Close Window

North Carolina State University: "An investigation of the inland intensification of tropical storm Danny"

Final Report


The ultimate goal of the project was to investigate several aspects of the un-anticipated intensification of Tropical Storm Danny as it crossed the state on 24 July 1997. To move toward this goal we were to develop tools to use WSR-88D radar data to recover and analyze the wind field within the storm and to estimate the evolution and distribution of latent heat release.

The participants in the project included Dr. Allen J. Riordan, NCSU Principle Investigator, Jason Caldwell, NCSU Research Assistant, Kermit Keeter, NWS SOO, Joel Cline and Ronald Humble, lead NWS forecasters. The work period of the project began 15 May, when an NWS SAC computer, an HP workstation, first became available, and extended to 15 August 2000. This period corresponded with a large free block of time when the university principal investigator and research assistant were available.

As detailed in section 4, we encountered several problems that prevented us from meeting our original goal within the time and funding limits, but we were able make substantial gains and develop tools that may prove useful for several types of future radar applications.

The first step in the project was to obtain the level II radar data from NCDC and install software developed at NCAR for the analysis of Doppler radar data. Level II radar data from Greer, SC and Wakefield VA were obtained and used to document tropical storm Danny's reflectivity and wind fields at two key times, the first several hours before the storm's sudden intensification and the second about midway through the process. NCAR software packages were to be used to process the radar data and obtain the wind fields. These packages included Solo, REORDER and CEDRIC.

The first package (Solo) allows perusal, editing and processing of reflectivity and velocity fields. We used it to identify volume scans appropriate for examination and to unfold the Doppler velocities. We also developed a Solo workbook describing how to use the software, since such did not exist. Nearly all of this effort was by Jason Caldwell who spent about two weeks familiarizing himself with Solo and developing a set of instructions for its use.

The next task was to obtain and install REORDER, a program to transform data from radar to Cartesian coordinates. Ron Humble and Jason Caldwell worked on this task and compiled and installed the software on the HP. We used REORDER to interpolate and grid the data for four volume scans: two volumes that included the storm core and placed at two locations roughly equidistant and nearest the Greer radar site and two similarly located volumes near Wakefield. REORDER was well documented, but we had to overcome a few problems and delays in the compilation. Without gracious help from Dick Oye at NCAR we would probably still be wondering how to compile the program.

Finally, CEDRIC was to be used for computing synthetic dual-Doppler winds from the pairs of velocity volumes at the two sites. However, we again encountered trouble that plagued us to the end of the summer at which time, Jason Caldwell and I had other project commitments. Ron Humble with much help from Jay Miller at NCAR did eventually succeed in installing and successfully running CEDRIC. Dr. Miller provided much guidance along the way and supplied us with a script for recovering the wind fields using his new synthetic dual-Doppler technique.

We now have all three software packages working and producing results. The software can be used to map the vector wind field within the eyewall region of TS Danny and we will likely dedicate some time to this task in the future. The software can also be used to analyze additional cases of interest for which level II Doppler radar data exists.


Our work was limited to the tool-development stage. When final results can be obtained, a training workshop in the use of the software can be held. The programs run in minutes on the HP workstation, so can potentially be used for near-real-time analysis of wind fields for systems of interest including landfalling hurricanes. However, initial testing and evaluation of results first need to be completed for some test cases. Further, training materials on the use of the software and techniques need to be developed.




Benefits of the project include the further development of a working relationship with NWS staff who helped throughout the project. We also were able to develop, document, and make available a potentially valuable analysis tool for future research endeavors.

Much of the problems encountered stemmed from the limited programming skills and systems knowledge of the university PI and research assistant. The first major stumbling block was the installation of CEDRIC. Without the help of Ron Humble at the NWS and Jay Miller at NCAR, this would have been impossible. Between Ron's operational shift schedule and e-mail communication with Jay, the installation took roughly two weeks.

The most time-consuming delay a problem of incompatibility of the program with the HP compiler. It took about two months for this to become obvious as CEDRIC appeared to be correctly installed, but would crash before running to completion. Upon each crash and we would e-mail Jay who in turn would suggest ways to diagnose or fix the problem. Finally in late October, at the end of the project, Jay completed development of an HP-friendly version of CEDRIC. This version now allows the program to run to completion.


The problems associated with software issues were very time consuming requiring considerable manpower hours. With better documentation of the software, the problems could have been significantly reduced.

Despite the frustrations, the collaborative spirit and cooperation between NWS and NC State University remains strong. The project's participants persevered through many unexpected difficulties and achieved considerable success in tool development. The results of the project do in fact provide potential for future radar applications. No doubt, those who follow and choose to use our results will find an easier path.

Detailed Review
Several software packages were needed by the project. Several installed on the Hewlett Packard workstation (HP) without problems.

One early problem related to REORDER software in that we needed the latest version of the program, but such was not available on the NCAR web page. It would be helpful in the future if documentation with each software package referred specifically to the version number of each dependent software package necessary for the target program to function.

The freeware NCAR graphics program would not compile on the HP. I believe this to be related to problems mentioned regarding CEDRIC (see below). NCSU has a University license for the graphics package and this was loaded from CD-ROM (version 4.1.1). However the install script referenced a "HP-UX" directory on the CD-ROM, when the proper directory name was "hpux". Installation failed and was completed manually.

No instructions were found for installation of CEDRIC in the CEDRIC "tarball" or on the web. Instructions for the program SPRINT were used and installation of the files to compile the final program was successful.

CEDRIC compiled without error on the HP but failed to run to completion without error. Jay Miller at NCAR was a patient and invaluable help in tracking down problems with the code. These problems were eventually related to the HP compiler and its interpretation of record size and treatment of local vs. global variables. Once these problems were solved, CEDRIC ran to completion without error.