Extended Abstract (432K)11A.3
Data assimilation strategies in the planetary boundary layer
Brian P. Reen, Penn State University, University Park, PA; and D. R. Stauffer
The assimilation of mass-field (temperature and water vapor) observations at the surface and within the planetary boundary layer (PBL) and their effects on PBL structure and depth is explored in 1D and 3D MM5 experiments. Surface and other within-PBL mass-field data are sometimes excluded from data assimilation due to difficulties in effectively assimilating these observations. Since there can be strong vertical gradients in mass fields near the surface (e.g., surface superadiabatic layers), the best method to apply surface observations to the lowest model level (LML), which is often well above the surface, requires further investigation. Additionally, representativeness and temporal variability at the surface and within the PBL are also often much different than above the PBL, and this also affects the assimilation methodology and the vertical spread of observation information. Assimilation methods are first tested in 1D MM5 experiments and the most promising techniques are then applied to 3D MM5 forecasts.
The 1D experiments are first designed to investigate computing surface mass-field innovations from surface observations and diagnostic 2-m above ground level (AGL) values from the model, compared to applying surface observations within the innovation for the LML directly. During free-convective conditions the PBL is well-mixed and so the surface innovation is applied throughout the PBL. This allows assimilation of surface mass-field observations to improve model simulations well above the surface. We also propose nudging the soil temperature using the innovation from the surface air temperature, because nudging the air temperature alone disrupts the surface energy budget. However, assimilating only surface mass-field observations can still have negative effects since surface innovations applied within the PBL can change the low-level stability and degrade the model PBL depth.
Improved 1D results are found by 1) computing the surface mass-field innovation at 2 m AGL rather than the LML and spreading it throughout the PBL during free-convective conditions, 2) limiting it's use to near the surface during stable conditions, 3) assimilating multi-level (radiosonde) temperature and water vapor 3-h observations within the PBL, and 4) assimilating the above-PBL data as commonly utilized. As expected, assimilation of more data in the profile improves the model results and removes the increased PBL depth errors found when assimilating surface mass-field observations alone. The 3D results further confirm the 1D testing and show improved mass-field predictions and model PBL depths that are so important for air-chemistry and atmospheric transport and dispersion applications.
Session 11A, Data Assimilation: Profiling and satellite data assimilation
Wednesday, 3 June 2009, 4:00 PM-5:30 PM, Grand Ballroom East
Previous paper Next paper
Recorded presentation