Shu-Chih Yang;
Jhe-Hui Lin
National Central University
The 8th EnKF Data Assimilation Workshop
Ensemble Transform Kalman Incremental Smoother and its impact on severe weather prediction
Talk:
Yang_8thEnKF_Canada.pptx
Data assimilation seeks to find the optimal analysis to represent the atmospheric state by combining the information of observation and short range forecast (background). However, the corrections derived from data assimilation can also induce negative impact on forecast, such as the imbalance or unrealistic high-frequency oscillation. Such detrimental effect is particularly severe when the difference between observation and background is large or when the model state changes rapidly, e.g. undergoing highly nonlinear dynamics. These lead to issues of model spin-down or spin-up and affect the forecast accuracy. The incremental analysis update (IAU) has been proposed and widely applied to deal with these issues. Recently, the four dimensional IAU (4DIAU) is further proposed to consider the time-varying increment during the gradual update. Based on the framework of local ensemble transform Kalman filter, no-cost smoother and the concept of 4DIAU, we propose the local ensemble transform Kalman incremental smoother (LETKIS) to derive the incrementally time-varying and flow-dependent update. Based on the results from models with simple dynamics, LETKIS not only allows the gradual increment change in time and also the increment can correspond to the gradual update in the background trajectory within the assimilation window. Therefore, LETKIS provides a better analysis accuracy for cases with high nonlinearity, even with model errors.
In this work, we investigate the feasibility of applying LETKIS with the WRF-LETKF framework to alleviate the spin-down issue during severe weather prediction and preliminary results with a heavy precipitation case will be presented.
Data assimilation seeks to find the optimal analysis to represent the atmospheric state by combining the information of observation and short range forecast (background). However, the corrections derived from data assimilation can also induce negative impact on forecast, such as the imbalance or unrealistic high-frequency oscillation. Such detrimental effect is particularly severe when the difference between observation and background is large or when the model state changes rapidly, e.g. undergoing highly nonlinear dynamics. These lead to issues of model spin-down or spin-up and affect the forecast accuracy. The incremental analysis update (IAU) has been proposed and widely applied to deal with these issues. Recently, the four dimensional IAU (4DIAU) is further proposed to consider the time-varying increment during the gradual update. Based on the framework of local ensemble transform Kalman filter, no-cost smoother and the concept of 4DIAU, we propose the local ensemble transform Kalman incremental smoother (LETKIS) to derive the incrementally time-varying and flow-dependent update. Based on the results from models with simple dynamics, LETKIS not only allows the gradual increment change in time and also the increment can correspond to the gradual update in the background trajectory within the assimilation window. Therefore, LETKIS provides a better analysis accuracy for cases with high nonlinearity, even with model errors.
In this work, we investigate the feasibility of applying LETKIS with the WRF-LETKF framework to alleviate the spin-down issue during severe weather prediction and preliminary results with a heavy precipitation case will be presented.
Thanks to McGill's Atmospheric and Oceanic Sciences department for hosting this web-site.