HR: 14:25h
AN: A33F-04    [Abstracts]
TI: Data Assimilation and Targeted Observation of Chemical Tracer Concentrations in a Sea Breeze Model Forecast Using an Ensemble Kalman Filter
AU: * Stuart, A L
EM: astuart@hsc.usf.edu
AF: University of South Florida, 13201 Bruce B. Downs Blvd, MDC 56, Tampa, FL 33612-3805 United States
AU: Aksoy, A
EM: aaksoy@tamu.edu
AF: Texas A&M University, Department of Atmospheric Sciences, College Station, TX 77843-3150 United States
AU: Zhang, F
EM: fzhang@tamu.edu
AF: Texas A&M University, Department of Atmospheric Sciences, College Station, TX 77843-3150 United States
AU: Nielsen-Gammon, J W
EM: n-g@tamu.edu
AF: Texas A&M University, Department of Atmospheric Sciences, College Station, TX 77843-3150 United States
AB: Ensemble-based Kalman filtering (EnKF) is a data assimilation technique that is undergoing significant investigation for many environmental modeling applications. Here, we study the use of EnKF for constraining meteorological uncertainties in a two-dimensional sea breeze and chemical tracer model forecast. For this work, we have augmented a nonlinear meteorological model of a sea breeze circulation with a tracer transport algorithm. With the coupled model, we perform three numerical experiments. First, we investigate the chemical tracer forecast uncertainties associated with meteorological initial condition error. We use a 50 member ensemble forecast in which member initial conditions were chosen by statistically sampling a time series of data, in a `climatological' initialization scheme, to represent the variance in the maximum diurnal heating (local noon). We find that the ensemble variance builds during the transition between land and sea breeze phases of the circulation. Second, we investigate the effects on the forecast variance and error of assimilating tracer concentration observations from a network of surface locations. Observations were extracted from an arbitrarily-chosen 51st member used as the truth simulation. We find that the EnKF assimilation reduces the variance and error in both meteorological variables (vorticity and buoyancy) and in chemical tracer concentrations. Finally, we investigate the potential value to the forecast of added targeted observations by post-processing analysis that determines an uncertainty norm throughout the spatial and temporal domain. The norm uses covariance information generated through the ensemble forecast to maximize the total decrease in model uncertainty summed over all state variables. By comparing the distributions of the uncertainty norm with and without data assimilation of the network observations, we find that the overall uncertainty decreased, but locations of optimal targeted observations remained similar. Our analysis also demonstrates the potential usefulness of EnKF data assimilation for planning targeted observations.
DE: 0345 Pollution: urban and regional (0305, 0478, 4251)
DE: 0368 Troposphere: constituent transport and chemistry
DE: 0520 Data analysis: algorithms and implementation
DE: 0545 Modeling (4255)
SC: Atmospheric Sciences [A]
MN: Fall Meeting 2005