The 16 ha study area shows a rolling topography supported by Cretaceous chalk of the SW Parisian Basin, France. Two types of landforms were identified: lynchets, similar to those described as soil terraces occurring on downslope field parts in other contexts, and undulations, linear, convex landforms that cut across fields. Accurate DEM construction and a detailed soil thickness survey were performed all over the study area. Soil samples were classified considering their location on specific types of anthropogenic landforms. The Classification Tree (CT) method was applied to assess whether lynchets and undulations can be discriminated through morphometric attributes (slope, curvature, profile curvature and planform curvature) and soil thickness (CT soil ) or through morphometric attributes only (CT topo ). The CT application establishes predictive classification models to map the spatial distribution of lynchets and undulations over the whole study area. The validation results of the CT soil and CT topo applications show model efficiencies of 83% and 67%, respectively. Both models performed well for lynchets. Errors arise mainly from difficulties in unequivocally discriminating gently convex undulations and undifferentiated surfaces, especially when soil thickness is not accounted for. Mean values of soil thickness are 1.08, 0.62 and 0.45 m in lynchets, undulations and undifferentiated areas, respectively. The general shape of the thickened soil is characteristic to each type of anthropogenic landform. Multi-temporal mapping of field border networks shows that undulations are linked to borders that were removed during the latest land consolidation. Lynchets are associated with current field borders. Lynchets and undulations, which cover 39% of the study area, define topographic indicators of human-induced soil accumulations. The method involves perspectives for efficiently mapping and quantifying the anthropogenically modified spatial variability of soil thickness on agricultural hillsides.

show abstract

Soil-landscape modelling and spatial prediction of soil attributes

Cited by 552 publications

References 14 publications

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

A generic framework for spatial prediction of soil variables based on regression-kriging

Classification and mapping of anthropogenic landforms on cultivated hillslopes using DEMs and soil thickness data — Example from the SW Parisian Basin, France

Contact Info

Product

Resources

About