在一个简化气候模式的混沌态上,针对影响实际预报准确度的多种因素,将协调多时次差分格式与提为时间演变的变分同化进行了全面的比较.结果发现协调多时次差分格式可以有效地纠正模式误差,尤为独特之处在于其还可以有效地纠正微分方程本身不完善所造成的误差,而这一点正是变分同化所不及的。
Numerical experiments are performed on the chaotic movement of a simplified climate model. Considering various factors related to the predictability, we numerically compare the harmonized multi-time difference scheme (HMTDS) with the variational assimilation. It is shown that the HMTDS corrects remarkably the model bias, especially the differential equation bias, which is just the drawbaek of the variational assimilation.