Extended Abstract (64K)15A.2
Sensitivity study of HRD's H*WIND surface wind analyses for tropical cyclones
Samuel H. Houston, NOAA/NWS/WFO, Honolulu, HI; and M. D. Powell
The Hurricane Research Division's (HRD's) H*WIND system has been used to generate surface wind analyses for tropical cyclones (TCs) in real-time in the last few years. These wind products are used routinely as guidance for operational TC forecast and advisory products, including the determination of wind radii (e.g., radius of 34, 50, and 64 kt winds by quadrant) by hurricane forecasters at the National Hurricane Center and the Central Pacific Hurricane Center. There has already been extensive development of the methodologies and techniques used in the analysis of all available surface wind observations to derive the TC wind fields that are generated by the H*WIND application. Considerable work with post-storm analyses of significant tropical cyclones, such as Hurricanes Andrew and Iniki of 1992, Hurricane Floyd of 1999 and Super Typhoon Paka of 1997, has led to significant improvements in the software used in the wind analysis algorithms contained in H*WIND.
The routine use of H*WIND to determine the characteristics of relatively weak and poorly defined TC surface wind fields during recent hurricane seasons has exposed some unexpected problems in the analysis of the surface winds. Asymmetric systems such as Central Pacific Hurricane Daniel (2000) and Atlantic hybrid TCs such as Tropical Storms Gordon and Helene of 2000 had particularly complicated surface wind structures that severely tested the algorithms contained within the H*WIND application.
This study uses "known" analytic wind fields sampled at discrete locations to simulate typical data sampling patterns near the surface in TCs. The output from H*WIND using these sampled "known" winds as input will then be compared with the original analytic fields to quantify the errors. Parameters involved in the objective analysis of the surface wind data, such as the analysis domain, the mesh sizes for the multi-nested grid used by H*WIND, filtering wavelengths, and the methods of filling in data void regions with subjective data will be varied to determine which values reduce the errors between the analyzed and the "known" wind field. This study will not only examine well developed and intense TCs, but will also look at the weaker and less well developed, but often more complicated, storm structures. Using the results of these sensitivity studies the optimal parameters as a function of TC strength and intensity for the objective analyses will be determined for the use of H*WIND during future hurricane seasons. An additional benefit from this work may be the proposal for new sampling strategies for aircraft reconnaissance flight patterns. For example, the benefits of rotating or modifying the "Figure 4" flight pattern may be potential outcomes within the framework of this study.
Session 15A, Tropical Cyclone Observations and Structure IV (Parallel with Sessions 15B and 15D)
Thursday, 2 May 2002, 4:00 PM-5:30 PM
Previous paper Next paper