Coeficients of Kostiakov, Modified Kostiakov and Philip Infiltration Models on the Basis of Soil Bulk Density and Initial Water Content

Determining of the coefficients in infiltration models often is accomplished by fitting experimental data (cumulative infiltration and time) to the models; the two important soil parameters, soil bulk density (Db), and Initial water content (W0) are overlooked. In the current study it was aimed to find out to what extent the coefficients in Kostiakov, modified Kostiakov and Philip infiltration models would depond on Db and W0 and by entering those two variables into the models how much the predicting accuracy may be improved. For this purpose soil columns using PVC tubes with 23 cm in diameter and 80 cm length were prepared with four Db ranging from 1.11 to 1.42 gcm-3 and three W0 of 0.106, 0.131 and 0.147 gg-1. Infiltration tests were run with each combination of Db and W0 with two replicates, and the cumulative infiltratoion (I) were recorded at various times (t). Constant presure head of 2.1 cm water was maintained on the soil surfuce during all tests until approaching steady infiltration rate when tests were terminated. At the first step of data analysis, I and tdata were fitted to Kostiakov, modified Kostiakov and Philip models and their coefficients were computed. Then, the coefficients were regressed against Db and W0 and the regression equations were constructed. An the second step by applying those equations, the coefficients in the models were substitueted in terms of Db and W0 and cumulative infiltration, (Ip) were predicted by using three variables of t, Db and W0. Comparing the corresponding set of I and Ip through computing linear correlation coefficient (R) and root mean square deviation (RMSD) was the final step of data analysis. Results indicated that the dependence of the coefficients of these models on Db was quite greater than on W0. Entering Db and W0 into the models inspite of making them relatively complicated, resulted in better prediction of cumulative infiltration at most combinations of Db and W0. In this respect Kostiakov model with Db and W0 entered to it, produced more accurate prediction comparing to Philip and modified Kostiakov models.