Closure to “Master Diagnostic Curve for Dispersion Coefficient of Soils” by Sushil K. Singh

Abstract

The author thanks the discusser for mentioning certain points. The clarifications to these points are given below. 1. No matching of diagnostic curve or type-curve matching has been proposed earlier for the estimation of dispersion coefficient. Further development of the method of Ogata 1958 is not very relevant to the original technical note; however, it is again stated that it requires the inverse of the error function to be computed, while the diagnostic curve method proposed in the original technical note does not. Its modification proposed by the discusser requires reading of values corresponding to X0.69, X0.31, or X0.84, X0.16, which adds to the subjectivity. 2. The spreadsheet programs can be used only as black-box models. The method proposed in the original technical note does not require at all the computations for such an inverse function and is a white-box model. Also, the procedure suggested by the discusser can not identify and quantify the hysteresis in the dispersion coefficient, while the method proposed by the writer can. The points plotted in Fig. 1 of the discussion do not show a straight line, especially for the points of data set 3. Data sets 4 and 5 have been inadvertently left out; these data show considerable hysteresis and would not plot as straight lines. Drawing straight lines through the points of data sets 3–5 would result in much subjectivity in the estimation of the dispersion coefficient. If the points do not plot on straight lines, a linear regression would not yield meaningful results. It could not be understood as to why the points having c 0.2 and 0.8 have not been considered by the discusser. 3. The same does not hold good for utilizing only a few data points, which has been illustrated in the original technical note. Fig. 1 of the original technical note shows that only few data points would yield an equally good estimate of the dispersion coefficient. The discusser’s statement that the same holds good for estimates utilizing only a few data points is speculative. The reliability of an estimated dispersion coefficient should be based on some reliability criterion rather than only on its values. Table 3 in the original paper has been given to illustrate the error in the estimated dispersion coefficient if only a single point is used. Both Fig. 1 and Table 3 in the original paper are correct and there is no inconsistency. The misunderstanding of the discusser arises because he is