The Ensemble Transform Sensitivity Method And The Study Of The Adaptive Observation Application Over The Tibet Plateau Based On The ETS

The Qinghai-Xizang(Tibet) Plateau(TP), as the “Third Pole”, can impact the weather and climate of East Asia or even the global world enormously. In recognizing the important roles of the TP, The China Meteorological Administration(CMA), the Chinese Academy of Sciences(CAS), and serveral other agencies, have finished the first and second Tibetan Plateau Atmospheric Scientific Experiment(TIPEX) in 1979 and 1998. Many important results were revealed. But the experiments are limited by observational parameters. The third TIPEX(TIPEX3) has started recently. One of the major missions of TIPEX3 is to establish a comprehensive observation system over TP and adjacent region. The lack of observations could cause large initial analysis uncertainties in the Numerical Weather Prediction(NWP) model. These initial uncertainties can be amplified by the atmospheric instability, lead to forecasting errors which are known as the internal error growth. The initial analysis fields of TP region have large uncertainties since the meteorological observational and operational stations are scarce over these areas. These errors can propagate to the downstream region too, such as: Sichuan basin, Yungui plateau, East of China, and limit the forecast skills of these areas. In this study, we showed that the initial errors of TP can propagate to the south-west China. The RMSEs of the downstream areas of TP are reduced after assimilate the real and “synthetic” observation data. Compared with “US” region forecasts, both the RMSEs results between WRF forecasts and GFS analysis/real observations imply that: the propagation of errors over TP region is larger and faster; the initial errors of TP region can propagate to the downstream areas after 24 hours’ integration.The sampling technique can be used for the strategy of establishing the observing networks during TIPEX3 too. For high-impact weather(HIW) events, adaptive mobile observation instruments or vehicles can be deployed to improve analysis quality and forecast accuracy. There are several approaches that have been developed to estimate sensitive areas, such as the singular vector method, the conditional nonlinear optimal perturbation method, and the adjoint sensitivity method. In general, an adjoint model is usually required in the above three approaches. In addition, ensemble-based methods, such as the ensemble transformation(ET) method, the ensemble transform Kalman filter(ETKF) method, and ensemble sensitivity are widely used in field campaigns. The ensemble-based methods are less demanding computationally and have been extensively employed in practical applications. These methods consider sensitivity in the subspace spanned by the ensemble forecasts and are computationally inexpensive in operational centers where ensemble forecasts are routinely produced. Among these methods, ET provides a practical method for adaptive observations. It has been used for targeted dropsonde deployments in winter storm reconnaissance(WSR). Later, ETKF was used to identify the sensitive region in WSR. The dropsonde data collected over these sensitive areas improved the weather forecasts over the continental United States and Alaska. However, the impact of dropsonde data may be limited in global forecasts, and high-resolution observation datasets are suggested for HIW. It is noted that ET is still expensive for high-resolution applications or those applications with large numbers of ensemble members. ET has been used at relatively coarse resolutions and a few vertical levels, e.g., usually three vertical levels at the National Centers for Environmental Prediction(NCEP), and a relatively small number of ensemble members(30--60). As resolutions increase for HIW applications, the computational cost grows exponentially. This is because ET, as well as the ETKF, has been implemented to exhaust all possible observation deployments. In order to further improve these methods for fine scale HIW applications, efficiency is an important factor to investigate. A new ET-based sensitivity(ETS) method is proposed in this paper to specify sensitive regions for adaptive observations. The newly proposed method possesses two advantages over the original ET method, better reflecting the analysis error variance in the sensitivity and computationally efficient. In this paper, ten extreme numerical experiments have demonstrated this subtle difference between ET and ETS. ETS method indeed showed larger sensitivity over the areas where analysis error variance is large but ET was not that sensitive. The ETS method is applied to the application of the sensitive region identify of TP areas. Fourteen south-west vortex cases are study in this paper. The sensitive results show the sensitive region of the south-west vortex are located near the initial source of the south-west vortex, such as: JiuLong ans XiaoJin areas. One moth numerical experiments are also provided. The sensitive areas from these experiments implicit the sensitive areas are located at the south of the TP areas. The OSSE results also indicated the additional radio-sound observations can decrease the forecast errors. These results could provide the possible strategy of the radio-sound observation networks. Here are the major conclusions:(1) The errors of TP region can propagate to the downstream areas in 1-days’ integration and the errors propagation is faster and larger over the TP region. The third Tibetan Plateau Atmospheric Scientific Experiment(TIPEX) is focus on establishing a comprehensive observation system over TP and adjacent region. Two Observing System Simulation Experiment(OSSE)s are setup to evaluation the observation data. The results imply that: the observation data over TP region can affect the forecasts of the downstream areas; additional “synthetic” observation data can improve the forecast skills. After assimilate the observation data of the upstream of verification areas, the OSSE forecasts get closer to the “nature”, especially for the horizontal wind and geopotential height. The OSSE have less RMSE because of the improvements of initial fields.(2) ETS is much faster than ET because ET needs to loop over all the possible elements in the state vector, especially when the number of ensemble prediction members(K) is large. The cost is less than 60 seconds with a fine resolution and few ensemble prediction members for ETS and ET. This is acceptable for the adaptive observations. However, the computational cost rises to about 1200 seconds with a 1° × 1° resolution in the horizontal direction, with three vertical levels and 50 ensemble prediction members. ETS only costs about 200 seconds. Overall, the computation time saved by ET was 60%--80%. If the computational domain is larger(particularly for a global model) with higher resolution in the horizontal and vertical directions(here the computations were conducted in three vertical levels only), the reduction in computational costs would be much more significant with ETS compared to ET. Overall, for the Hurricane Irene(2011) case, both ET and ETS identify one sensitive region; for the rainfall case, one region with global maximum signals and two local regions with local maximum signals are identified. The differences between ET and ETS are acceptable, since the targeting observation focuses on the sensitive areas with a maximum(the center of the signals). Generally the signals from ET and ETS are similar.(3) By examining the ET method proposed by Bishop and Toth(1999), an improvement is considered and a new ETS formulation is proposed in this paper. Contrast to ET formulation, the ETS formulation clearly shows the targeting observation sensitivity is proportional to the analysis error variance while the original ET formulation results in a reciprocally proportional relation. In real atmospheric applications, the variance might vary significantly due to observation distribution and accuracy, and weather conditions. For example, the variance over the ocean might be much larger than those over continental regions due to relative dense observations on land. The actual meteorological sensitivity might be larger over regions associated with higher values of analysis error variance since higher variance usually means more “accurate” observations are needed in these areas. A good adaptive observation scheme needs to take advantage of the actual analysis error variance. The numerical experiments results also implicit that the signals are indeed proportional to the analysis error covariance in the ETS; The ET are not sensitive to the enlarged of the analysis error variance. the signals from ETS can represent the higher error variance more appropriately.(4) A set of sensitive areas estimation experiments are provided in this paper. The cases are the fourteen south-west vortex. The sensitive regions identified by the ETS show the south-west vortex system are sensitive to the source areas of the vortex such as: JiuLong and XiaoJin areas. Another set of numerical experiments are provided which last for one month. The results show the major sensitive areas of TP and its surrounding areas are located the south of the TP. The OSSE results also confirm the results from adaptive observations:The additional radio-sonde observations could improve the forecast skill of the TP. And the results from the adaptive observations could provide deploying strategy for the additional radio-sonde networks.