Principal Component Analysis of the Spatiotemporal Pattern of Soil Moisture and Apparent Electrical Conductivity. Issue 10 (11th May 2017)

• Record Type: Journal Article
• Authors: Martini, Edoardo; Wollschläger, Ute; Musolff, Andreas; Werban, Ulrike; Zacharias, Steffen
• Is Part Of: Vadose zone journal. Volume 16:Issue 10(2017)
• Page Start: 1
• Page End: 12
• Publication Date: 2017-05-11
• Subjects: Soil science -- Periodicals; Zone of aeration -- Periodicals; Groundwater flow -- Periodicals; Groundwater flow; Zone of aeration; Periodicals; Electronic journals; 631.4
• DOI: 10.2136/vzj2016.12.0129
• Languages: English
• ISSNs: 1539-1663

Abstract

Abstract : Core Ideas: PCA identified patterns within collocated time‐lapse measurements of θ and ECa. The factors controlling the observed spatial patterns of θ and ECa were quantified. Results demonstrate the nonstationary control of the spatial pattern of θ and ECa. Characterizing the spatial and temporal patterns of soil properties and states such as soil moisture (θ) remains an important challenge in environmental monitoring. At the Schäfertal hillslope site, the spatial patterns of θ measured by a distributed monitoring network and those of apparent electrical conductivity (ECa) measured by electromagnetic induction were characterized based on an integrated monitoring approach, and their possible controlling factors were investigated. With this study, we aimed to quantify the factors controlling the observed spatial patterns of θ and ECa and their interrelation. A principal component analysis was used to identify patterns within a data set comprising θ measured on seven dates within one hydrological year at 40 locations (three depths each) and ECa extracted from spatial maps for the same positions and dates. The first three independent principal components were all important for characterizing the spatial organization of topsoil moisture and its temporal changes. The dominant pattern responded to time‐invariant soil attributes such as spatial soil properties and terrain attributes and could explain the spatial organization of ECa only on four of the seven measurement dates. The second and third principal components described the spatial reorganization of the patterns in response to θ dynamics within the soil profile and water removal processes, respectively, and showed distinct time‐varying effects on the spatial pattern of θ and ECa. Our results can help with designing field monitoring campaigns and improving modeling approaches by providing insights into the nonstationary control of static and dynamic attributes on the spatial pattern of θ and ECa.