LABEX Project Data Quality Review

Overall ISS Data Quality

There were a few shortcomings in the surface data collection. There were two significant gaps in surface data collection at the Clewiston ISS site (see below). The net-radiation data collected at all the sites were impacted by an error in the surface data ingestion software (voltage range specification) which cut off data above a value of approximately 335 W/m2. This affected daytime net-radiation data collected at all sites.

The profiler operation and data recovery was quite successful. The only problem was short lived. That was an error in beam orientation specification at ISS3. It was corrected the first day of the project (1844 GMT 7 July 1995). Data taken prior to that time has been corrected (see details below). C

Most of the soundings made had good data recovery. Only a relatively small number of those soundings were missing any significant data. There were occasional periods with bad wind data as well as a problem with the low level winds (below 1 km.) at the Pahokee site. The Pahokee low level winds were affected by an error in the site configuration (see section Loran-C Navigation at Pahokee (ISS4)).

Surface Meteorological Data

The problem was an error in software specification of beam orientation at the Okeechobee (ISS3) site. That error was discovered early and corrected. The error was such that only the wind direction was affected. Given the known beam orientation and the known error, the wind data were easily corrected. Both ASCII and netcdf consensus files were corrected and both the corrected and uncorrected files were supplied to the investigators. (The corrected ASCII filenames start with "z" rather than "w" and the corrected netcdf filenames have ".qc." included in them.)

Consensus wind data from each site's profiler were compared to wind data obtained from the balloon sounding. The profiler wind data consensus nearest the balloon launch time was used in that comparison. Note that this comparison compares instantaneous wind data (actually smoothed over one minute) with wind data averaged over time (25 minutes) and space. Plots of these comparisons for each sounding at all sites are included in Appendix C.

Consensus Parameters, Radar Parameters, and Beam Information.

Profiler data were processed to provide 25 minute consensus wind data files. These files are provided in both ASCII and netcdf format. The profiler wind data consensus files were generated from data obtained between five and thirty minutes past the hour and between thirty-five minutes past the hour and the next hour. In clear air operation, the consensus winds required that 55% of the data be within a 2 m/s window. Vertical motion subtraction was enabled.

Radio acoustic sounding system (RASS) temperature profiles were generated as five minute consensus files. Those files were generated from data obtained between the hour and five minutes past the hour and between thirty minutes and thirty-five minutes past the hour. In clear air operation, the consensus data required that 65% of the data be within a 6m/s window. Again, clear air vertical motion subtraction was enabled. The temperature profiles are provided in both ASCII and netcdf format. The RASS vertical temperature profile was obtained using the vertical beam specified in Appendix B using radar parameter set #4.

The profiler/RASS radar parameter sets, including beam information, are given in Appendix B for each site. The information provided gives the radar parameters for each beam, the number of beams used, individual beam orientation, and the beam sequence used. Unless a change is noted in the appendix, that information is valid for the entire project.

Each profiler operating system can store four different sets of radar parameters. Any one beam can be set up to utilize one of these parameter sets (see tables 1B, 2B, and 3B in Appendix B). In the presentation in appendix B, if a parameter set is not used that set is left blank. Note that radar parameter set #4 is always the parameter set used for RASS operation. The vertical beam used for the RASS consensus is indicated by an "R" in tables 1B, 2B, and 3B in Appendix B.

Appendix B summarizes the radar operating modes for each of the three ISS sites during LABEX. Part A of each table describes four sets of radar system parameters - information on the timing signals and the sampling intervals resulting from these signals. Part B shows the sequence of beam directions and the parameter set chosen for each beam. The radar operating mode is specified by the information in this table as data is collected for each beam in the order and with the parameters chosen. The vector wind is found by combining information from several beam directions.

The first 6 rows of part A show derived intervals and values based on other system parameters. SCALE is the Nyquist interval, or maximum positive and negative radial velocity observable (along the beam) with the sampling rates chosen; DWELL is the time spent collecting samples before moving to the next beam; RANGE shows the lowest and highest along-beam range (above the antenna, not above sea level) for which observations will be made; IPP is the inter-pulse period or time between transmitted pulses expressed in kilometers, which constrains the highest altitude to which data may be sampled; and PULSE is the transmitted pulse length in meters, which sets the range resolution.

The next three rows of this table are values chosen by the operator, rather than derived values. IPP again is the inter-pulse period, but now in micro-seconds. Changing the IPP will affect all the derived parameters in the first 6 rows with the exception of PULSE. PW is again the pulse width, now expressed in nano-seconds. The wind profiler is designed with four possible PW values 400, 700, 1700, and 3300 corresponding to spatial pulses of 60, 105, 255 and 462 meters. CODE BITS allows for increased sensitivity but its use is limited by the maximum transmitter duty cycle. Possible values are 0, 1, 2, and 4.

The final 6 rows of part A describe more parameters chosen by the operator. NHTS is the number of ranges sampled; DELAY in nano-seconds affects the waiting time before the first range sample - it specifies the lowest measured altitude; SPACE in nano-seconds is the distance between ranges sampled. SPACE is limited to the same four values as PW, and typically it is chosen to be the same as PW, but over-sampling or under-sampling is possible. COH AVG is the number of coherent averages. Coherent averaging can increase the profiler sensitivity, and also will affect the SCALE and DWELL. SPEC AVG is the number of spectral averages. This affects signal detectability, and will affect the DWELL. Finally, NPTS is the number of spectral points used in the Fourier transform. It affects the velocity resolution of the Doppler spectra and will affect DWELL.

Part B of the table shows the sequence of beam directions and parameter sets used. The profiler has 5 beam directions - four oblique beams at 90 degree intervals in azimuth and with a given elevation angle (67 or 69 degrees for the systems in LABEX), and a vertical beam (with two possible polarizations). Each row in part B shows the beam name, the number of repetitions (consecutive dwells) on that beam (one in all cases for LABEX), the parameter set used (as defined in part A), and the elevation angle of that beam. The profiler sequences through each row and then returns to the top in a continuous cycle.

Radiosonde Data

There are times when there are clearly bad winds associated with acceptable "Q" values (see section Individual Sounding Data Quality). This does occur from time to time and is a function of the fact that the "Q" value is nothing more than the standard deviation of the wind data over the smoothing interval used for the wind data calculation. (See "SSSF Observing Facilities: Description and Specifications".) It is possible for the navigator to generate bad data over a sufficient period to produce bad winds with an acceptable standard deviation. The investigator should be aware of this situation and use the "Q" values with caution.

The thermodynamic sensors performed reliably throughout the project. There was virtually no lost data due to a pressure sensor failure. The temperature and humidity sensor performance was good as well.

Data recovery or availability is represented graphically in Appendix D. This presentation is derived from the "Ten-Second Data" file. Sounding data losses can be associated with problems in various system components. In the case of no data, or intermittent data, it is likely due to a failure of some type in the sonde, commonly a transmitter problem or an antenna problem. In the case of missing or bad wind data the problem is most likely associated with the Loran-C reception by the sonde or relay of that signal to the ground station.

Loran-C Navigation at Pahokee (ISS4)

The ANI "knows" the position of the Loran transmitters being used and using its initial position (in error in this case) it figures distances to the stations and what their signal "time of arrival" (TOA's) should be. In a situation such as this, the error in position confuses the ANI and complete lock is hard to achieve. (The real distance to the nearest station, Jupiter, FL, is in reality about 50 km, but with this error the ANI "thinks" that distance is over 300 km.)

After sufficient time, the Loran will ultimately figure out where it really is and achieve NAV lock resulting in good winds MOST OF THE TIME. However, as seen by the presentation below this didn't always occur soon enough and that resulted in compromised data, particularly in the lowest kilometer. This is unfortunate as this project was concerned with lake breezes and thus the low level winds are important. (It is fortunate that the profiler performed reliably and that lowest kilometer is covered by the profiler winds.)

The summary below shows what happened on all the Pahokee flights. In a number of cases, there were bad winds in the lowest kilometer due to the situation described above. In most of those cases a position update was ultimately achieved and the remaining winds were reasonable. In some cases no position update was obtained and a wind solution was obtained with 3 or more stations in RCV (receive) lock. However, in these cases bad winds are encountered as the stations in RCV lock drop in and out as the ANI is still struggling with the position. In a few cases the position update and NAV lock was achieved at launch and most all winds were good (7/13, 7/14).

NOTE: The word "suspect" below is meant to indicate that there are bad wind data at times during the segment of the flight indicated.

ANI 7000 Status: Position never updated; Less than 3 Loran stations in RCV lock most of flight.

Wind Data Quality: What little wind data that are available are suspect.

ANI 7000 Status: Position updated at 130250 at which point NAV lock was obtained on at least 3 stations. Prior to 130250 - 1-3 stations in RCV lock.

Wind Data Quality: Wind data are suspect prior to 130250 (below approx. 1 km).

ANI 7000 Status: Position updated 1602; NAV lock obtained 1603.

Wind Data Quality: Wind data below 800m are suspect.

ANI 7000 Status: No lock of any kind at launch. One station in RCV at 185840. Five stations in RCV lock at 1901. Position updated 191700.

Wind Data Quality: Wind data below 6.5 km are suspect. They are clearly bad (oscillation in u and v components) between 3 km and 6.5 km.

ANI 7000 Status: No lock at launch. One station in RCV lock at 1006. Position updated 100810.

Wind Data Quality: Wind data are suspect through the first km. There are definitely bad wind data between 500 and 600 m.

ANI 7000 Status: Position updated at launch. Good lock not obtained until 1.4km. Lock lost for periods later in flight.

Wind Data Quality: Wind data are suspect below 1.4 km. Wind data are bad at the following altitudes: 6->8 km, 9->10 km, 11->12 km, and near 14 km.

ANI 7000 Status: Position never updated. Three stations in RCV lock most of flight. Only 2 in RCV lock at times.

Wind Data Quality: Wind data probably not as good as it could be. No wind data between 3 and 5 km.

ANI 7000 Status: Position updated and NAV lock obtained almost immediately

Wind Data Quality: Wind data good with possible exception of the first two or three points.

ANI 7000 Status: Position updated at launch, but no lock. One station in RCV lock at 1008; Two at 100850; Three stations in RCV lock at 1012.

Wind Data Quality: Wind data are suspect prior to 1012. Also winds are suspect between 8 and 12 km.

ANI 7000 Status: Position never updated. Not enough stations in receive lock until 1603.

Wind Data Quality: Wind data are suspect below 800m. See u and v component behavior.

ANI 7000 Status: No Loran lock through entire flight. High atmospheric noise (thunderstorms).

Wind Data Quality: NO WIND DATA.

ANI 7000 Status: Position updated and NAV lock obtained immediately. Lock lost for a period above 4 km.

Wind Data Quality: Wind data are definitely bad between 4 and 11 km.

ANI 7000 Status: Five stations in RCV lock at launch. Position updated and NAV lock achieved at 1600.

Wind Data Quality: Winds probably OK. Slight shift at about 800m - improvement perhaps.

ANI 7000 Status: Three stations in RCV lock at launch. Position updated within a minute of the launch time.

Wind Data Quality: Winds definitely suspect between 300 and 400 m. Bad winds between 1.5 and 1.8 km.

ANI 7000 Status: Position updated and NAV lock obtained at launch.

Wind Data Quality: Wind data are likely very good overall. Some bad wind data at 200 m. Wind data otherwise agrees with other sites.

ANI 7000 Status: Position updated and NAV lock obtained at launch.

Wind Data Quality: Wind data are likely very good.

ANI 7000 Status: Position updated at launch. Between 2 and 4 stations in RCV lock after that.

Wind Data Quality: First few wind data points are suspect. Bad winds near 4.7km and 7.5km.

ANI 7000 Status: Site configuration updated with better position. NAV lock at launch.

Wind Data Quality: Wind data are good.

SUMMARY:

Given the ANI performance described above, had the correct position been provided to the ANI in these cases, NAV lock would have most likely been easily achieved and maintained prior to launch. This would then have resulted in much improved low level (1 km) winds. In the situation described herein at Pahokee, the correct position was found using the ANI. However, when the site configuration was re-done to correct the site ID for use with the ISS, a "typo" occurred - 29 degrees latitude was entered instead of 26 degrees latitude.

The site configuration error also had a direct impact on the position information obtained from the sonde. Values for "Range" and "Azimuth" in the sounding data file products will be in error due to this problem.

Stability Parameters

Note that the stability parameters included with the Skew-T plots of Appendix E are generated from temperature and moisture data over the lowest 50mb (near surface) of the sounding. If the low level temperature and moisture data are not accurate, the stability parameters will be directly affected. Anytime there is a questionable stability parameter, that will be documented in the individual sounding data quality section which follows.

Individual Sounding Data Quality

The following information provided for each sounding is in reference to the "Interpolated Data" file produced. The Skew-T plots provided (Appendix E) are also plotted from that "Interpolated Data" file. The filenames for each listed sounding follow the standard naming convention: imddhhmm.xxx. Here, i stands for "Interpolated Data" file, m is a one digit indicator of the month, dd is the day, hh is the hour, mm is minutes past the hour, and xxx is a site id.

The "COMMENTS ON DATA QUALITY:" are included to address and explain problems encountered in the sounding data. Although effort is extended to cover as many data problems as possible, it is certain that not every bit of spurious data will be addressed herein. It is hoped that the comments provided will give the investigator using this data a good understanding of the possible problems that are sometimes encountered and an overall feel for the system and sensor performance. General comments about the data, not necessarily regarding problems, are also included with these comments.

In addition, the operator comments, when pertinent, are included in the following section.

Site: ISS2 -- Clewiston, Florida (SW "shore" of Lake Okeechobee)