Study On The Ensemble Transform Kalman Filter-based Adaptive Observation And Applications

Adaptive observation is a new idea raised in recent years internationally to improve the quality of medium and short term weather forecast, the key content in international THORPEX, and is a frontier hot problem in numerical prediction field. No matter in the theory of adaptive observation or in observational experiment that aims to resolve high impact weather, the research in China is in its junior stage. For taking part in international THORPEX scientific plan program and at the same time, promoting our skill in research of adaptive observation and improving the outcome of the forecast of high impact weather, this paper will explore research on scientific problems related with adaptive observation.Taking the latest international research results of adaptive observation and according to ETKF theory, this paper develop ETKF adaptive observation system and GRAPES global ensemble prediction system that based on the ensemble transform Kalman filter (ETKF) initial perturbation scheme, especially considering the specialty of China's typical high impact weather system. Focusing on China's tropical typhoon and high impact weather that lead to heavy rainfall in Jianghuai area, this paper using data from GRAPES global ensemble forecast model and international TIGGE ensemble forecast to study the structural properties of ETKF signal variance, discover the relation between signal variance and the number of ensemble, the forecast time scale and meteorological variables, and still discover the dependence of that relations on different model ensemble forecast. Besides, this paper will intend to investigate the influence of adaptive observation to China's typical high impact weather system. The innovations and the main scientific conclusions of this paper are as follows:(1) Develop and build ETKF adaptive observation system. Combining with GRAPES 3DVAR, ETKF adaptive observation system suitable for China's high impact weather is built based on ETKF theory. Choose the total energy that can reflect exactly the prediction uncertainty as a measurement variable to distinguish targeting observation area. The ETKF adaptive observation system is of strong ability to predict signal variance and can identify targeting observation area.(2) The structural properties of targeting area and its relationships with influential factors. The number of ensemble has apparent influence on forecast veracity of tropical cyclone targeting observation area in low and middle latitude. Basically, ensemble with more than 37 members can identify exactly targeting observation area. The structure and reliability of targeting observation area can vary with forecast length, and have strong dependence on different ensemble forecast, and so, those show that the quality of ensemble prediction are important to ETKF adaptive observation.(3) Research on adaptive observation of China typhoon forecast. Research show that targeted observations of tropical typhoon in West Pacific is mainly in steering flow area that located by the north side of typhoon center. Whereas, targeted observations cannot ascertain that analysis error can definitely be improved. It changed with different levels and variables, and can reduce the analysis quality. Targeted observations can help to improve analysis error and forecast error of the geopotential height and those of temperature, but have different effect to wind forecast on different levels. It can make the forecast error of typhoon track and center strength smaller in 24 hr but the situation is not so after that. Another point is that targeted observations have better improvements on track forecast than on strength forecast.(4) Application of adaptive observation on heavy rainfall forecast in Jianghuai. Targeting observation area of heavy rainfall synoptic system in Jianghuai is basically located at ahead of depression system trough that will bring heavy rainfall. Targeted observations can improve both analysis and forecast, but will partially reduce those in some instances. For different analysis variables and layers, the effect is totally different. For forecast of some part of variables, improvement is mainly in low middle layer of troposphere, especially to geopotential height. Targeted observations play a certain part in predicting rainfall area and the heavy rain center, but do not take effect in improving the magnitude of the rainfall.(5) Developing GRAPES global ensemble prediction system which based on ETKF initial perturbation scheme. GRAPES ensemble initial perturbations can present main structure of analysis error variance in the North Hemisphere. The perturbation magnitude and ensemble spread is reasonable. However, the effect in south hemisphere is not that well. Perturbation increase is accord with increase of forecast error, and ensemble variance can correctly explain more forecast error variance. Ensemble forecast can reasonably express the developments status of real atmosphere. GRAPES global ensemble prediction system which based on ETKF initial perturbation scheme would better potential and values.This paper firstly built ETKF adaptive observation system that can be used both in scientific research and operational forecast, and provide a platform for China to enthusiastically join the collaboration on adaptive observation in international. This paper also gives first understanding on some scientific questions about targeting observation of China high impact weather, builds a foundation for future consistent study, and proposes an available method to improve the forecast quality of high impact weather. GRAPES global ensemble prediction system provides us a new scheme for global operational ensemble forecast as well.