During their analysis of the original DYCOM-II data set, Don Lenschow and Marie Lothon discovered what appeared to be atypical shapes in 3-D wind component spectra and a systematic bias in calculations of momentum fluxes.  They informed the RAF of their concerns about the data in July 2005 via a note summarizing their findings (DYCOMSII.issues.pdf).  In response to this issue the RAF formed an investigative committee that included both Don and Marie, Dick Friesen, Al Cooper, Jørgen Jensen and Allen Schanot to look into possible causes and corrections.

In a separate effort to improve wind data from the C-130 platform, the RAF took the opportunity in February 2004 to characterize the performance of the C-130 research pressure measurement system against a trailing cone pressure reference system.  Historically, and continuing on through today, the trailing cone is considered to be the standard pressure reference for aircraft avionics and research measurements.  Basically all aircraft-based pressure measurements are affected by the motion of the platform through the atmosphere.  Corrections (know as Pcors) to the raw measurements are determined empirically and are both platform and mounting location specific.  Standard theory has been that the Pcors are linear functions of the raw dynamic pressure (QCX) and are applied as:

QCXC = QCX - Pcorq(QCX)
PSXC = PSX + Pcorp(QCX)
        where PSX is the raw static pressure and QCXC and PSXC are the corrected values.

This form of Pcor was derived for the various C-130 systems during its initial testing, and the resulting Pcors were used on all research programs conducted through calendar 2003.  The results from the January 2004 trailing cone inter-comparison indicated that a more accurate parameterization of the C-130 Pcors could be achieved, including the effects of aircraft angle of attack (AKRD).  Such action was taken, and all subsequent field deployments (post July 2004) have used the new Pcors in the calculation of QCXC and PSXC.  At the time it was felt that the corrections to these parameters were too small to warrant any reprocessing of previous data sets.

First off, the DYCOMS-II data were reviewed for possible causes for the problems in the wind calculations.  It became clear that a small altitude-dependent bias in true airspeed (TASX) had been missed in the original data processing.  The systematic offset corrections to aircraft true heading (THDG) and corrected dynamic pressure derived from the standard calibration maneuvers for each C-130 deployment were determined for an altitude of 15,000 ft - as is standard practice.  The atypical presence of the altitude-dependent bias in the static and dynamic pressures resulted in TASX errors at the lower flight altitudes targeted by the DYCOMS-II research flight operations.  A maximum TASX offset error of 0.8 m/s was noted at an altitude of 300 ft AGL.  Such errors resulted in corresponding heading-dependent variations in the wind speed (WSC) of ± 0.8 m/s and wind direction (WDC) of ±7.0°.  While an adjustment of TASX resulted in better mean wind field data, it had little effect on the basic spectra shape and momentum flux issues raised by Lenschow and Lothon.

As the next step in the review process, previous C-130 data sets were checked to determine if the broader problems were of long standing or project specific.  It was quickly apparent that the altitude-dependent TASX bias was isolated to DYCOMS-II and the following C-130 field deployment (EPIC2001).  While some of the issues with the shape of the wind-component spectra seemed tied to possible water accumulation in the pressure lines (deposited during the prolonged in-cloud flight legs conducted during DYCOMS-II, the turbulent boundary-layer differences in the slopes of the longitudinal and latitudinal wind components have been consistent throughout the lifetime of the C-130 radome wind gust system.

The C-130 research pressure system was run through a series of special pressurization tests, and a problem was found with an o-ring seal, thus confirming the cause for the altitude bias.  The airspeed bias correction was then adjusted using some low-altitude maneuvers available in the DYCOMS-II data set to minimize the errors in the lowest 5,000 ft.  Then the data were reprocessed using the new angle-of-attack-dependent Pcors to see if that had any effect on the wind-component spectra and momentum fluxes.  Lenschow and Lothon re-ran their analyses on the newly re-processed data and found some significant improvements.  The results of that analysis are summarized by Lothon in an internal report (DYCOMS.report1_win07.pdf, January 2007).  Finally, both the 1 sample-per-second and 25 sps data sets from DYCOMS-II were reprocessed using both the corrected TASX and the new Pcors derived from the trailing cone reference test.

While many of the initial concerns raised by Lenschow and Lothon appear to have been addressed, the slope differences of the wind-component spectra remain an issue, and work continues in this area.  In the mean time, it was determined that the modified data sets provided much improved 3-D wind field and turbulence data and should be released to the community.  It should be noted that by reducing the airspeed bias error at the lower altitudes, the errors in the 3-D winds at the higher altitudes (above 10,000 ft) have been correspondingly increased.  It is likely that a slight heading dependence (± 0.8 m/s over 180° of heading) in the horizontal wind components (UXC, VYC, UIC, VIC) may be observed at these altitudes in the new data set.