Effective incorporation of single-level observations, especially those of air temperature near the earth's surface, to accurately determine modeled initial atmospheric conditions represents a major challenge in numerical weather prediction. The exact reasons for this difficulty remain unclear, though inadequate representation of the diurnal cycle is thought to play an important role. This is particularly true in regions of complex terrain, where sharp variations of elevation and corresponding surface temperature (which are imperfectly resolved by coarse model grids) may lead to large differences between model "first guess" fields and inserted local observations. This challenge stands in the way of capitalizing fully on the bounty of new observations coming from expanded surface observing networks. The research supported here focuses on the use of observing system simulation experiments (OSSEs) in conjunction with the Weather Research and Forecasting (WRF) model to address this problem. OSSEs will be used to supply synthetic observations at sites where actual observations are being made. After incorporation of known, representative errors these synthesized observations will in turn be assimilated using both traditional variational (3DVAR) and more modern (but computationally expensive) Ensemble Kalman filter (EnKF) techniques. Ensuing model forecasts will be compared with actual observations at these selected sites to quantify the impact of such errors as well as assess methods for their reduction. The goals of this effort are thus to (1) identify and understand fundamental problems interfering with the inclusion of surface observations in weather forecast models, and (2) design and conduct numerical experiments to overcome these obstacles and thereby improve forecast accuracy.
The intellectual merit of this work centers upon identification of leading sources of weather forecast errors and design of methods for their mitigation. Broader impacts of this work will include: significant improvements to the community-based WRF model; more complete and efficient utilization of data emerging from a growing array of surface observational networks; and the education of a graduate student under supervision of a PI from an underrepresented group.
Effective incorporation of single-level observations, especially those of near-surface air temperature and wind fields, to accurately determine initial atmospheric conditions represents a major challenge in numerical weather prediction (NWP). The primary goals of this project, as addressed in the proposal, are 1) to understand the fundamental scientific problems associated with surface data assimilation, and 2) to conduct basic scientific research and numerical experiments to explore ways to overcome the major obstacles in assimilating surface observations. Specifically, we propose to apply observing system simulation experiments (OSSEs) to understand the impact of two unique issues in the assimilation of surface data: 1) variable terrain and 2) diurnal surface variations. To accomplish the proposed tasks, a series of OSSEs was performed using an advanced research version of the Weather Research and Forecasting (WRF) model and its data assimilation systems—a 3-dimensional variational data assimilation (3DVAR) system and Data Assimilation Research Testbed (DART) ensemble Kalman filter (EnKF) system. It was found that the 3DVAR method has problems generating realistic analysis increments from assimilating near-surface temperature and wind fields in regions of complex terrain. Next, we conducted the first study to comprehensively examine the ability of the EnKF method to assimilate surface observations. Results indicated that the EnKF method is an effective way to assimilate these observations. Its flow-dependent background error covariance term helps the data assimilation produce more realistic analysis increments in the atmospheric boundary layer and over complex terrain. In addition, over complex terrain, model terrain mismatch can cause problems in surface data assimilation, but EnKF is better able to handle mismatched terrain. Further studies also found that the proper assimilation of surface observations can significantly reduce errors in diurnal variations in analysis and forecasting. However, under strong synoptic forcing, assimilation of surface observations is still not able to overcome large diurnal errors in the near-surface layers of the atmosphere, implying that the model needs to be improved in order to fully address the problems with surface data assimilation. In addition to the proposed studies with OSSEs, research also extended to real cases and assimilation of surface mesonet observations. The impact of assimilation of surface mesonet data on numerical simulations and forecasts of mesoscale convective systems was examined. Results demonstrated that the assimilation of surface mesonet observations led to significant improvement in the predictability of mesoscale convective initiation and its subsequent evolution. A case study also showed that assimilation of surface observations with EnKF resulted in improved forecasts of a hurricane track during its landfall. The outcomes from this proposal should benefit the meteorological community and other related disciplines by increasing theoretical and practical knowledge of the effective use of available surface observations. Publications: Pu, Z. and J. Hacker, 2009: Ensemble-based Kalman filters in strongly nonlinear dynamics, Advances in Atmos. Sci. (SCI), 26, 373-380, dio: 10.1007/s00376-009-0373-9 Zhang, H.* and Z. Pu, 2010: Beating the uncertainties: Ensemble forecasting and ensemble based data assimilation, Advances in Meteorology, 2010, Article ID 432160, 10pp., doi:10.1155/2010/432160. Ma, M.*, Z. Pu, S. Wang and Q. Zhang, 2011: Characteristics and numerical simulations of extremely large atmospheric boundary layer heights over an arid region in Northwest China. Boundary Layer Meteorology, 140, 163-176. Zhang, L.**, and Z. Pu, 2011: Four-dimensional assimilation of multi-time wind profiles over a single station and numerical simulation of a mesoscale convective system observed during IHOP_2002. Mon. Wea. Rev., 139, 3369-3388. Pu, Z. and H. Zhang*, J. A. Anderson, 2012: Ensemble Kalman filter assimilation of near-surface observations over complex terrain: Comparison with 3DVAR for short-range forecasts. Tellus, (forthcoming).
