The effect of rate-limited adsorption on transport of environmental contaminants is difficult to characterize at the field scale. This study investigated transport, during unsaturated water flow, of pulse inputs of bromide, simazine (2-chloro-4,6-bis(ethylamino)-s-triazine), and MS-2 coliphage in a field lysimeter (0.8 m ¿ 0.8 m square) containing undisturbed Tujunga loamy sand (mixed, thermic, Typic Xeropsamment). Sixty-four fiberglass wick soil solution samplers collected drainage fractions from the exit surface (30 cm depth) following daily 2-cm water inputs applied at 0.5 cm h-1. After 19.7 cm of cumulative drainage, the soil above 10 of the 64 locations was sampled to determine final depth distributions of simazine and virus. Most of the bromide was leached from the transport volume, while the sorbing pesticide and virus remained in the soil. Variance analysis indicated that local dispersion processes contributed more to the observed bromide spreading than did differences in local water velocities.

A linear, first-order, kinetic adsorption submodel was incorporated into a generalized linear transport model relating the bromide flux concentrations to the simazine and virus final resident concentrations. Least squares fitting showed that area-averaged bromide transport could be described reasonably well by the two-parameter convection-dispersion model (CDM), although the mobile-immobile water model provided a slightly better representation of effluent tailing. The CDM parameters fitted to the bromide data were then held constant while the two parameters of the adsorption submodel were varied to fit the pesticide soil concentrations at the end of the experiment at 10 days. A good fit was obtained for simazine, and the fitted value 0.54 d-1 of the rate coefficient was in the range characterizing nonequilibrium adsorption. A batch adsorption/desorption experiment produced Freundlich isotherms describing nonlinear adsorption (exponent m=0.85) and hysteresis in desorption. There was poor agreement between the retardation factor (R) estimated from a linearized batch distribution coefficient Kd and the R fitted to lysimeter data. Virus concentrations fitted to the model yielded coefficients implying strong adsorption (R=254) and rapid inactivation (inactivation rate coefficient of 1.64 d-1), whereas the laboratory sorption study implied that the virus should be very mobile in soil. The difference in field and laboratory sorption may be due to air-water interfacial forces in the unsaturated field experiments. ¿ American Geophysical Union 1995