『Abstract
　A method for computing loads of total phosphorus and nitrate from a small catchment in the Czech Republic has been presented, given continuous measurements of water levels, turbidity and an ion-specific NO3^--N probe together with a limited number of discharge measurements and analyses of grab samples. Given the observed variability in the observations, log-log regressions are used to convert water levels to discharge and to derive continuous estimates of total phosphorus and nitrate. Integrations to estimate loads using predicted values of the true discharges and concentrations are then carried out by a Monte Carlo method to avoid strong assumptions of independence and distribution. Coefficients of variation for the estimated monthly loads range from 5％ to 23％ for TP and 12-36％ for NO3^--N. At the annual time scales, coefficients of variation are of the order of 5％ for both TP and NO3^--N. The variability in the estimated annual loads between the years was, however, significantly larger than the uncertainties estimated within each year.

Keywords: Uncertainty estimation; Rating curve; Monte Carlo sampling; Nutrient loads』

１. Introduction
　1.1. Uncertainty in load estimation
2. The catchment
3. Evaluating the load
　3.1. Predicting the instantaneous load
4. Results
　4.1. Modelling discharge
　4.2. Modelling phosphorus
　4.3. Modelling nitrate
　4.4. Load predictions
5. Conclusions
Acknowledgement
References