Atmospheric data assimilation which accommodates weather descriptions as observations of solar radiation

Evaluation of performance with a modern one-year case

天気記述を日射量としてとりこむ大気データ同化: 現代の1年の事例による性能評価

JpGU-AGU 2020 [M-IS28-P05]

Questions and answers

• [Q] Were tropical storms (including typhoons) reproduced by the assimilation?
• [A] Tropical storms are not well reproduced in our assimilation experiment, though we did not mention in the poster because the cases seem to be too few. One of the reason may be the fact that the spatial scale of tropical storms is smaller than that of extratropical cyclones. But, more likely reason is that the small ovarlap between the track of a tropical storm and the area of main four islands of Japan where we assume availability of observations.
• [A (cont.)] Even if there is some overlap, just weather information does not seem effective enough to reporduce the pattern of tropical storms. Collecting many written documents of winds, Hirano was able to reconstruct the track of a tropical cyclone (See [An article in the JCDP site] (in Japanese)). If this kind of information can be input to data assimilation, it can probably reproduce a tropical cyclone. Still there is difficulty that pre-modern description of winds are not standardized.
• [Q] Are such experiment as yours can be used to estimate how dense the observations (weather descriptions) need to be in order to reproduce each of various kinds of weather phenomena?
• [A] We cannot yet answer such questions, but we make effort to become able to answer them.