I have 6 samples of water quality parameters taken on a 10 km stretch downstream (Cooum River in Chennai), so it's basically the estuary.
The river discharge and flow is extremely low, therefore I assumed the current to be not an important factor.
Which interpolation method would make sense in this set up?
Since the number of sample points is extremely low I moved away from IDW.
At the moment I am considering Indicator Kriging, which was recommended for pollution data in general.
I would recommend the Indicator Kriging approach as well. Using indicator kriging to produce a probability or standard error of indicators map assumes an unknown constant mean. In addition, Indicator kriging can use either semivariograms or covariances, which are the mathematical forms you use to express autocorrelation.
The observed binary data is given by the open squares. The unknown mean for all indicator variables is shown by the dashed line, and it is µ. This can be compared to ordinary kriging. As with ordinary kriging, you assume that ε(s) is autocorrelated. Notice that because the indicator variables are 0 or 1, the interpolations will be between 0 and 1, and predictions from indicator kriging can be interpreted as probabilities of the variable being 1 or being in the class that is indicated by 1. If a threshold was used to create the indicator variable, the resulting interpolation map would show the probabilities of exceeding (or being below) the threshold.
There is more information Understanding indicator kriging
