Clouds play a critical role in regulating air pollution. Their direct radiative impact alters the boundary layer photochemistry and inhibits ozone formation in favor of long range transport of ozone precursors (Pour-Biazar et al., 2007). Clouds also vent boundary layer air into the free troposphere, impact heterogeneous chemistry, wet deposition, aerosol recycling, and re-distribution of pollution over air, land, and water. Due to their importance, improving cloud simulations in meteorological and air quality models has been the focus of many studies. While the current air quality models show a significant improvement in cloud simulation compared to past practices, errors in timing and location of model clouds remain a significant source of uncertainty in air quality simulations.

Under previous TCEQ funding, University of Alabama in Huntsville (UAH) developed a technique to assimilate satellite cloud observations in the Weather Research and Forecasting (WRF) model. The technique has proved to not only improve the model performance with respect to cloud prediction, but also improve other atmospheric state variables. However, the technique relies on a series of scripts and programs that work outside WRF system to perform the adjustments needed and interacts with WRF through WRF FDDA option by providing a modified nudging field. In order to make the cloud adjustment technique developed by UAH more accessible to the broader user community this project explores the integration of UAH technique with WRFDA (WRF Data Assimilation) system.

WRF Data Assimilation (WRFDA) system (Barker et al., 2004; Huang et al., 2009; Barker et al., 2012) is an evolving system. As the new observational data is being incorporated into the system, the system is being modified and the new additions incorporated into the system. While FDDA (the option currently used by UAH assimilation system) is still a part of WRFDA, most of the recent developments have focused on the use of three dimensional variation (3DVAR) and four dimensional variation (4DVAR) techniques. Both these techniques are based on minimizing a cost function that describes model error as compared to the observations. 3DVAR can be considered a special case of 4DVAR where there is no time dimension. Basically, 3DVAR assimilates the observations to create a balanced initial condition for a forecast, while 4DVAR creates an analysis field by blending observations with model forecast.

The system originally started with the conventional data assimilation that emphasized the synoptic scale balance. Originally the convective-scale analysis did not receive the proper attention. However, in recent years new observational data have been assimilated into the WRF modeling system (Sun and Wang, 2013; Jones et al., 2013; Wang et al. 2012), and some have been adopted in the standard release of WRF. For example, recent attempts at assimilating radar data (Wang et al., 2012) have resulted in some core modifications in WRFDA, while the radar assimilation system resides outside WRFDA core. The radar assimilation technique incorporates the radial velocities retrieved from Doppler radar observation into the WRF data assimilation system.

Since the implementation of the UAH technique has some similarities to the radar assimilation, we propose to explore a similar approach in integrating UAH technique in WRFDA. UAH technique uses satellite cloud observations to estimate a target vertical velocity. Then using this target vertical velocity, the technique uses a variational technique to construct a 3D wind field over the modeling domain. Our first attempt will focus on providing the estimated horizontal wind field from UAH technique as observed wind field for WRFDA. Since WRF 4DVAR is using the model as the forward operator, this first attempt can lead to elucidating the effectiveness of 4DVAR versus FDDA in improving cloud simulations. 4DVAR relies on using the model as a forward operator and developing linear regression equations to be used in the cost function, while FDDA constrains the model forecast by observed value and relies on the model to create a balanced atmosphere. The results from this study will allow the evaluation of the effectiveness of these two techniques with respect to cloud simulation.

UAH also explores a more direct approach for assimilating estimated target vertical velocity or estimated vertical wind components into the WRFDA. Currently, UAH technique uses a one dimensional variation technique that adjusts the divergence estimates to achieve the estimated target velocity. However, these divergence estimates must be blended in with horizontal wind components. We will explore the possibility of using WRFDA to directly take in the estimated divergence components.

UAH will be performing the following tasks:

Task 1: Establishing the base case performance.

UAH will be using the latest release of WRF model and WRFDA system. Using this version of the model UAH will perform a baseline simulation for August 2006. WRFDA is an evolving system and as it evolves more physics options and model configurations are incorporated into the system. Previous versions of WRFDA have posed limitations on the model configuration. Use of the latest version will ensure the availability of a configuration more closely inline with UAH and TCEQ model configurations.

Deliverable 1.1: Interim report about the baseline simulation, configuration limitations, and justification for the configuration used.

Deliverable Date: 3 months after the start of work

Task 2: Performing Cloud assimilation using UAH current technique.

UAH will perform cloud assimilation using the current UAH technique with the configuration used in the baseline simulation. This will ensure that the estimated wind fields and error statistics conform to the same baseline configuration.

Deliverable 2.1: Interim report describing cloud assimilation simulation and evaluation of the results.

Deliverable Date: 6 months after the start of work.

Task 3: Preparation of WRFDA simulations.

WRFDA system will be installed and tested for the available data for August 2006. The system will be used to perform 3DVAR and 4DVAR on the model forecasts from baseline simulation. The estimated wind field in Task2 is based on the discrepancies between baseline simulation and satellite observation; therefore it conforms to the fundamental assumptions in WRFDA. These wind fields will be provided as observations to be used in 3D/4D VAR.

Deliverable 3.1: Interim report documenting results of 3D/4DVAR simulations. The analyses fields will be evaluated against surface observations for August 2006.

Deliverable Date: 9 months after the start of work.

Task 4: Performing a simulation using 3DVAR analyses fields.

UAH will perform a new simulation using 3DVAR analyses field as initial condition in each segment of model simulation. This task will quantify the impact of cloud assimilation in improving model forecast. The evaluation of the results from this task compared to 4DVAR results also indicates the model ability in retaining the improved cloud information. The results from this task will lead to developing an optimal time scale for assimilation.

Deliverable 4.1: Interim report on model simulations using 3DVAR analyses fields for initialization.

Deliverable Date: 12 months after the start of the work.

Task 5: Exploring the feasibility of direct assimilation of target vertical velocity in WRFDA.

UAH will investigate the feasibility of assimilating the estimates of target vertical velocity directly in WRFDA. This task will involve the examination of WRFDA code structure to estimate the difficulty of modifications needed to directly digest the target wind velocities that are estimated based on satellite observation. UAH also investigates the feasibility of other options (similar to that of radar assimilation) in which an additional cost function can be added to the standard cost function used in 3D/4DVAR calculations.

Deliverable 5.1: Interim report documenting the findings in Task5 and recommending an alternative.

Deliverable Date: 12 months after the start of work.

Task 6: Final report

UAH will document the results from all the tasks described above in a final report for submission to TCEQ.

Deliverable 6.1: Final report.

Deliverable Date: 12 months after the start of work.

References:

Barker, D. M., W. Huang, Y.-R. Guo, A. Bourgeois, and Q. N. Xiao (2004). A three-dimensional variational (3DVAR) data assimilation system for MM5: Implementation and Initial Results. Mon. Wea. Rev., 132, 897–914

Barker, D. M., W. Huang, Y.-R. Guo, and A. Bourgeois, 2003: A three-dimensional variational (3DVAR) data assimilation system for use with MM5. NCAR Tech. Note. NCAR/TN-4531 STR, 68 pp. [Available from UCAR Communications, P.O. Box 3000, Boulder, CO 80307.]

Barker, D. M., X. Y. Huang, Z. Liu, T. Auligné, X. Zhang, S. Rugg, R. Ajjaji, A. Bourgeois, J. Bray, Y. Chen, M. Demirtas, Y. R. Guo, T. Henderson, W. Huang, H. C. Lin, J. Michalakes, S. Rizvi, and X. Zhang (2012). The Weather Research and Forecasting Model's Community Variational/Ensemble Data Assimilation System: WRFDA. Bull. Amer. Meteor. Soc., 93, 831–843.

Huang, X. Y., Gao, F., Jacobs, N. and Wang, H. (2013). Assimilation of wind speed and direction observations: a new formulation and results from idealized experiments. Tellus A, 65, 19936, doi:10.3402/tellusa.v65i0.19936.

Huang, Xiang-Yu, Q. Xiao, D. M. Barker, X. Zhang, J. Michalakes, W. Huang, T. Henderson, J. Bray, Y. Chen, Z. Ma, J. Dudhia, Y. Guo, X. Zhang, D. J. Won, H. C. Lin, and Y. H. Kuo, (2009). Four-dimensional variational data assimilation for WRF: Formulation and preliminary results. Mon. Wea. Rev., 137, 299–314.

Jones, Thomas A., D. J. Stensrud, P. Minnis, R. Palikonda (2013). Evaluation of a Forward Operator to Assimilate Cloud Water Path into WRF-DART. Mon. Wea. Rev., 141, 2272–2289. doi: http://dx.doi.org/10.1175/MWR-D-12-00238.1

Pour Biazar, A., McNider, R.T., Roselle, S.J., Suggs, R., Jedlovec, G., Byun, D.W., Kim, S., Lin, C.J., Ho, T.C., Haines, S., Dornblaser, B. and Cameron, R. (2007). Correcting photolysis rates on the basis of satellite observed clouds. Journal of Geophysical Research, 112, D10302, doi: 10.1029/2006JD007422. issn: 0148-0227.

Sun, Juanzhen, and Hongli Wang, “WRF-ARW Variational Storm-Scale Data Assimilation: Current Capabilities and Future Developments,” (2013). Advances in Meteorology, vol. 2013, Article ID 815910, 13 pages. doi:10.1155/2013/815910

Wang, Hongli, J. Sun, X. Zhang, X.-Y. Huang, T. Auligné (2013). Radar Data Assimilation with WRF 4D-Var. Part I: System Development and Preliminary Testing. Mon. Wea. Rev., 141, 2224–2244. doi:http://dx.doi.org/10.1175/MWR-D-12-00168.1