Improving severe storms forecasts by better constraining initial conditions

为了做出更好的天气预报，向计算机提供对当前天气状况的最佳描述是必不可少的。由于我们不能测量从地面到整个大气的所有大气参数，如温度和湿度，我们依靠观测和不同大气参数之间的物理关系的组合来获得当前天气状况的最佳图景。这种技术被称为数据同化。尽管数据同化越来越频繁，目前还不清楚我们是否尽我们所能做到最好。什么是最好的观测组合？我们应该使用其他工具吗？我们使用的观测周期是太短还是太长？在这项工作中，我们将使用基于计算机的方法来模拟数据同化过程，以调查哪些类型的观测最能全面地描述大气。我们将模拟现有和计划中的仪器如何在这一过程中帮助或不帮助。最后，我们将利用这次模拟的结果，通过实际观测和天气预报来证明，我们已经确定了执行数据同化的最佳仪器组合。

In order to make better weather forecasts, it is essential to provide the computer with the best description of the current weather situation.  Because we cannot measure all atmospheric parameters such as temperature and humidity everywhere for the whole atmosphere from the surface up, we rely on a combination of observations and physical relationships between the different atmospheric parameters to get the best picture of the current weather situation.  This technique is referred to as data assimilation.Although data assimilation is done more and more often, it is not clear that we are doing it the best that we can.  Are the best combinations of observations used?  Should we use other instruments instead?  Are we using observations over too short or too long a period?In this work, we will simulate the process of data assimilation using a computer-based approach to investigate what types of observations do the best job at describing the atmosphere as completely as possible.  We will simulate how existing and planned instruments help in this process or not.  Finally, we will use the results of this simulation to demonstrate using real observations and weather forecasting that we have determined the best combinations of instruments to perform data assimilation.