Model bias correction for dust storm forecast using ensemble Kalman filter

摘要

First attempt to correct model bias in a dust transport model using ensemble Kalman filter (EnKF) assimilation targeting heavy dust episodes during the period of 15–24 March 2002 over north China is successfully performed. The uncertainty of dust emissions and surface wind fields are taken into account individually and simultaneously to correct their biases. The 24-h surface forecasts are significantly improved with the root mean square error reduced by more than 45% on 20 March and by 50% on 21 March after correcting the biases. The results indicate that there are high biases due to the dust emissions and surface wind fields. These biases converge to the values similar with those obtained in previous sensitivity analyses indicating that the EnKF can accurately correct the bias. The corrected total dust emissions are decreased more than 33%. However, when considered simultaneously, they do not converge to the same results as those considered individually. This indicates that the two biases can compensate for each other in terms of predicted surface dust concentration.