Premium
Statistical Methods for Evaluating Soil Salinity Spatial and Temporal Variability
Author(s) -
Douaik Ahmed,
Van Meirvenne Marc,
Tóth Tibor
Publication year - 2007
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj2006.0083
Subject(s) - environmental science , salinity , spatial variability , soil science , soil salinity , hydrology (agriculture) , common spatial pattern , spatial ecology , soil water , geology , statistics , ecology , mathematics , geotechnical engineering , oceanography , biology
Monitoring soil salinity requires knowledge of its magnitude and its spatial and temporal variability. To characterize the spatiotemporal variability of soil salinity in a native sodic grassland in the east of Hungary, we applied several statistical methods. Within a 25‐ha study area, soil samples were taken repeatedly from 13 to 20 locations at 19 dates between November 1994 and June 2001 (with intervals between 2 and 9 mo). Electrical conductivity was measured both in the laboratory in a 1:2.5 soil/water suspension (EC 2.5 ) and in situ using a four‐electrode probe (EC a ). These measurements were converted, via calibration regressions, into predicted EC 2.5 *, which were compared with EC 2.5 in their ability to characterize the spatiotemporal variability of soil salinity. The temporal change in the mean soil salinity level between each subsequent two dates was evaluated using a paired t ‐test, a test of significance of the regression parameters based on the concept of temporal stability, and a temporal mean shift test. The static–dynamic (uniform–nonuniform) nature of the temporal change in the spatial pattern of soil salinity between two dates was evaluated using the same concept of temporal stability and a spatial shift test. For either type of temporal change (mean level or spatial pattern), the methods agreed for some pairs of dates and did not for others, but these differences were partly due to differences in data input. The method to use depends on the data availability and the aim of the study. The joint use of temporal stability and temporal mean shift and spatial shift tests could result in a drastically reduced sampling effort.