COMPARISON OF THREE LABORATORY AND ONE REGRESSION KRIGING METHOD FOR QUANTITATIVE AND QUALITATIVE ASSESSMENT OF SOIL SALINITY IN THE HARRAN PLAIN, SE TURKEY


BİLGİLİ V., AYDEMİR S., SÖNMEZ O. , Cullu M. A.

FRESENIUS ENVIRONMENTAL BULLETIN, cilt.22, ss.1339-1350, 2013 (SCI İndekslerine Giren Dergi) identifier identifier

  • Cilt numarası: 22 Konu: 5
  • Basım Tarihi: 2013
  • Dergi Adı: FRESENIUS ENVIRONMENTAL BULLETIN
  • Sayfa Sayıları: ss.1339-1350

Özet

The Harran Plain occurring in southeastern Turkey, has faced salinity problems since the beginning of irrigated agriculture. Soil salinity is generally most accurately determined from a soil saturation paste (SP) extract. In this study three laboratory (SP, 1:1 and 1:2.5 soil to water ratios) and one kriging technique have been used for the assessment of soil salinity. A total of randomly selected 210 locations, 60 in data set I and 150 in data set II were sampled in 2009 and 2010, respectively, and analyzed for soil electrical conductivity (ECe dS m(-1)), sodium adsorption ratio (SAR) and the soluble cations (Ca2+, Mg (2+), Na+ and K+). Regression analysis was used for quantitative assessments and a classification approach used for qualitative evaluation of salinity. The kriging of residuals and the values from regressions between 1) soil EC(e)s obtained from different soil and water ratios (auxilary variables) and 2) soil salinity variables obtained from saturation paste (primary variables) were combined under regression kriging with the goal of estimating soil salinity parameters. Despite significant correlations among different methods, the results of paired t test showed that averages of soil salinity variables measured with different methods were mostly statistically different (P=0.01). The 1:1 soil water ratio produced the closest results to SP, especially after classification of soils into different salinity groups which provided regression R-2 values up to 0.99. Using a validation with independent samples per cent classification accuracy and kappa statistics of 91 %, 0.72 (p=0.001) were obtained. Kriging combined with regression under regression kriging improved the estimations of K+, Mg2+ and ECe slightly but did not show any improvement over different soil to water ratios for the estimation of SAR, Na+ and Ca2+.