Comparative analysis and prediction of water quality indicator parameters by using geostatistical river network models: A case study Coruh and Kucuk Menderes river basins

Within the scope of this study, by using river network models based on hydrologic distance and upstream-downstream relations that have brought a new breath to classical geostatistical methodologies in recent years, the comparative analysis, prediction, and basin-based spatial characterization of the water quality indicator parameters measured in two distinct basins of Turkey has been carried out. Water temperature, dissolved oxygen, alkalinity, and chloride water quality indicator parameters were modeled in Kucuk Menderes and Coruh River Basins by evaluating 7 parameters within the framework of torgegram, correlogram, principal component analysis, and exhaustive search algorithm. Additive function values produced by runoff curve numbers for spatial weighting, which is one of the model components, were used as comparatively and it was found that they affect the model accuracy positively. It has been approved that anthropogenic activities are more effective in Kucuk Menderes River Network than Coruh River Network in terms of autocorrelation based on hydrological distances, spatial covariates, the effect of model component and the mean standard error values of the perdictions. RMSPE error values of prediction results on the basis of sampling sites highlight alkalinity and chloride parameters in Kucuk Menderes River network and water temperature and dissolved oxygen parameters in Coruh River network. It has been found that alkalinity and water temperature estimates are more accurate in both river networks than the estimates of chloride and dissolved oxygen.