Water quality and influence of interpolation procedure on visualization of selected parameters in a headwater stream, in Ayepe-Olode, southwestern Nigeria

Data interpolation – construction of new data points within range of a discrete set of known data point – is an important modeling activity in geographical studies. In this study, three commonly applied interpolation methods (nearest point, kriging and moving average) were examined in an assessment of the varying dispersion of selected physical and chemical parameters of stream-borne effluents from palm oil processing area in a growing commercial centre in Ife South local government area in Nigeria. Specific objectives were to examine selected physiochemical properties of a stream that receives palm oil effluent, and compare results of a kriging interpolation using derived variogram values with that which was based on the accepted parametric default in a popular geographical information system. The study also presents visualised results of interpolation of selected parameters based on ordinary kriging, moving average and nearest point interpolation. Analysis were achieved using PAST 3 and ILWIS GIS software. Result showed that although the stream is vulnerable to contamination by the palm oil processing activities around the area, it also receives contaminants from other non-source points that were not investigated in this study. It also indicated that the different point interpolation methods did not produce similar results. Whereas the values of conductivity were interpolated to vary as 120.1 – 219.5 μScm-1 with kriging interpolation, it varied as 105.6 – 220.0 μScm-1 and 135.0 – 173.9 μScm-1, with nearest point and moving average interpolations, respectively. Also, whereas the computed variogram model produced the best fit lines with Gaussian model, the Spherical model was assumed default for all the distributions in selected GIS software, such that the value of Nugget was assumed as 0.00, when it actually varies with data locations distribution. Conclusively, procedure of estimating spatial variation always produce results that are influenced by data distribution and model assumptions, and as such the data characteristics rather than GIS software’s defaults are appropriate for consideration in geospatial evaluation.