Introduction

The climate models used for modeling the transport of the short and longwave radiation in the atmosphere require a knowledge of the aerosol and cloud optical properties. Because the optical properties of cirrus clouds are not well known, the measurements of cloud optical depth, phase function, and particle size provide important information. Clouds affect the radiative balance of the earth and its atmosphere by reflecting incoming solar radiation and trapping outgoing longwave radiation. Cirrus clouds have been found to have an important role on this process [1,2]. Cirrus clouds consist of large, nonspherical ice crystals and they are generally found between altitudes of 4 and 15 km. Compared to water clouds, cirrus clouds are generally optically thin. Studies based on climate models suggest that presence of cirrus may produce either a heating or cooling effect depending on the cloud optical properties and altitude [1,2].

The University of Wisconsin High Spectral Resolution Lidar (HSRL) measures optical properties of the atmosphere by separating the Doppler-broadened molecular backscatter return from the unbroadened aerosol return. The molecular signal is then used as a calibration target which is available at each point in the lidar profile. This calibration allows unambiguous measurements of aerosol scattering cross section, optical depth, and backscatter phase function. Also measurements of depolarization and multiple scattering can be performed. In this study, clear air aerosols, stratospheric aerosols, and cloud particles are all referred to as aerosols. Similar measurements of cloud optical parameters can be made with a Raman-lidar [3,4], but because the Raman backscatter cross section is about 1000 times smaller than the Rayleigh backscatter cross section, the HSRL has a significant signal strength advantage over the Raman-technique. Another advantage of the HSRL is that it can provide daytime measurements while sky noise background limits the measurements of the weak Raman signal to night time.

The basic idea of an HSRL was originally presented by Fiocco and DeWolf [5]. They proposed the measurements of atmospheric aerosols by interferometrically separating the backscatter signal. They demonstrated the principle of an HSRL with laboratory measurements of scattering from natural aerosols and artificially produced dense fog. A later experiment with the HSRL technique was performed by Schwiesow and Lading [6], who used Michelson interferometers in an attempt to measure atmospheric temperature. Their evaluation showed that a Michelson interferometer based measurements would theoretically produce accurate measurements of atmospheric temperature, but development of a functional system was limited by the quality of available optical components.

An investigation done by Shipley et al. [7] for a shuttle borne lidar experiment to measure global distribution of aerosols and their effects on the atmospheric heat budget started the University of Wisconsin HSRL research. The demonstration of the first University of Wisconsin HSRL was published by Shipley et al. [8]. The paper by Sroga et al. [9] presented the first results measured with the same system. The transmitter was based on a dye laser that operated at 476.8 nm wavelength. This laser provided only a 2--4 mW output and it operated at 100 Hz repetition rate. The backscatter signal was collected by a 0.35 telescope with a 320 rad receiver field of view. The filtering of the solar background was performed by using an interference filter and three low resolution etalons with a total bandpass of 2 pm. The separation between aerosol and molecular backscatter signals was based on a high resolution etalon with 0.5 pm bandpass. The HSRL was mounted in an aircraft. The system pointed down to ground and it therefore covered only about 2 km measurement range. The range dependence of the receiver spectrometer bandpass due to the angular sensitivity of the etalon transmission complicated the measurements.

The aircraft system was later modified for ground based operations and the changes were reported by Grund [10]. The problems due to the range dependence of the spectrometer bandpass were reduced by using a fiber optics scrambler [12]. The transmitter laser was changed to a CuCl laser operating at 510.6 nm wavelength. The laser output power was 50 mW at 8 kHz repetition rate. The receiver telescope and the high resolution etalon remained the same, but the number of low resolution etalons was decreased to two providing a 2.5 pm bandpass. This system was capable of measuring cirrus cloud optical parameters up to 12 km altitude. Later, the development of lasers enabled the use of an injection seeded, frequency doubled Nd:YAG-laser (532 nm) as the HSRL transmitter. The new laser provided a 0.7 W output at 4 KHz repetition rate. This change was reported by Grund and Eloranta [11].

After all the development in the HSRL, problems with the insufficient calibration accuracy and environmental sensitivity of the system operation remained. The HSRL produced accurate measurements of clear air and thin cirrus, but the measurement accuracy was not sufficient for the cases where the aerosol signal was large compared to the molecular signal. Eventhough the measurement accuracy was not high enough to provide direct measurements of the optical depth profile inside a cloud, the total cloud optical depth was obtained. The cloud optical depth was determined from the decrease in the molecular signal across the cloud. The calculations of optical depth profile inside a cloud were made by assuming a constant backscatter phase function with altitude. The measured molecular profile was fitted with a profile derived from the known molecular backscatter cross section that had been corrected for the atmospheric extinction. The extinction profile was obtained by applying the Bernoulli solution to the aerosol backscatter profile [12]. The measured optical depth was used as a constraint.

The accuracy of the HSRL calibration was improved and the sensitivity for environmental changes was reduced in the next University of Wisconsin HSRL (Eloranta and Piironen [13] ). In order to make the system operation more stable, the system was moved to a temperature controlled and vibrationally isolated environment. Participation in field experiments was made possible by moving the lidar to a semitrailer. The diameter of the receiver telescope was increased to a 0.5 m and the field of view was decreased to 160 rad. A new photon counting data system was designed and extensive computer control of the system operations was implemented. Depolarization and multiple scattering measurement capabilities were added. With the improved system, the measurements of cloud optical properties were performed at altitudes up to 30 km.

During the 18 first years of the University of Wisconsin HSRL, the separation between aerosol and molecular scattering was based on a Fabry-Perot etalon with a 0.5 pm bandpass. With this etalon, the signal was divided into two channels. The signal reflected from the etalon surface was measured with one channel. This channel contained the the wings of the molecular spectrum and the part of the aerosol spectrum that did not pass the high resolution etalon. The other channel was used to measure the part of the aerosol spectrum and the central part of the molecular spectrum that transmitted through the etalon. The basic idea of the spectral separation of a high resolution etalon based HSRL is presented in Fig 1.

  
Figure 1: The spectral response of a high resolution etalon based HSRL. The received signal is a combination of the Doppler-broadened molecular backscatter spectrum and the unbroadened aerosol spectrum. The aerosol signal and the center of the molecular signal that transmits through the etalon is detected with one channel. The other channel detects the signal reflected from the etalon.

For the aerosol backscatter signal, only a 2:1 separation ratio between aerosol and molecular channel is achieved by a etalon. The Fabry-Perot etalon based HSRL produces accurate measurements of clear air, thin cirrus, and stratospheric aerosols. However, when the system is used to probe dense water clouds, the aerosol signal becomes on the order of larger than the molecular return. Therefore, the inversion coefficients used to separate the aerosol and molecular signals must be known with better than 0.01% accuracy or otherwise some of the aerosol return will appear in the separated molecular return. Since the inversion coefficients for the etalon based system are known with only 0.1% accuracy, the measurements of dense water clouds are subject to error. The performance of a Fabry-Perot etalon is limited by its finesse and the angular distribution of incoming light. The etalon must be operated in pressure and temperature controlled environment, since better than a 0.1 mbar pressure tuning accuracy and 0.1 C temperature stability are required to keep the filter performance stable.

Shimizu et al. [14] proposed the use of an narrow-band atomic absorption filter in an HSRL and She et al. [15] reported high spectral resolution lidar measurements of temperature and aerosol extinction coefficient made by using a barium atomic absorption filter. These studies, and later studies from the same group [16,17], have shown, that an absorption filter provides a high rejection against aerosol scattering and therefore it makes the separation between molecular and aerosol scattering easier. Another advantage of an absorption filter is the stability of the absorption characteristics [14]. Furthermore, the transmission characteristics of an absorption filter are not dependent on the mechanical alignment of the filter [14] or the angular dependence of the incoming light. Also, a wide dynamic range in rejection against aerosol scattering is achieved by simply changing the vapor pressure [14] or the length of the cell.

This study presents an HSRL employing an iodine absorption filter. The spectrum of the electronic transition in molecular iodine has more than 22 000 absorption lines in the visible wavelengths [18], and 8 of them are easily reached by thermally tuning a frequency doubled Nd:YAG laser output [19]. Compared to the barium, the advantage of iodine is that instead of requiring a dye laser, a narrow bandwidth, frequency doubled Nd:YAG laser can be used. Also, a strong absorption is obtained in a short cell at room temperature. Even though iodine has extensive hyperfine structure, the absorption line width is similar to the barium line width, which is broadened by operating at a temperature of 500 C.

A large number of iodine absorption lines have been used as reference for Doppler-limited spectroscopy studies and also numerous spectroscopic studies of the line structure and hyperfine structure have been performed. Liao and Gupta [20] reported a use of an iodine absorption vapor cell as a narrow band filter for fluorescence spectroscopy. Lately Miles et al. [21] reported the measurements of flow field properties based on an iodine absorption filter and Filtered Rayleigh Scattering technique. The first iodine absorption filter based HSRL is presented here.

The basic idea of an iodine absorption filter based HSRL is presented in Figure 2. The received backscatter signal is divided to two channels. One channel detecting a sample from the total backscatter spectrum and the other channel the spectrum filtered by the iodine absorption filter. This signal contains information about the wings of the molecular spectrum and a small aerosol cross-talk signal. The first measurements made with the iodine absorption based HSRL were presented by Piironen and Eloranta [22].

  
Figure 2: The spectral response of an iodine absorption filter based HSRL. The detected backscatter spectrum is divided to two channels. One containing the information from the total backscatter spectrum and the other the wings of the molecular spectrum and a small aerosol cross-talk.

The work presented in this thesis is organized as follows. A short introduction to the HSRL theory is given in Chapter 2. After that, the instrumentation of the new iodine absorption based HSRL is given in Chapter 3. This chapter also presents the basic principles of the polarization and multiple scattering measurements. A more detailed description of the iodine absorption cell is given in Chapter 4. The system calibration and laser wavelength locking are discussed in Chapter 5. In Chapter 6, examples from the data obtained with the new HSRL are given. This chapter gives a more detailed description of the depolarization measurements and also shows the effects of multiple scattering on the measured depolarization ratio. In addition to the depolarization measurements, the measurements of scattering ratio, aerosol backscatter cross section, and optical depth are discussed. The measurements show, that the use of an iodine absorption filter enables accurate measurements of cloud optical parameters. Because the cross talk between channels can be accurately corrected and because the 160 rad field of view of the HSRL effectively suppresses multiple scattering, the optical depth inside a cloud can be measured. This makes future studies of scattering phase function possible. As a final example from the HSRL measurements the Chapter 6 shows an atmospheric temperature profile obtained by the HSRL. The accuracy of the HSRL measurements is discussed in Chapter 7 and the error analysis is presented.