A New Soil Sampling Design in Coastal Saline Region Using EM38 and VQT Method

Abstract

Spatial sampling design based on the variability and distribution of soil properties is an important issue with the progress in precision agriculture and soil ecology. Electromagnetic induction (type EM38) and variance quad-tree (VQT) method were both applied to optimize the sampling scheme of soil salinity in a coastal reclamation field in north Jiangsu Province, China. Apparent soil electrical conductivity (EC a ) measured with EM38 was used as an ancillary variable and the spatial distribution of EC a was used as priori information. The process and result of VQT algorithm analysis was illustrated and the obtained sampling strategy was validated using observed soil salinity. Then the spatial precision and sampling efficiency were evaluated. The result indicated that the spatial distribution of soil salinity produced with the VQT scheme was quite similar to that produced with total sampling sites, while sampling quantity of the former was reduced to approximately 1/2 of the latter. The spatial precision of VQT scheme was considerably higher than that of traditional grid method with respect to the same sampling number, and fewer samples were required for VQT scheme to obtain the same precision level. A 17.3% increase in sampling efficiency was achieved by VQT over grid method at the precision level of 90%. The VQT method was proved to be more efficient and economical because it can sample intensively or sparsely according to variation status in local areas. The associated application of EM38 and VQT method provides efficient tools and theoretical basis for saving sampling cost and improving sampling efficiency in coastal saline region and enriching soil ecology.

Field‐scale apparent soil electrical conductivity

“…(2000) response survey sampling design (Corwin et al., 2010; Fitzgerald, Lesch, Barnes, & Luckett, 2006; Guo et al., 2015), Van Groenigen et al. (2000); Lesch, 2005a) minimization of a weighted means of the shortest distance (Barca et al., 2015; Brus & Heuvelink, 2007; Castrignanò, Morari, Fiorentino, Pagliarin, & Brenna, 2008; Debba, Van Ruitenbeek, Van Der Meer, Carranza, & Stein, 2005; Scudiero, Deiana, Teatini, Cassiani, & Morari, 2011), Minasny, McBratney, and Walvoort (2007) variance Quad‐Tree algorithm (Li, Zhou, Wu, Feng, & Li, 2007; Yao et al., 2012), balanced sampling (Brus, 2015), and a special case of the latter known as Minasny and McBratney's (2006) conditioned Latin hyper‐cube (Clifford, Payne, Pringle, Searle, & Butler, 2014; Kidd, Malone, McBratney, Minasny, & Webb, 2015; Ließ, 2015).…”

Review of soil salinity assessment for agriculture across multiple scales using proximal and/or remote sensors

“…Numerous sampling strategies have been presented in the literature, including, but not limited to: (i) Lesch et al's (2000) response surface sampling design (Fitzgerald et al, 2006;Guo et al, 2016;Lesch, 2005), (ii) Van Groenigen et al's (2000) minimization of a weighted means of the shortest distance (Barca et al, 2015;Brus and Heuvelink, 2007;Debba et al, 2005;Scudiero et al, 2011Scudiero et al, , 2016a, (iii) Minasny et al's (2007) variance Quad-Tree algorithm (Yan et al, 2007;Yao et al, 2012), (iv) stratified random sampling , balanced sampling (Brus, 2015), and (v) a special case of the latter known as Minasny and McBratney's (2006) conditioned Latin hypercube (Clifford et al, 2014;Kidd et al, 2015;Ließ, 2015). Currently, the most widely used sampling strategy for the characterization of soil spatial variability with EC a -directed soil sampling is the response surface sampling design in the ESAP software developed by Lesch et al (2000).…”

The Application of EM38: Determination of Soil Parameters, Selection of Soil Sampling Points and Use in Agriculture and Archaeology

“…However, the reduction of sampling points from 116 to 30 to 15 points resulted in a loss of accuracy, but this loss may not have an economic or management consequence to the producer. Yao et al [ 242 ] described a completely new method based on Minasny and McBratney [ 246 ]. The authors developed the application of the VQT (variance quad-tree) method on sampling design with the digital elevation model and its derivatives and Landsat TM images.…”