Elsevier

Pedosphere

Volume 21, Issue 3, June 2011, Pages 339-350
Pedosphere

Application of a Digital Soil Mapping Method in Producing Soil Orders on Mountain Areas of Hong Kong Based on Legacy Soil Data

https://doi.org/10.1016/S1002-0160(11)60134-3Get rights and content

Abstract

Based on legacy soil data from a soil survey conducted recently in the traditional manner in Hong Kong of China, a digital soil mapping method was applied to produce soil order information for mountain areas of Hong Kong. Two modeling methods (decision tree analysis and linear discriminant analysis) were used, and their applications were compared. Much more effort was put on selecting soil covariates for modeling. First, analysis of variance (ANOVA) was used to test the variance of terrain attributes between soil orders. Then, a stepwise procedure was used to select soil covariates for linear discriminant analysis, and a backward removing procedure was developed to select soil covariates for tree modeling. At the same time, ANOVA results, as well as our knowledge and experience on soil mapping, were also taken into account for selecting soil covariates for tree modeling. Two linear discriminant models and four tree models were established finally, and their prediction performances were validated using a multiple jackknifing approach. Results showed that the discriminant model built on ANOVA results performed best, followed by the discriminant model built by stepwise, the tree model built by the backward removing procedure, the tree model built according to knowledge and experience on soil mapping, and the tree model built automatically. The results highlighted the importance of selecting soil covariates in modeling for soil mapping, and suggested the usefulness of methods used in this study for selecting soil covariates. The best discriminant model was finally selected to map soil orders for this area, and validation results showed that thus produced soil order map had a high accuracy.

References (27)

Cited by (15)

  • Digital soil mapping in a low-relief landscape to support wetland restoration decisions

    2020, Geoderma
    Citation Excerpt :

    Topographic wetness indices are useful in mapping the potential spatial distribution of soil saturation (Beven and Kirkby, 1979), a measure of important wetland processes (Lang et al., 2013; Rampi et al., 2014; Rodhe and Seibert, 1999). To date, DSM techniques used to classify soils have been applied predominantly in areas of heterogeneous topography and geomorphology (Häring et al., 2012; Nauman et al., 2014; Nauman and Thompson, 2014; Sun et al., 2011). Elevation data used to derive predictor variables are typically at resolutions of 10 m or coarser.

  • Digital soil mapping based on wavelet decomposed components of environmental covariates

    2017, Geoderma
    Citation Excerpt :

    Thus, other predictor selection methods might be helpful to make better use of wavelet components. Predictor selection is also an issue for DSM based on multi-scale predictors generated using other methods (Behrens et al., 2010a, 2014; Miller et al., 2015), as well as standard predictors (Lark et al., 2007; Sun et al., 2011). A data mining method usually has an ability to select predictors during modelling, but Miller et al. (2015) worried about whether it could find the optimal combination of ones.

  • Cladistic analysis of Chinese Soil Taxonomy

    2017, Geoderma Regional
    Citation Excerpt :

    As such, mapping Chinese soil is required for securing the resource for sustained agricultural and ecological benefits (McBratney et al., 2014) or as a contribution to the international collaboration for global environmental protection (e.g., Global Digital Soil Property Map project, Minasny and McBratney, 2010). To assist and improve soil mapping in China, the CST system has been widely used to map carbon (e.g., Yu et al., 2007; Shi et al., 2015) and salinity (e.g., Sheng et al., 2010) as well as class identification and prediction (e.g., Zhu et al., 2010; Sun et al., 2011). However, compared with the extensive research using the CST system to map Chinese soil, the CST has been little studied from a perspective of a natural classification.

  • Dealing with spatial outliers and mapping uncertainty for evaluating the effects of urbanization on soil: A case study of soil pH and particle fractions in Hong Kong

    2013, Geoderma
    Citation Excerpt :

    Soils of this study area were derived from seven kinds of parent materials: alluvial, alluvial and colluvial, boulder fan, boulder fan enriched by more recent alluvium, colluvium and alluvial fan, in situ, and mangrove swamp alluvial (Fig. 1(c)), according to Grant (1962). The soils are similar to Luviosls, Cambisols, Anthrosols and Acrisols of the World Reference Base for Soil Resources (Sun et al., 2011). Soil sampling was based on the soil series map established by Grant (1962), with totally 31 soil series.

  • Spatial disaggregation of complex soil map units: A decision-tree based approach in Bavarian forest soils

    2012, Geoderma
    Citation Excerpt :

    Therefore, we developed an approach applying comparably simple but class-specific models for the delineation of sub-areas. Almost all studies in which categorical map units were disaggregated considered a smaller number of classes (i.e. less than 10 classes: Behrens et al., 2010; Brus et al., 2008; Sun et al., 2011; 10 to 20 classes: Debella-Gilo and Etzelmüller, 2009; Hengl et al., 2007; Kempen et al., 2009; Moonjun et al., 2010; more than 20 classes: Grinand et al., 2008; Smith et al., 2010; Stum et al., 2010). Only Smith et al. (2010) predicted more than 100 classes, however not in a single model but with knowledge-based fuzzy classification rules for every class separately.

  • Sensitivity of digital soil maps based on FCM to the fuzzy exponent and the number of clusters

    2012, Geoderma
    Citation Excerpt :

    Then, empirical best linear unbiased predictor with residual maximum likelihood (REML-EBLUP) (Lark et al., 2006) was used instead of a regression model by maximum likelihood (Lark, 1999) to construct soil maps based on transformed memberships. Considering that not all of the terrain attributes influence the spatial distribution of SOM (Chai et al., 2008; Sun et al., 2011), REML-EBLUP was applied based on those memberships that were statistically significantly correlated with SOM. An analysis of REML-EBLUP was performed using geoR (Ribeiro and Diggle, 2001) in R (R Development Core Team, 2006).

View all citing articles on Scopus

Supported by the Public Policy Research of the Research Grants Council of Hong Kong, China (No. 2002-PPR-3), the Knowledge Innovation Program of the Chinese Academy of Sciences (No. KZCX2-YW-409), the National Natural Science Foundation of China (Nos. 40625001 and 40771092), and the Mini-AOE (Area of Excellence) Fund from the Hong Kong Baptist University, China (No. RC/AOE/08-09/01).

View full text