Routine High Spectral Resolution Measurements of Extinction: Challenges and Progress

The University of Wisconsin High Spectral Resolution lidar (HSRL) systems operate as Internet appliances in remote untended deployments. Lower cost versions of these systems may be widely deployed in future networks. Comparsion of the backscatter and depolarization profiles made with the UW ground based HSRL and the NASA Langley airborne HSRL will be shown. These demonstrate the robust absolute calibration of these HSRL measurements. HSRL systems also provides direct measurements of atmospheric extinction. Many applications require knowledge of the extinction cross section because of it's fundamental role in radiative transfer and it's more direct relationship to particle cross sectional area. Extinction measurements are more difficult than backscatter and depolarization and have larger errors. Backscatter and depolarization are derived from ratios of HSRL signals that cancel out range sensitive instrumental artifacts, while extinction measurements are derived from the molecular backscatter return alone. Furthermore, extinction is derived from the slope of the molecular signal. This amplifies noise due to photon counting statistics. Signal strengths decrease by nearly 6-decades within the ~20 km useful clear air range of the lidar or within just a few hundred meters in a dense water cloud. This paper will examine the sources of error in UW HSRL extinction cross section measurements and describe our progress in controlling them.

The most sensitive term in the lidar equation is the geometric correction that corrects for that fraction of the backscattered light that arrives at the field stop of the receiving telescope but fails to pass through the stop. Part of the light may be blocked because the image spot is not centered on the stop or because it is out of focus. The University of Wisconsin HSRL design uses a single telescope for both the transmitter and the receiver such that the transmit beam and the receiver field of view are coaxial. As a result the image spot is always centered on the field stop. However, the receiving telescope is focused on infinity and thus, the image of the scattering volume is poorly focused in the near range and is partially obstructed by the field stop until the lidar pulse is nearly 5km from our 100 microradian field-of-view receiver. The molecular extinction cross-sections in the boundary layer is ~ 1e-5 (1/m). Aerosol extinction on clear days is often less than the molecular. Thus, it would desirable to measure extinction cross-section values less than 1e-5 (1/m). The slope of the geometric correction thus must be known to at least 1e-6 (1/m) and ideally would be known to less than ~1e-7 (1/m). In the UW HSRL systems the slope of the geometric correction varies from ~5e-3 (1/m) at a range of 200 m to ~1e-4 at 1400 m and continues to contribute a correction up to nearly 5 km. This means that the slope of the correction must be known to at least 0.5% at 200 m and 1% at 1400 m and would ideally be known to better than 0.05% at 200 m and 0.1% at 1400 m. These are demanding error limits which are easily exceeded due to receiver optical alignment or laser beam shape modifications caused by small temperature changes.

By virtue of it's airborne vantage point, the NASA Langley HSRL is able to measure boundary layer aerosols with little influence from near range geometric corrections. Comparisons with the Langley data with our gr ound based data illuminate our geometric correction errors. The largest errors appear in the range interval between 3 and 5 km. This is where the focal spot on the field stop transitions from partial blocked to fully transmitted. In the course of a hot summer day, the slope of the correction appears to vary by as much 1e-5 (1/m) in this sensitive range interval. This occurs even though optical components other than the telescope are temperature controlled to ~1 deg C and the telescope uses carbon fiber members and a zerodur primary mirror to minimize it's thermal responses. Because we have not yet succeeded in suppressing this source of error in the primary HSRL channels, we have added a wide-field-of-view molecular channel receiver with a ~1 milliradian field-of-view. This transfers to the overlap sensitivity to a ranges of less than 400 m. Our progress in utilizing this data will be described.

Extinction errors can also be created shifts in the spectral purity of the transmitting laser, changes in the receiver bandpass, or failure of the laser wavelength to be exactly locked to the spectral filters in the receiver. We will describe a new seed laser and control electronics that have greatly reduced these error sources.

This paper will also discuss errors created by photon counting statistics and uncertainties in the knowledge of the temperature profile used in the HSRL inversion. A temperature sensing channel based on measuring the width of the molecular backscatter has also been added to the HSRL. This channel may eventually eliminate the need for an externally provided temperature measurements.