This file contains a summary of the instrument documentation to help you use the HALOE Level 2 and Level 3 data held at the BADC.
Here is a brief outline of the HALOE instrument and the measurement and data processing techniques which are employed. Users requiring a more detailed description of the design and operation of the instrument should consult the paper by Russell et al. (1993a).
HALOE is a satellite solar occultation experiment designed to monitor the vertical distributions of HCl, HF, CH4, and NO by gas filter correlation radiometry and H2O, NO2, O3, and temperature versus pressure using CO2 absorption by broadband filter radiometry (Russell et al., 1993a). Aerosol extinction is directly measured in each of the gas filter channel wavelengths which range from 2.5 to 5.26 m (Hervig, et al., 1993). The broad band radiometer measurements range from 2.7 to 10 m wavelength. The absorption of solar energy at selected spectral bands is used to infer vertical profiles of atmospheric temperature, pressure, and mixing ratios of key gases involved in ozone chemistry. The HALOE instrument includes both broadband and gas filter channels covering the spectral range from 2.45 to 10.04 microns.
HALOE uses the solar occultation technique to make measurements of vertical profiles of atmospheric parameters. The satellite observes 15 sunrises and 15 sunsets per day. For each of these events a vertical scan of the atmosphere is obtained by tracking the position of the Sun during the occultation. The technique is illustrated schematically in Figure 1 .
As each event proceeds, the instrument locks onto the sun and goes through a sequence of operational modes, which can be divided into two distinct phases :
An Exo-Atmospheric Reference Phase:In this phase the instrument observes the sun above the atmosphere. An intensity profile for the sun is measured by scanning across the solar disk, the gas-correlation channels are balanced (for sunsets only), and a sequence of calibration measurements are made in each spectral channel.
An Atmospheric Tracking Phase: The solar position is tracked through the atmosphere, and the atmospheric absorption is measured in each channel as a function of the pressure at the tangent height (defined as the altitude of the closest approach of the ray path to the Earth's surface).
The order in which these two steps are performed depends on whether the event is a sunrise or a sunset.
Gas mixing ratio profiles are then determined by taking the ratio of the solar intensity which has been attenuated by its passage through the atmosphere with the unattenuated intensity measured outside the atmosphere. The use of a relative measurement means that the instrument is virtually self-calibrating.
The instrument uses Gas Filter Correlation Radiometry to make measurements of HCl, HF, CH4 and NO, and Broadband Filter Radiometry for measurements of O3, H2O, NO2 and the temperature as a function of pressure using a CO2 channel. Aerosol extinction is also measured in each of the gas filter channels.
1.2.1 Gas Filter Correlation Radiometry (HCl, HF, NO, CH4)
Solar energy enters the gas correlation section of the instrument and is divided into two paths for each channel. The first path contains a cell filled with a sample of the gas to be measured; the second is a vacuum path. The output signals from the detector in each of these paths are fed into a differencing amplifier, where a differenced signal is calculated.
In the reference phase of an occultation event, when the instrument observes the sun above the atmosphere, the gains of the two channels are balanced, so that the difference signal is close to zero.
During the atmospheric tracking phase, absorption by the target species in the atmosphere changes the spectrum of the incoming radiation in a way which is strongly correlated with the absorption line spectrum of the gas cell. This introduces a difference signal, from which the concentration of the target gas in the atmosphere may be derived.
Since absorption by aerosol is spectrally broad, it may be taken as flat across the bandpass of a single HALOE channel. affecting both the gas-cell path and the vacuum path equally. As a result it does not contribute to the differenced signal. Once the mixing ratio of the target gas has been derived, the broadband signal in the vacuum path is used to derive the aerosol transmission.
1.2.2 Broadband Filter Radiometry (H2O, NO2, O3,CO2)
These channels are conventional broadband filter radiometers. The instrument signal is a direct measurement of the energy absorbed during occultation. The ratio of the atmospheric signal to the exo-atmospheric signal is a measure of the atmospheric transmission in the optical path. Aerosol absorption has a large effect on the radiometer measurements but signal correction methods have been developed (Hervig et al., 1993) and used successfully to derive mixing ratios.
1.3 Data Processing Techniques
The UARS data processing is carried out at the Central Data Handling Facility at the Goddard Space Flight Center using software supplied by the Instrument's Principal Investigator group.
The data processing for UARS instruments consists of a progression through a sequence of `levels' from the raw telemetry at level 0 to geophysical quantities interpolated onto standard grids at level 3. The processing steps for HALOE are outlined below : Level 0-1 At the level 0-1 processing step, instrument-specific effects are removed and a set of calibrated data are derived in physical units (eg. voltages and radiances) tagged with their locations.
Level 1-2 The level 1 data are then processed further to produce the level 2 product which contains vertical profiles of temperature, pressure, mixing ratios of chemical constituents and aerosol extinctions at the measurement positions. This step involves a complex `onion-peeling' algorithm in which species are retrieved layer-by-layer from the top of the profile downwards, using the data from the layers above to retrieve at lower levels.
For each event the pressure and temperature retrievals are performed first, followed by the mixing ratios and aerosol extinctions. The order in which the retrievals are carried out is important, since some species are spectral contaminants in other channels. Where this happens, retrievals are repeated using the derived profiles from the contaminant channel to improve the original retrieval.
The level 1-2 processing step also requires data on spectral line parameters and spectral filter functions for the gas-filter functions and band averaged emissivity tables for the radiometer channels.
Level 2-3A The level 2 data are profiles located at the measurement positions which are determined by the Sun-satellite geometry. The level 2-3A processing step takes these data and interpolates them onto a standard set of vertical levels evenly spaced in pressure, and onto standard times (level 3AT) and standard latitudes (level 3AL).
The data available through the BADC are in level 2 files. These consist of day files containing vertical profiles for all of the parameters measured. There are typically 29 or 30 profiles per day corresponding to each occultation event. The HALOE level 2 data are not public and users who require access to the HALOE data must apply for a web account at the BADC.
2.1 Spatial Coverage
2.1.1 Vertical CoverageThe parameters available in the files are listed below, together with their approximate vertical ranges
Vertical Altitude Parameter Resolution Range ________________________________________________________________________________________ Hydrogen Chloride (HCl) 4.5 km 10-60 km Hydrogen Fluoride (HF) 4.5 km 10-60 km Methane (CH4) 4.5 km 15-75 km Nitric Oxide (NO) 4.5 km 10-130 km (below 78 km) 6.5 km (above 78 km) 10-20 km Aerosol 2-3 km 10-50 km (HCl, HF CH4, NO Channels) Ozone (O3) 2-3 km 10-85 km Water Vapour (H2O) 3-4 km 10-75 km Nitrogen Dioxide (NO2) 2-3 km 10-55 km Temperature 4.5 km 10-130 km _________________________________________________________________________________________Note that these data have not been interpolated; the level 2 profiles are provided at the location of the measurements and are not on a regular pressure grid.
The vertical coverage is illustrated in Figure 2 .
2.1.2 Horizontal Coverage
As HALOE only makes measurements during sunrise and sunset it does not operate continuously.
The locations of the HALOE profiles varies with the spacecraft-Earth-Sun geometry. For a given day, the locations tend to be in two distinct latitude bands with 14 - 15 sunrises in one band and 14 - 15 sunsets in the other. These bands are typically in opposite hemispheres. The longitudinal separation of profiles in each band is approximately 24°, and a day's measurements therefore covers the full longitude range.
The latitudes of successive events of one type (e.g. sunsets), shift over the course of a day. The magnitude of the shift varies with latitude. Near the equator it may be as much as 10° per day, whereas at high latitudes measurements lie almost along zonal circles. The total sweep from high latitude in one hemisphere to high latitude in the other takes between 2 and 6 weeks, and the maximum latitude reached varies, the extremes being 80° S and 80°N. The latitude range covered by the measurements is shown in the BADC catalogue, and a typical range over the course of a year is shown in Figure 3.
A polar projection of the measurement points shows a pattern of measurements which appears to spiral towards the poles before changing direction and spiraling away.
2.2 Temporal Coverage
HALOE began operations on the 11th October 1991 (UARS day 30) until 21st November 2005. Since launch there have been a number of periods during which data collection has ceased. These are tabulated below, with comments.
Year dates UARS days Comments _______________________________________________________________ 1991 21 - 30 Dec 101 - 110 High beta 1992 1 - 10 Mar 172 - 181 High beta 9 - 14 Apr 211 - 216 High beta 3 - 11 Jun 266 - 274 Solar Array Problem 15 - 26 Dec 461 - 472 High beta 1993 24 Feb - 1 Mar 532 - 537 High beta 13 - 24 Jun 641 - 652 High beta 4 - 8 Aug 693 - 697 Solar Array Anomaly 27 - 28 Aug 700 - 701 High beta 18 Sep 738 Solar Array Anomaly 10 - 20 Dec 821 - 831 High beta 1994 8 - 18 Jun 1001 - 1011 High beta 23 - 25 Aug 1077 - 1079 High beta 7 - 15 Dec 1183 - 1191 High beta _______________________________________________________________
2.3 Resolution
The vertical resolution of the data varies with channel and is not equivalent to the vertical spacing in the level 2 file. Several factors contribute to the resolution. Instrument optics, signal measurement bandwidth and data processing procedures are the primary factors. Although the field of view projects to roughly 1.6 km at the limb, when combined with the electronic bandpass and solar sink/rise rate, the natural instrument resolution varies between about 2 and 3 km, (depending on the sink rate). Events viewed parallel to the orbital plane have faster sink rates than events viewed at large angles to the orbital plane (large beta angle)
The resolution can be further degraded by the processing. In fact, nearly any attempt at improving the stability of retrievals has the consequence of smoothing results in the vertical. The lower the signal-to-noise (S/N) the more instability, requiring more smoothing.
Each HALOE channel is processed with point spacing and smoothing chosen to give a good balance between stability and resolution. The low S/N regions (mainly upper altitudes) are smoother than the high signal regions. For the radiometer retrievals (NO2, H2O, O3 and aerosols) the advantages of the highly over sampled HALOE data are used to achieve results very close to the natural instrument resolution in regions of sufficient signal. The computationally intensive gas channels (HF, HCl, CH4 and NO) have been limited to 3 km point spacing in the processing, which limits effective resolution to about 4.5 km. At high altitude (dependent on channel, but always above the stratopause), an additional smoothing of the signal is performed, before processing, that is inversely proportional to S/N. This smoothing starts when signals drop below a S/N of 20 and is limited to no more than a 5 km smoothing. Consequently, resolution at the higher altitudes will be closer to 6 or 7 km, quickly improving as signal increases at lower altitudes. Although variable smoothing can violate conservation of signal rules, the error is a very small fraction of total estimated error.
Estimated resolutions for each species are tabulated in the instrument team's data quality document.
2.4 Data Quality
The data have undergone extensive validation. They have been compared to a variety of correlative measurements, as well as data from other UARS instruments. Although all channels have produced high quality results, there are known and suspected errors; these are outlined below, using information taken from documentation supplied by the HALOE team.
For each mixing ratio profile in the level 2 file, there is an accompanying quality profile which is a measure of the `random' error in that species, expressed in standard deviation.
2.4.1 General considerations
The single most difficult problem to address in processing HALOE data has been the increased aerosol extinction in the lower stratosphere due to the Mt. Pinatubo eruption (June 1991, 12 °N). The Aerosol cloud rose to ~40km in January 1992 and spread globally, before subsiding. The worst problems occurred in the equatorial lower stratosphere in the first few months of operation.
The aerosol has induced problems through two distinct mechanisms; contaminate absorption and degraded solar tracking.
Results from the radiometer channels may be in error by as much as 20-30% due to error in estimating the aerosol extinction under the heavy aerosol conditions prevailing after the Pinatubo eruption. Since the aerosol can contribute up to 80% of the absorption in some channels when aerosol levels are high (typically the tropics and subtropics from October 1991 and throughout 1992 around the aerosol peak).
The second mechanism, degradation of the ability to track the sun, leads to a reduced altitude registration accuracy. Early in the mission, HALOE profiles often cut-off with a minimum altitude in the mid stratosphere. There is an additional problem with the tracking mechanism when the sun is within a small angle of thick aerosol layers, which leads to errors in the mixing ratio profiles which are near constant or decrease with pressure, principally H2O and CH4. This problem can manifest itself as pockets of apparently dry air at the lower altitudes in the tropics and sub-tropics. The effect is typically under 10%, but can be as high as 30%.
The tracking fidelity also degrades at low altitude (below 18 km) for a variety of other reasons, making results below the 450K surface quite variable for all channels and all products.. The measurement at the bottom point in a profile is made close to the time at which the Sun lock was acquired or lost, so the time delay in the instrument response can result in a corrupted datum for this point. Considerable effort has been devoted to this problem, but there is some evidence that occasional rogue values persist.
Above the aerosol layer the data quality is limited primarily by the random errors tabulated in the data files. However, certain conditions related to both the atmosphere and instrument can be more error prone than others. Major systematic errors are mostly limited to the lower stratosphere and troposphere.
2.4.2 Individual species
For information about the quality of the measurements of individual species, users should refer to the documentation produced by the instrument team which gives a detailed discussion of measurement errors and vertical resolution for each species measured by the instrument.
2.5 Units
The units for the parameters present in the HALOE data files are tabulated below :
Parameter Units __________________________________________ Temperatures Kelvins Mixing Ratios By volume Quality Standard Deviation Column Sum molecules/cm(+2) Pressures mbar Transmission (none) Aerosol extinction km(-1) Altitude km Refraction factor (none) __________________________________________
3.1 Data Description
Level 3AT data consists of atmospheric profiles on a vertical spacing corresponding to the standard UARS pressure grid. In many cases this vertical spacing is much broader than the HALOE Level 2 point spacing. For this reason, HALOE Level 3AT data is not always representative of the HALOE Level 2 data from which it was produced. Therefore, care should be taken in interpreting the HALOE Level 3AT data between the UARS standard pressure levels. Also, the temperature profiles are retrieved on a 1.5 kilometer grid, interpolated to a 0.3 kilometer grid, and then interpolated to the UARS Level 3AT pressure grid. The temperature standard deviation values, on the other hand, are taken directly from the 1.5 kilometer gridded values. The standard deviation value of -999.0 indicates a temperature value from NMC or climatology.With the last few points of all HALOE Level 2 profiles being suspect, large variations may occur at the bottom of the profiles. On rare occasions the Level 3AT interpolation will wash out these variations and they will be reduced or not appear at all in the interpolated 3AT profile. In cases where this occurs, the comment associated with the quality number will not accurately describe the bottom of the profiles.
Level 3AT data is organized into thirteen files per day, one each for temperature, mixing ratios of H2O, NO2, O3, NO, CH4, HCl, and HF, and aerosol extinction values for CO2, NO, CH4, HCl, and HF channels. Each file contains the retrievals for the sunrise and sunset events for the day (typically 29 or 30 total), and files are organized by event. The two Level 3TP files are latitude and longitude for each pressure level of each event for the day.
3.2 Spatial Coverage and Resolution
Latitudinal coverage is from 80°S to 80°N over the course of a year and includes extensive observations of the Antarctic region during spring. The latitude and longitude locations of the actual retrievals vary with each retrieval point due to motion of the spacecraft during the length of the occultation event. For this reason, two HALOE level 3TP parameter files are produced, one containing the latitude and one containing the longitude at each UARS pressure level for each HALOE event. Also the level 3AT data are assigned a latitude/longitude location corresponding to that of the 30km retrieval point.
Horizontal resolution is 4° for level 3AL files and about 495 km along the orbital track for level 3AT files.
Vertical resolution for level 3A files is about 2.5 km between pressure surfaces.3.3 Grid description
All HALOE level 3A data have been referenced to the UARS standard pressure grid. The index of the data array defines the pressure level (in millibars) given by:
P(i) = 1000 x 10**(-i/6) mb, where i=0,1,2,...54 3.4 Temporal Coverage
HALOE began operations on the 11th October 1991 (UARS day 30) until 21st November 2005. Since launch there have been a number of periods during which HALOE data are missing or unavailable:Days Since Launch Date Ranges _____________________________________________________________________ 33 (14-OCT-1991) 35 (16-OCT-1991) 100 to 110 (20-DEC-1991 to 30-DEC-1991) 174 to 179 (03-MAR-1992 to 08-MAR-1992) 211 to 216 (09-APR-1992 to 14-APR-1992) 264 (01-JUN-1992) 266 to 303 (03-JUN-1992 to 10-JUL-1992) 360 (05-SEP-1992) 460 to 472 (14-DEC-1992 to 26-DEC-1992) 477 (31-DEC-1992) 533 to 538 (25-FEB-1993 to 02-MAR-1993) 540 (04-MAR-1993) 641 to 651 (13-JUN-1993 to 23-JUN-1993) 658 (30-JUN-1993) 694 to 697 (05-AUG-1993 to 08-AUG-1993) 716 to 719 (27-AUG-1993 to 30-AUG-1993) 744 (24-SEP-1993) 821 to 831 (10-DEC-1993 to 20-DEC-1993) 895 to 898 (22-FEB-1994 to 25-FEB-1994) 902 (01-MAR-1994) 1001 to 1012 (08-JUN-1994 to 19-JUN-1994) 1077 to 1082 (23-AUG-1994 to 28-AUG-1994) 1181 to 1194 (05-DEC-1994 to 18-DEC-1994) 1252 to 1263 (14-FEB-1995 to 25-FEB-1995) 1316 to 1321 (19-APR-1995 to 24-APR-1995) 1342 to 1349 (15-MAY-1995 to 22-MAY-1995) 1354 (27-MAY-1995) 1362 to 1374 (04-JUN-1995 to 16-JUN-1995) 1388 to 1395 (30-JUN-1995 to 07-JUL-1995) 1418 to 1419 (30-JUL-1995 to 31-JUL-1995) 1435 to 1443 (16-AUG-1995 to 24-AUG-1995) 1450 to 1451 (31-AUG-1995 to 01-SEP-1995) 1495 to 1497 (15-OCT-1995 to 17-OCT-1995) 1527 to 1528 (16-NOV-1995 to 17-NOV-1995) 1541 to 1553 (30-NOV-1995 to 12-DEC-1995) 1568 to 1569 (27-DEC-1995 to 28-DEC-1995) 1605 to 1623 (02-FEB-1996 to 20-FEB-1996) 1674 to 1676 (11-APR-1996 to 13-APR-1996) 1706 to 1709 (13-MAY-1996 to 16-MAY-1996) 1721 to 1734 (28-MAY-1996 to 10-JUN-1996) 1747 to 1749 (23-JUN-1996 to 25-JUN-1996) 1776 to 1784 (22-JUL-1996 to 30-JUL-1996) 1794 to 1808 (09-AUG-1996 to 23-AUG-1996) 1815 to 1820 (30-AUG-1996 to 04-SEP-1996) 1854 to 1860 (08-OCT-1996 to 14-OCT-1996) 1883 to 1891 (06-NOV-1996 to 14-NOV-1996) 1902 to 1914 (25-NOV-1996 to 07-DEC-1996) 1928 to 1932 (21-DEC-1996 to 25-DEC-1996) 1958 to 1959 (20-JAN-1997 to 21-JAN-1997) 1973 to 1984 (04-FEB-1997 to 15-FEB-1997) 1997 to 2000 (28-FEB-1997 to 03-MAR-1997) 2034 to 2039 (06-APR-1997 to 11-APR-1997) 2046 (18-APR-1997) 2062 (04-MAY-1997) 2066 to 2069 (08-MAY-1997 to 11-MAY-1997) 2083 to 2094 (25-MAY-1997 to 05-JUN-1997) 2101 to 2103 (12-JUN-1997 to 14-JUN-1997) 2108 to 2111 (19-JUN-1997 to 22-JUN-1997) 2130 to 2136 (11-JUL-1997 to 17-JUL-1997) 2139 to 2140 (20-JUL-1997 to 21-JUL-1997) _____________________________________________________________________3.5 Resolution
The gridded Level 3a product is put on UARS standard pressure grid points that result in roughly a 2.5 km point spacing. This can severely alias the HALOE O3, H2O, NO2 and aerosol data which often resolve real features under 2 km in height (although at reduced amplitude). Therefore, if the higher resolution information is important to the user, the Level 2 product should be used. It is planned for future releases to include a product for delivery to the DAAC with a format nearly identical to 3a but with a grid spacing closer to 0.5 km. This will coincide with future HCl, HF, CH4 and NO releases at point spacings near 1 km to attain a resolution closer to the natural 2-3 km instrument resolution. (Note that for this discussion we define a resolution, say 4 km, as a Gaussian apodization of the perfectly resolved signal transform in which the amplitude of an 8 km wave (needed to detect 4 km features) would be attenuated by 50%. Therefore even 1 km features may be detected with sufficient sampling, although the amplitude would be reduced by 94%.)3.6 Data Quality
The data have undergone extensive validation. They have been compared to a variety of correlative measurements, as well as data from other UARS instruments. Although all channels have produced high quality results, there are known and suspected errors; these are outlined below, using information taken from documentation supplied by the HALOE team.3.6.1 General considerations
The single most difficult problem to address in processing HALOE data has been the increased aerosol extinction in the lower stratosphere due to the Mt. Pinatubo eruption (June 1991, 12 °N). The Aerosol cloud rose to ~40km in January 1992 and spread globally, before subsiding. The worst problems occurred in the equatorial lower stratosphere in the first few months of operation.
The aerosol has induced problems through two distinct mechanisms; contaminate absorption and degraded solar tracking.
Results from the radiometer channels may be in error by as much as 20-30% due to error in estimating the aerosol extinction under the heavy aerosol conditions prevailing after the Pinatubo eruption. Since the aerosol can contribute up to 80% of the absorption in some channels when aerosol levels are high (typically the tropics and subtropics from October 1991 and throughout 1992 around the aerosol peak).
The second mechanism, degradation of the ability to track the sun, leads to a reduced altitude registration accuracy. Early in the mission, HALOE profiles often cut-off with a minimum altitude in the mid stratosphere. There is an additional problem with the tracking mechanism when the sun is within a small angle of thick aerosol layers, which leads to errors in the mixing ratio profiles which are near constant or decrease with pressure, principally H2O and CH4. This problem can manifest itself as pockets of apparently dry air at the lower altitudes in the tropics and sub-tropics. The effect is typically under 10%, but can be as high as 30%.
The tracking fidelity also degrades at low altitude (below 18 km) for a variety of other reasons, making results below the 450K surface quite variable for all channels and all products.. The measurement at the bottom point in a profile is made close to the time at which the Sun lock was acquired or lost, so the time delay in the instrument response can result in a corrupted datum for this point. Considerable effort has been devoted to this problem, but there is some evidence that occasional rogue values persist.
Above the aerosol layer the data quality is limited primarily by the random errors tabulated in the data files. However, certain conditions related to both the atmosphere and instrument can be more error prone than others. Major systematic errors are mostly limited to the lower stratosphere and troposphere.
3.6.2 Individual species
For information about the quality of the measurements of individual species, users should refer to the documentation produced by the instrument team which gives a detailed discussion of measurement errors and vertical resolution for each species measured by the instrument.
3.7 Units
The units for the parameters present in the HALOE data files are tabulated below :Parameter Units __________________________________________ Temperatures Kelvins Mixing Ratios By volume Quality Standard Deviation Column Sum molecules/cm(+2) Pressures mbar Transmission (none) Aerosol extinction km(-1) Altitude km Refraction factor (none) __________________________________________
For more background information, you may read the following documents: