To reduce the uncertainty in calculation of unsteady flows, a multivariate alternate updating method is proposed based on the ensemble Kalman filter. This method updates water stage and discharge data alternately to calibrate unsteady flow, using the observed information without the large matrix calculating; meanwhile, scaling transformation is used in order to improve the water level filter precision. Numerical experiments emphatically investigate the effects of measurement accuracy and water level transformation coefficient on forecast precision of the method. The results show that the forecast error increases as the measurement accuracy decreases; the water level transformation coefficient can obviously improve the effect of the multivariate alternate updating method, the larger the water level transformation coefficient is, the higher the forecast precision will be; the multivariate alternate updating method has good calibrating performance and can improve forecast accuracy of unsteady flows in open channel.