Spatial variability and geostatistical analysis of selected soil

Abstract

A study was conducted to explore the spatial variability of major soil nutrients of Agricultural fields in South-western region of Bangladesh. From the study area, 40 surface soil samples were collected by a random sampling strategy using GPS. Then soil physico-chemical properties i.e., pH, electrical conductivity (EC), organic matter (OM), total nitrogen (TN) N, soil available nutrients (P, K and S) were measured in laboratory. After data normalization, classical and geo-statistical analyses were used to describe soil properties and spatial correlation of soil characteristics. Spatial variability of soil physico-chemical properties was quantified through semi-variogram analysis and the respective surface maps were prepared through ordinary Kriging. Spherical model fits well with experimental semi-variogram of pH, EC, OM, TN, available P, K and S. Soil pH, available phosphorus (Av P), potassium (Av K) and sulfur (Av S) have the moderate spatial dependence, with nugget/sill ratios of 41.13% to 72.21%. The others have the strong dependence with nugget/sill ratios of less than 25%. The spatial variability of estimating soil properties varies within range of 0.0142 for Av P to 0.0383 for Av S and this model can calculate the un-sampled within neighboring distance of about 12.65 m for Av S to 150.82 m for TN, respectively. Cross validation of kriged map shows that spatial prediction of soil nutrients using semi-variogram parameters is better than assuming mean of observed value for any un-sampled location. Therefore, it is a suitable alternative method for accurate estimation of chemical properties of soil in un-sampled positions as compared to direct measurement which has time and costs concerned.

Bangladesh J. Sci. Ind. Res.54(1), 55-66, 2019