Development of a Variogram Procedure to Identify Spatial Outliers using a Supplemental Digital Elevation Model

Abstract

When using a ground water elevation dataset for the development of a ground water model, it is prudent to first evaluate the quality of the data before using it in a ground water model. However, it may not be practical to evaluate every data point when working with large datasets associated with a regional model. To isolate misrepresentative points in a large dataset, a graphical technique has been developed which examines the variance of ground water elevation values from an unconfined aquifer to identify points with high variance. The potential outliers identified using the graphical variogram process are subsequently evaluated by reviewing well borings, well installation records, and available time series of water level measurements to retain or reject outlier status. Supplemental ground elevation data from a digital elevation model is used to create a threshold on the experimental variogram of the ground water elevation data. This process is verified using a developed synthetic ground water dataset, then applied to a case study at the Fort Leonard Wood Military Reservation, Missouri. The method showed good results in identifying points that were justified for removal upon inspection of the available records and provides recommendations based on common causes of error. With this method's reliance on both, an experimental variogram of measured water levels and a binned variogram of ground elevation measures, it naturally fits as a preprocessing step that can be applied prior to kriging.

Department(s)

Geosciences and Geological and Petroleum Engineering

Comments

This work was supported by the U.S. Army Corps of Engineers grant no. W912DQ-14-2-0003-001.

Keywords and Phrases

Gradient; Outlier detection; Spatial outlier; Supplemental topographic data; Variogram

International Standard Serial Number (ISSN)

2589-9155

Document Type

Article - Journal

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2019 Elsevier B.V., All rights reserved.

Publication Date

01 Apr 2019

Share

 
COinS