home » about nssl » research divisions » forecast r&d » models and assimilation
Data Assimilation
It is necessary to assimilate real data onto model grids to make a forecast based upon the initial state, or "picture" of the atmosphere. Traditional data sources, such as surface and upper air observations, are more widely spaced than individual grid points in high resolution models. Other observations, such as radar and satellite, do not include direct information on all the standard variables contained within modal equations. Scientists at NSSL are examining data assimilation techniques and evaluating whether incorporating new data and non-standard data improves the forecast. More about the problem of assimilating real data onto model grids »
Phased array radar observed radial velocity at 21:47 UTC on 2 June 2004. The aliased velocity areas are highlighted by yellow circles.
PHASED ARRAY RADAR DATA ASSIMILATION
Assimilating phased array radar data into models should improve numerical analyses and predictions of severe storms. In an initial effort to achieve this goal, scientists are taking a comprehensive approach to attack problems in three important aspects: (i) velocity dealiasing in radar data quality control, (ii) error covariance estimation, and (iii) radar data assimilation using the estimated error statistics. More about phased array data assimilation techniques »
RADAR DATA QUALITY CONTROL
By using the Bayesian statistical decision theory, a probabilistic QC technique was developed to identify and flag migrating-bird contaminated sweeps of level II velocity scans at the lowest elevation angle using three identified QC parameters. The QC technique can use either each single QC parameter or all the three in combination. While the single-parameter QC technique is useful for evaluating the effectiveness of each QC parameter, the multi-parameter QC technique is much better than any of the three single-parameter QC techniques (as verified by using independent ground truth information). When the multi-parameter QC technique is used for real applications (with no ground truth information), the probabilities of wrong decision can be also estimated quite reliably (as indicated by the tested percentages of wrong decision). The detailed techniques were presented in two recent publications (Zhang et al. 2005, Liu et al. 2005, J. Atmos. & Oceanic Technology).
DATA ASSIMILATION FOR MILITARY AVIATION
NSSL is working in conjunction with the Naval Research Laboratory (Monterey), the Desert Research Institute (Reno, NV), and the NOAA Western Region Climate Center to improve the short-range forecasting of severe weather at the Fallon (NV) Naval Air Station. Especially critical to the operation of this jet aircraft training facility is the presence or absence of strong winds, obscuration from dust, and cloud. The research combines data assimilation with regional mesoscale prediction to make 0 - 6 hour forecasts of the critical parameters. John Lewis, research meteorologist at NSSL, is responsible for the data assimilation component of the research effort. The Office of Naval Research is supporting this 3-year project.
LIGHTNING DATA ASSIMILATION
NSSL scientists are studying how including lightning observations in mesoscale forecast models could improve the forecasts of initial conditions. Lightning data could be important because it can pinpoint the location of current convection and may be a measure of its intensity. Using lightning data could also improve the effects of prior convection on the initial conditions of the forecast period. More about lightning data assimilation techniques »
