Evaluation of unsaturated hydraulic conductivity of soil based on soil moisture retention data

Abstract

Hydraulic conductivity is one of the most important hydraulic properties that governs water flow in the soil. Groundwater recharge and contaminant transport mainly depend on the hydraulic properties of soil. The present study has been undertaken for the determination of unsaturated hydraulic conductivity and for the performance evaluation of different closed form equations. The experiments were conducted at three sites namely BUET campus, BARI campus and CIP area. Unsaturated hydraulic conductivity (K) was measured by using internal drainage method which follows Darcy's law. At each side the unsaturated hydraulic conductivity was determined for two columns of unit cross sectional area and depth from 35 - 45 and 45 - 60 cm respectively. The unsaturated hydraulic conductivity was determined from the measured value of hydraulic gradient and rate of change of moisture content. The moisture content and soil moisture suction at different times were measured by using Time Domain Reflectrometry (TDR) technique and Tensiometer respectively. Textural classifications of soil were done by Hydrometer method. The unsaturated hydraulic conductivity of SiC soil of BARI campus varied from 0.2 mm per day corresponding to moisture content of 36 percent to 24.0 mm per day corresponding to moisture content of 49 percent. The value of this parameter of SiCL soil of BUET campus was found to verify from 0.07 mm per day corresponding to moisture content of 35 percent to 25.0 mm per day corresponding to moisture content of 48 percent. The HC varied from 40 to 0.8 mm per day corresponding moisture content 51 percent to 34 percent for SL soil of CIP area respectively. The results are comparable to those of Joshua and Rahman (1983). The short ranges of soil moisture suctions (0 to 840 cm of water) and soil moisture content were measured by Tensiometer and TDR in this experiment. Residual moisture contents corresponding to soil moisture suction of I 5000cm were needed for the evaluation of empirical equations. So in this case, evaluation could not be performed based on this experimental data. On the other hand, empirical equations were developed based on a long range of soil moisture suction and soil moisture content. So, evaluations were done based on soil moisture retention data which were collected from field experimental by Joshua and Rahman (1983). The data were collected from three types of soils namely silty clay (SiC), silty clay loam (SiCL), and silty loanl (SiL). The field values of unsaturated hydraulic conductivity estimated by Joshua and Rahman (1983) were compared with the values estimated by four popular closed form models namely, Brooks-Corey (1966), Campbel (1974) and Bresler (1978) and Van Genuchten (1980). Parameters of each model were estimated based on soil-moisture characteristic curves. Evaluation of unsaturated hydraulic conductivity models was done based on statistical analysis and graphical interpretation. Statistical analyses were done using four different measures namely, average percentage of measured value, Standard Error of Estimates (SEE), correlation coefficient and level of significance for goodness of fit. Model predicted values of unsaturated hydraulic conductivity varied . from 90 to 181 percent of the measured value. The VG model was found to give least deviation from the measured value (10%) among the four selected models. SEE of model value varied from 1.5 to 4.8. The least value (1.5) was given by VG. The correlation coefficient varied from 0.96 to 0.97. The highest value (0.97) was found for VG model and lowest value (0.96) for Campbel model. The level of significance for the goodness of fit varied from 40% to 70% in case of Chi-square test. VG model fitted to the measured values at about 70% level of significance whereas Campbel, Bresler fitted at about 50% and 42% level of significance respectively. The level of significance for Brooks-Corey model found to be only 40% which is not satisfactory. On the basis of the above analyses, it can be concluded that VG model can be considered as most suitable among the four selected models and BC model was not found to give satisfactory results.