Skip to Main Content
 

Global Search Box

 
 
 

ETD Abstract Container

Abstract Header

Statistical Identification of Best Representative Examples of a Common Soil Series Using Environmental Covariates

Omar, Abdelmatloub
http://rave.ohiolink.edu/etdc/view?acc_num=osu1703616363658244

Abstract Details

2024, Doctor of Philosophy, Ohio State University, Environment and Natural Resources.
Soil profile sampling and description are an important part of soil survey, mapping, and classification. In soil mapping, representative soil profiles (pedons) are used to characterize the properties of the series. A representative soil profile is the one that has normative properties within the defined range. In traditional soil survey, the surveyor has to walk in the field, expose and describe a large number of pedons and reach a conclusion about the range of properties and those that are most representative of the series. Candidate representative pedons are then sampled and morphologically described for the classification purposes and may be sampled for detailed laboratory analysis. However, designating and describing the representative soil profile is challenging for many reasons. The variability in soil properties is one of the most important reasons. In many cases, “representative pedons” may not express normative properties, and there are few examples of representative pedons being selected statistically. The traditional way of soil survey is time-consuming, and it requires resources which have become limited. The technological advance in hardware, software, and acquiring and saving data paved the road to move from the conventional to the digital soil mapping. The environmental covariates driven from the digital elevation model have widely contributed to this transfer. The purpose of this study is to adapt a spatial-statistical model by which the best representative soil profile will be designated, spatially located, and morphologically described. In this study, we used a clustering algorithm trained with environmental covariate data and identified medoids from which the best representative combinations of covariates could be identified. The gSSURGO is the best database as a source of digital soil information related to the US Cooperative Soil Survey. It offers soil information in spatial and tabular formats which can be analyzed using modern methods. We took the advantage of the availability of the soil data from gSSURGO, and the Ohio soil dataset was downloaded. We collected seventeen environmental covariates that were widely used in previous studies for soil predictive models. These environmental covariates included general curvature, maximal curvature, minimal curvature, plan curvature, profile curvature, tangential curvature, total curvature, multiresolution ridge top flatness index (MRRTF), multiresolution valley bottom flatness index (MRBVF), slope, mis-slope, LS factor, valley depth, flow accumulation, stream power index, topographic wetness index, and convergence index. Using the ArcMap software, the soil raster was converted to point feature shapefile format. The point shapefile and the seventeen environmental covariates were simultaneously opened in the ArcMap. In ArcMap, the Extract Multi Values To Point tool was used to extract the information from the covariates. This process resulted in the attribute table of the Blount soil being expanded and all the information related to the environmental covariates were included in the table. The Blount soil shapefile was opened in R software. Due to memory limitations, a subset of 30,000 locations of Blount soil was randomly sampled. After editing and removing N/A values, the final number of locations was 26,721. The partitioning approach using CLARA algorithm was applied for clustering the soil dataset. We obtained eight clusters. For each cluster, the best one hundred observations were extracted based on the similarity to the medoids. All the clusters’ groups were exported in spatial format as shapefile to be explored in Arc-Map. We used the box plots to describe and compare the obtained clusters. We propose this approach for defining the best example of soil profiles at the catena level.
Rattan Lal (Committee Member)
M. Scott Demyan (Committee Member)
Brian K. Slater (Advisor)
122 p.
Agriculture; Environmental Studies; Natural Resource Management; Soil Sciences
Representative soil profiles; environmental covariates; CLARA algorithm

Recommended Citations

Citations

  • Omar, A. (2024). Statistical Identification of Best Representative Examples of a Common Soil Series Using Environmental Covariates [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1703616363658244

    APA Style (7th edition)

  • Omar, Abdelmatloub. Statistical Identification of Best Representative Examples of a Common Soil Series Using Environmental Covariates. 2024. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1703616363658244.

    MLA Style (8th edition)

  • Omar, Abdelmatloub. "Statistical Identification of Best Representative Examples of a Common Soil Series Using Environmental Covariates." Doctoral dissertation, Ohio State University, 2024. http://rave.ohiolink.edu/etdc/view?acc_num=osu1703616363658244

    Chicago Manual of Style (17th edition)

osu1703616363658244
15
© 2023, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.