HRDI Data Quality The following information applies to HRDI level 3a data, version number 7 thru 10. Mesospheric zonal and meridional winds: Altitude Range The daytime altitude range differs according to modes run and time period - there are three main periods: 1) before UARS day 201 (29-March-1992) HRDI provided mesospheric data from two different modes, one of which gave results from 80 to 105 km (strato-meso mode), while the other covered from 65 to 105 km. These two modes were generally alternated on a daily basis during this period with our stratospheric mode, so that (for example) we might, during a three day period, have 80-105km coverage, no mesospheric coverage, then 65-105 km coverage. 2) 65 to 105 km between UARS day 201 and UARS day 620 (22-May-1993), 3) 50 to 115 km after UARS day 620. Please note that some data are available above these altitude ranges, at lower quality. Nighttime winds are only available at the peak of the O2 Atmospheric (0,0) band, which is approximately 94 km but varies (and is always reported as 96 km in our data). In the HRDI level 3a products, if there is only a single altitude reported, the data are nighttime winds results. Estimated Precision For these data, precision information is available in the 'data quality' field of the file, which holds the standard deviations of the measurements. These values depend on viewing conditions - at 95 km with low solar zenith angle, they can be less than 3 m/s - under 'average' conditions (solar zenith angle about 30 degrees) the standard deviation of the data varies (depending on interpolation from level 2 to level 3 locations) from 6 to 12 m/s at 111 km, from 3 to 7 m/s at 95 km, from 8 to 15 m/s at 69 km, from 20 to 40 m/s at 50 km. Estimated Accuracy The accuracy of the data set depends on how well we have determined the 'zero wind' position for the current mode of instrument operation. We have determined the 'zero wind' position to within 5 m/s for data after UARS day 201. Before UARS day 201 (30-March-1992) there are some days during which we ran different modes which, because of the small size of the data set, cannot be validated/calibrated to the same degree as the rest of the data set - these are UARS days: 63-79 (13-November-1991 thru 29-November-1991), 113-120 (2-January-1992 thru 9-January-1992), 134-149 (23-January-1992 thru 7-February-1992), 165-173 (23-February-1992 thru 2-March-1992), 182-189 (11-March-1992 thru 18-March-1992). The accuracy of these data is at or near the 10 m/s level. Resolution HRDI collects a 'scan', consisting of a series of measurements at different altitudes above the same position on the earth, every 30 seconds, on average, during the 'day' (when the spacecraft is on the sunlit side of the orbit). These data are then linearly interpolated onto the UARS standard grids (every 4 degrees in latitude for L3AT data, or every spacecraft 'minute' (about 65 seconds) for L3AL data). HRDI makes measurements every 2.5km in altitude, which is a slightly finer grid than the UARS level 3 standard (3km from 60km to 120km, 5km elsewhere). Before UARS day 394 (9-Oct-1992) HRDI made nighttime wind measurements approximately only one day out of three; after day 394 nighttime winds were measured nearly every day. Stratospheric zonal and meridional winds: Altitude Range The daytime altitude range is from 15 to 40 km in the stratosphere. Estimated Precision For these data, precision information is available in the 'data quality' field of the file, which holds the standard deviations of the measurements. These values depend on viewing conditions. Estimated Accuracy Same as in the mesosphere. Resolution Same as in the mesosphere. Systematic Effects: Our level 2a product wind measurements along the instrument line of sight are at about our required precision, but because of noise added in the inversion process, we need to filter out high frequencies in our inverted profiles to create our level 2b and level 3a products. We accomplish this thru the use of sequential estimation, in effect taking a weighted average of the current measurement and those near it in time and location. As a result, some small scal features in the real winds may be smoothed out by this process. We continue assembling data to verify that our estimating algorithm is optimum. No sequential estimation is used in determining the nighttime winds. The HRDI instrument has had two hardware failures, and has also been somewhat effected by the UARS spacecraft solar array problem, and by HRDI's exposure to the space environment. After each of the instrument's hardware failures (which occurred on UARS days 201 (30-March-1992) and 510 (2-February-1993)) we were able to develop work-arounds, but as a result we were required to change the tuning of our instrument somewhat, which means that raw winds measured after each failure were shifted somewhat from winds measured beforehand. We have been able to calculate and compensate for these shifts by comparing with correlative measurements. These compensations are part of the version 7 (and beyond) data files, but we are still in the process of refining our compensation for the day 510 shift. When the UARS solar array drive first halted (UARS days 266 thru 313, 3-June-1992 thru 20-July-1992) our instrument was powered down and got very cold - this also caused a change in the way the instrument worked, and a shift in the measured winds. Compensation for this shift is also included in the version 7 (and beyond) data. We believe that the space environment has also affected the measured winds. We see a long term drift in the measured winds (and in the position of on-board calibration lamp lines), which we attribute to drying and outgassing of the instrument optics (including coatings). Compensation for this drift is also included in the version 7 (and beyond) data files. The raw winds measured by the HRDI instrument also depend on the temperature of the instrument. We have attacked this problem in two ways, first, we have tried to calculate the wind drift with temperature, and have applied this compensation in the version 7 (and beyond) data files; we have also developed new instrument control software, to narrow the temperature range over which the instrument is operated. Both of these efforts are still underway. Caveats: HRDI started making good wind measurements on UARS day day 53 (3-Nov-1991), before that the instrument control coefficients were not sufficient to make good wind measurements. UARS days 85-90 (5-December-1991 thru 10-December-1991), 202-226 (31-March-1992 thru 24-April-1992), 510-519 (2-February-1993 thru 11-February-1993), and 583-587 (16-April-1993 thru 20-April-1993) contain invalid winds. No HRDI measurements are available for UARS days 266-313 (3-June-1992 thru 20-July-1992), 694-698 (5-August-1993 thru 9-August-1993), 737-743 (17-September-1993 thru 23-September-1993), 753-775 (3-October-1993 thru 25-October-1993), 1290-1303 (24-March-1995 thru 6-April-1995), and 1314-? (17-April-1995 thru the present) due to problems with the UARS solar array. HRDI level 2b data are interpolated to create both L3AT and L3AL data files. They are also available on both an altitude and pressure grid. We recommend using the altitude gridded data for a number of reasons: 1) the HRDI instrument makes its measurements on the altitude grid, 2) the pressure gridded data are produced using NMC pressure/altitude data for the given day, a somewhat smoothed product, and 3) the UARS standard grid for pressure has a top level of 1.47e-3 mb, or about 90 km, while HRDI makes measurements well above this level.