1 / 8

Statistical Considerations for ABFM Radar Parameters

Statistical Considerations for ABFM Radar Parameters. Francis J. Merceret 2d ABFM/LAP Joint Workshop 16 May 2003. Overview. Core Issues Relevant Results to Date Things Planned for Near Term. Core Issues. Peak or average? What kind of average? dBZ, Z, or Z-derived LWC?

kamil
Télécharger la présentation

Statistical Considerations for ABFM Radar Parameters

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Statistical Considerations for ABFM Radar Parameters Francis J. Merceret 2d ABFM/LAP Joint Workshop 16 May 2003

  2. Overview • Core Issues • Relevant Results to Date • Things Planned for Near Term

  3. Core Issues • Peak or average? • What kind of average? • dBZ, Z, or Z-derived LWC? • Truncated (where)? • How big a box?

  4. Peak or Average? • Peak values are extremely sensitive to statistical sampling fluctuation. • See Reflectivity Averaging: A Monte Carlo Study under the reports link at ABFM Home Page for details • Use of average is strongly recommended.

  5. What Kind of Average? - 1 • Averages of Z are extremely sensitive to peaks and share disadvantages of peak values. • Averages of Zn such as LWC (n ~ 0.5) behave much like averages of Z • Averages of dBZ seem more consistent and repeatable as shown in Monte Carlo report.

  6. What Kind of Average? - 2 • Averages of data truncated at a lower bound are biased upward. Bias is severe if actual mean is within one std. dev. of boundary • Without truncation, how handle • Clear air (dBZ <<< 0) ? • Radar noise floor (range dependent) ? • Recommendation: Truncate at highest noise floor in the range of interest but no higher than two standard deviations below LLCC trigger threshold • See Monte Carlo report for quantitative analysis

  7. How Big a Box? • Box should be much bigger than radar beam position uncertainties. These are of order 1 Km. See Radar Beam Geometry Effects for ABFM under report link for details. • Box should be of same order as correlation length of phenomena being averaged. This is of order 10 – 20 Km. See Autocorrelation Presentation under report link for details.

  8. Things Planned for Near Term • Spectral analysis to determine scales of features most contributing to variance and covariance of radar and electric fields • Improved correlation analysis to determine degree of independence of the data • Correct QC error in preliminary work • Process entire data set • Include new averaged variables

More Related