Simulation model of tritium kinetics in a freshwater marsh

A simulation model of tritium kinetics was developed from data collected in a Lake Erie marsh from November, 1973 to October, 1974. Approximately 11 curies of tritiated water (HTO) were released into the marsh, and the loss of HTO was monitored for one year. Models were fitted to these data with nonlinear least squares to provide maximum likelihood estimates of parameters. The initial model of the marsh water was the simple differential equation dW/dt = -aW, where W is the marsh water HTO concentration, a is the loss rate constant, and t is time. The solution to this equation was a boundary condition for the partial differential equation used to describe HTO kinetics in the sediment, (delta S/sub c/)/(delta t) = D(delta2S/sub c/)(delta/sub x/2), where S/sub c/ is the sediment HTO concentration, x is the distance into the sediment, and D is the diffusion constant. A second model of the water-sediment system was developed to allow environmental variations to influence tritium kinetics. A 31-compartment system was assumed. HTO concentration in the marsh water was effected by HTO levels in the atmosphere, marsh water temperature, marsh depth, and rate of change of marsh depth. Organisms were described by a two-compartment model, a body water compartment, and a tissue compartment, with interchange between compartments. Estimated parameters for crayfish (Procambarus blandingi) showed that much of the tritium in tissue comes from food. A new parameter estimation technique was developed which allowed both sampling error and variation in the observed process. This method removed the autocorrelation found in the residuals when the process variation was significant relative to the sampling error. Simulated data were used to verify the technique