|RAMMB CIRA 2nd Quarter Report|
January February March 2008
A new parameterization for estimating SST cooling beneath a tropical cyclone’s inner core has been developed. In this parameterization, SST cooling is dependent on storm intensity, storm translational speed, and ocean heat content. These values were computed from reruns of the HWRF model for Atlantic tropical cyclones from 2004-2006, and a multiple regression analysis was used to determine a simple linear relationship. Next, this parameterization will be tested in the SHIPS model against the current SST cooling relationship to determine whether or not it provides any improvement to the SHIPS intensity forecasts. (A. Schumacher)
The Tropical Cyclone Formation Probability (TCFP) product has been verified for the 2007 season. Analysis of Brier Skill Scores and Relative Operating Characteristic Skill Scores indicate that the TCFP outperformed climatology alone in all three basins covered by the extended domain, which includes the Atlantic, E. Pacific and W. Pacific basins. Subjective analysis of the sub-basin time series plots indicates that the TCFP product is showing peaks of enhanced TC formation probability within the 24-hour prior to TC formation the majority of the time. (A. Schumacher)
2007 time series plots of cumulative TC formation probability over 2 sub-basins in the western N. Pacific basin. The black line represents climatological formation probability, the blue line represents the TCFP product estimated formation probability, and the open red circles represent all times 24-hour prior to the formation of a TC.
Assistance has been provided to NESDIS Satellite Analysis Branch members in the moving of the operational TCFP product to the NSOF facility. The TCFP product is now running temporarily on a linux system (gp60) at NSOF and work continues to get the product and its graphical displays running properly on the AIX machine (satepsdev3) that will run the product from now on. (A. Schumacher, J. Knaff, D. Molenar)
A global water vapor imagery archive has been identified for use in planned improvements to the TCFP product. This archive is part of the NOAA HURSAT project, and will be provided by Ken Knapp. This archive consists of 3-hourly, near-global water vapor images created from various geostationary satellites. The domain extends from 60 N to 60 S and from 0 to 360 E, and the data has already been inter-calibrated to account for differences between the satellites. Ken Knapp has provided this data for 2004 and expects the full dataset, which will extend from 1998-2006, to be available early this summer. (A. Schumacher)
For several years, GFS data have been ingested and converted from grib1 format to a packed ASCII format for use in tropical research. In January, the real-time GFS data switched from grib1 to grib2 format. The software which performs the ingest was modified to process the new format, and the transition to grib2 went smoothly. (J. Dostalek)
In an effort to improve the tropical cyclone formation product, large-scale vertical motion will be added as an additional screening parameter, and its affect on the skill of the forecast measured. The large scale vertical motion field will come from an omega equation valid over the entire sphere. The input will come from the GFS model. The development of the software used to compute the vertical motion is underway. (J. Dostalek)
Rapid intensification is being reexamined using the IR imagery. Twenty-five extremely rapidly intensifying tropical cyclone cases with IR imagery have been collected 1987-2006. The average starting intensity of these cases is ~60 kt and after 24 hours the final intensity is ~110 kt. Techniques to remap and spatially filter the brightness temperatures to a storm motion relative, cylindrical coordinate system have been developed. The cylindrical data has also been interpolated to a half-hourly temporal resolution. Two dataset were created, one with and one without the axisymmetric mean brightness temperatures removed from each image. Finally, the 2-d time series is then temporally smoothed using a binomial filter and a complex principle component analysis is applied. (J. Knaff)
Results suggest that the first five complex empirical orthogonal functions are related to 1) the mean, 2) cyclonic outflow, 3) pulsing central convection, 4) anticyclonic outflow and 5) rotating core convection/cloud free moat formation. The spatial amplitudes and phases are shown in Figure 1. Wind barbs give the phase direction and relative magnitude (note the change in sign produces erroneous wind barbs). Similar results were found with the dataset with the azimuthal mean brightness temperatures removed, but nearly symmetric features like CEOF 3 were difficult to interpret.
Figure 1: Spatial Amplitudes (shaded) temporal phase (contoured), and phase speed/direction (wind barbs) of the first three complex principle components of the azimuthal brightness temperatures found in 25 rapidly transitioning tropical cyclones. A 48-hour period is used for each storm encompassing 24 hours prior and the 24-hour during these rapid transitions. The intensity at the beginning (end) of these rapid transitions was 60 (108) kt.
Composite means of the amplitude and phases of these patterns from the 25 rapidly intensifying cases suggest that CEOF2, CEOF3, and CEOF5 may be helpful for predicting rapid intensification. These results form the basis for a conceptual model of how the spatial patterns of IR brightness temperatures evolve during rapid intensification. These results will be presented next quarter at the 28th AMS Conference on Hurricanes and Tropical Meteorology. (J. Knaff)
Vertical wind shear (VWS) is an important factor in determining the potential for tropical cyclone intensity change that is often estimated using analyses and forecasts from numerical prediction (NWP) models. While this approach has been shown to be rather successful, it is not clear if the NWP models capture all the variability of VWS. AMSU-derived wind fields offer an independent dataset from which VWS can be calculated. Such calculations are possible using balanced winds calculated from AMSU-derived height fields, which are already an operational product. Such independent information may also be useful to tropical cyclone forecasters. To investigate the reliability and usefulness of AMSU-based VWS calculations, the magnitude of the 200 minus 850 hPa VWS has been calculated using both AMSU- and NCEP analysis- based wind fields in a storm-centered circular area within 600 km of the storm. These VWS calculations are then compared to vertical wind shear from the NCEP analyses (Figure 3). Results show that there is a good agreement between VWS values calculated using AMSU-derived wind and those calculated from the NCEP operational analyses. The relationship is shown below for all tropical cyclones in the North Atlantic 2004-2007. Further evaluation results will be presented by R. Zehr at the AMS Conference on Hurricanes and Tropical Meteorology. (J. Knaff, R. Zehr)
Figure 3: Scatter diagram comparing the magnitude of the 200 to 850 hPa vertical wind shear calculated using the NCEP operational analyses and the 2-D non-linear balance winds produced by the NOAA operational AMSU tropical cyclone intensity and structure algorithm. The comparison, which indicates that the AMSU-based vertical winds shear explains 70% of the NCEP VWS variance, includes data from all tropical cyclones that occurred in the North Atlantic during 2004-2007.
Previous studies have indicated that recurving western North Pacific tropical cyclones, initially westward moving tropical cyclones that turn toward the east, often reach their maximum intensity close to the time of recurvature. Those results have often been cited in the literature and sometimes inferred to be valid in other tropical cyclone basins. This study revisits this topic in the western North Pacific, North Atlantic and Southern Hemisphere tropical cyclone basins. The timing of lifetime maximum intensity associated with recurving tropical cyclones is examined using best track datasets from the United States’ Joint Typhoon Warning Center and the National Hurricane Center, Miami during the period 1980-2006. Results reveal that tropical cyclones are less likely to experience peak intensity within ± 12h and ±24h of recurvature than has been previously reported in the western North Pacific. It is also shown that tropical cyclones that become most intense (i.e., intensities greater than 52 ms-1) have a greater tendency to reach peak intensity before recurvature than weaker storms save for in the South Pacific where the most intense storms have a greater probability of reaching their maximum intensity following recurvature. While it appears that weak tropical cyclones (i.e., peak intensities less than 33 ms-1) often reach peak intensity prior to or close to recurvature in all tropical cyclone basins as others have reported, the cumulative distributions of maximum intensity with respect to the time of recurvature can be quite different for other intensity ranges suggesting that a universal relationship between peak intensity and time of recurvature does not exist. This study was submitted to the International Journal of Climatology. (J. Knaff)
Return to the current RAMMB CIRA Quarterly Report