Abstract
The aim of the study was to analyse spatial variability of selected parameters of subsurface waters in the area of approximately 10 ha, located in the valley of the Ciemięga River in the village of Snopków, near Lublin, Poland. For the purpose of this study, nine sections were delimited, each with four points of collecting groundwater. In the groundwater samples, there were measured {text{NH}}_{4}^{ + }, {text{NO}}_{3}^{-} , and {text{NO}}_{2}^{-}. Due to the small number of samples, the analysis was limited to deterministic interpolation methods. The following methods were compared using leave-one-out cross-validation procedure: triangulation, inverse distance weighting, radial base function, and modified Shepard’s method. The methods which proved to be optimal were used to create spatial variability maps of the analysed parameters. Spatial interpolation and visualization of the results were performed in Surfer ver.16, and other calculations were conducted using R software.
Highlights
Geostatistical methods have become a key element of studies in the areas of science where spatial aspects play an essential role
Whilst there was not a universal superior method, the most accurate results were provided by Co-ordinary kriging (OK). In contrast to these related works, the present paper focuses only on deterministic interpolation methods
Our study revealed that the performance of Radial Basis Function (RBF) and MS methods was superior to inverse distance weighting (IDW)
Summary
International Journal of Environmental Research (2019) 13:679–687 on the minimum sample size, sampling schedule, distribution of an examined feature, etc. There are several variants of stochastic methods, which are referred to as kriging methods. The choice of an appropriate variant should be determined by the type of spatial relation of the examined phenomenon and fulfilling of the assumptions (appropriate number of samples, data distribution, stationarity, and isotropy). The advantage of stochastic approach is that it can be used both to interpolate the values at unsampled locations and to model uncertainty or errors of the estimated surface. Deterministic interpolation methods do not incorporate such errors. They predict solely the values at unsampled locations within region of interest using appropriate criteria for selecting the optimal degree of smoothing or similarity
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
