Development of Laser, Detector, and Receiver Systems for an Atmospheric CO2 Lidar Profiling System

A ground-based Differential Absorption Lidar (DIAL) is being developed with the capability to measure range-resolved and column amounts of atmospheric CO2. This system is also capable of providing high-resolution aerosol profiles and cloud distributions. It is being developed as part of the NASA Earth Science Technology Office's Instrument Incubator Program. This three year program involves the design, development, evaluation, and fielding of a ground-based CO2 profiling system. At the end of a three-year development this instrument is expected to be capable of making measurements in the lower troposphere and boundary layer where the sources and sinks of CO2 are located. It will be a valuable tool in the validation of NASA Orbiting Carbon Observatory (OCO) measurements of column CO2 and suitable for deployment in the North American Carbon Program (NACP) regional intensive field campaigns. The system can also be used as a test-bed for the evaluation of lidar technologies for space- application. This DIAL system leverages 2-mum laser technology developed under a number of NASA programs to develop new solid-state laser technology that provides high pulse energy, tunable, wavelength-stabilized, and double-pulsed lasers that are operable over pre-selected temperature insensitive strong CO2 absorption lines suitable for profiling of lower tropospheric CO2. It also incorporates new high quantum efficiency, high gain, and relatively low noise phototransistors, and a new receiver/signal processor system to achieve high precision DIAL measurements. Atmospheric tests of the laser have been conducted by operating it locked to the CO2 absorption line center, with off-set locking in the side-line mode, and in the off-line position. The reference laser is locked to center of absorption line within 390 kHz. This improves the level of stabilization by factor of 10 compared to earlier configuration. The detector has been characterized in the laboratory and by conducting atmospheric tests at The National Center of Atmospheric Research (NCAR), Boulder, Colorado. The receiver uses an F2.2 all aluminum 40 cm diameter telescope and the system is designed to focus light onto a 200 mum size detector. Subsystem level integration and testing has been completed in the second year. System level testing is planned in the third year along with validation in the late spring of 2008 that involves comparisons with ground-based and aircraft in situ CO2 sensors.

This DIAL system leverages 2-micron laser technology developed under a number of NASA programs to develop new solid-state laser technology that provides high pulse energy, tunable, wavelength-stabilized, and double-pulsed lasers that are operable over pre-selected temperature insensitive strong CO 2 absorption lines suitable for profiling of lower tropospheric CO 2 . It also incorporates new high quantum efficiency, high gain, and relatively low noise phototransistors, and a new receiver/signal processor system to achieve high precision DIAL measurements.
Atmospheric tests of the laser have been conducted by operating it locked to the CO 2 absorption line center, with off-set locking in the side-line mode, and in the offline position. The reference laser is locked to center of absorption line within 390 kHz. This improves the level of stabilization by factor of 10 compared to earlier configuration. The detector has been characterized in the laboratory and by conducting atmospheric tests at NCAR, Boulder.
The receiver uses an F2.2 all aluminum 16"diameter telescope and the system is designed to focus light onto a 200 micron size detector. Subsystem level integration and testing has been completed in the second year. System level testing is planned in the third year along with validation in the late spring of 2008 that involves comparisons with groundbased and aircraft in situ CO 2 sensors.

INTRODUCTION
The atmospheric burden of CO 2 is increasing in response to widespread anthropogenic combustion of fossil fuels. Roughly half of the emitted CO 2 is absorbed by the Earth's oceans and terrestrial ecosystems [1]. This uptake [2] varies annually from 1 to 6 PgC yr -1 . Understanding source/sink processes and the geographic patterns of carbon fluxes are primary goals of carbon cycle science. Uncertainty in predictions of the carbon cycle is one of the leading sources of uncertainty in projections of future climate [3]. A double-pulsed DIAL system operating in the 2.05-micron band of CO 2 is being developed for profiling CO 2 in the low-to-mid troposphere. There are several advantages of this system over passive remote sensing systems including day/night operation, reduction or elimination of interference from clouds and aerosols, and direct and straight forward inversion that leads to better quality data and faster retrievals with few assumptions. A ground-based lidar profiling system with ability to delineate atmospheric boundary layer (ABL) CO 2 from the free tropospheric CO 2 is needed that can operate during day or night. CO 2 distributions in the troposphere are linked to transport and dynamical processes in the atmosphere and are associated with near-surface sources and sinks. Annually averaged, inter-hemispheric, and continental to marine boundary layer CO 2 mixing ratio differences are on the order of 1 to 3 ppm [4]. Thus 0.2 ppm has long been a benchmark for required instrumental precision. Achieving this level of precision is difficult with remote sensors. Much larger mixing ratio differences emerge, however, at smaller spatial and temporal scales. In many instances exchange of ABL CO 2 with the free troposphere takes place through convective activity and passage of weather fronts. Hurwitz et al. [5], describe several synoptic passages and document 10 to 20 ppm mixing ratio changes that result from frontal passages. Thouret et al. [6], have shown that there is a high probability of observing more than one layered structure above the boundary layer at any time. Airborne sampling shows that the majority of the vertical structures in CO 2 mixing ratios are found within the lowest 5 km of the troposphere [4].
Thus, the requirements of this DIAL system development are: 0.5% (1.5 ppm) precision for vertical differences in the 30 minute mean mixing ratio resolved every 1 km from 0.5 to 5 km above ground.
Progress during the last two years in the design, development, atmospheric testing, and performance modeling associated with the development of this DIAL system is presented in this paper. Field evaluation of the DIAL will be conducted in year 3 of the project in coordination with field observations of the North American Carbon Plan (NACP) and OCO pre-validation activities.

THE CO 2 DIAL PROFILING SYSTEM
The DIAL system incorporates a high pulse energy, tunable, wavelength-stabilized, and double-pulsed laser that operates over a pre-selected temperature insensitive strong CO 2 absorption line in the 2.05-μm band. It incorporates a newly developed low noise and high gain InGaAsSb/AlGaAsSb (AstroPower) infrared heterojunction phototransistor (HPT) with a 200 μm sensitive area diameter. It is planned to operate it in the direct detection mode by taking advantage of a large 16" diameter telescope, which increases the photons collection at the detector by a factor of 16 over our previous system [7] in order to make CO 2 measurements above the boundary layer. The key lidar system parameters are given in Table 1. The Ho:Tm:LuLF tunable laser is configured to operate at the 2053.204 nm CO 2 absorption line and later on the laser is planned to operate on the more temperature in sensitive line [8] at 2050.967 nm using the Ho:Tm:YLF configuration. The CO 2 line will be fully characterized using a tunable highresolution (New Focus Model #6335) laser diode at the Jet Propulsion Laboratory [9]. Accurate spectroscopic parameters will be derived that are critical to realizing ground-based CO 2 lidar detection strategy. The lowpressure line position is known to an uncertainty less than 6×10  -5 cm -1 [9]. The ambient temperature line strength will be determined to 2%, the line width to 3%, and the atmospheric pressure shift to 5×10 -4 cm -1 using a multispectral fitting technique.

LASER DEVELOPMENT AND TESTING
Recent improvements in performance of the laser transmitter include double-pulse operation as demonstrated in the past with other DIAL systems. The double-pulse is injection seeded with different on-off wavelength for each pulse of the doublet. The wavelength switching is accomplished by having two injection seed lasers that can be rapidly (in under 1 μs) switched by an electro-optic device controlled by a simple logic signal. One of the seed lasers is tuned to the CO 2 line and the second is tuned to off line. The on-line laser is referenced to a CO 2 absorption cell at low pressure, and recent work has improved the performance of the wavelength locking to a level within 390 kHz standard deviation over hour-long time periods. This level of stabilization to line center reflects a factor of 10 improvements over our previous implementation, realized by converting to an external frequency modulation technique rather than wavelength dithering of the laser cavity length. An option now exists for tuning the on-line laser to the side of the line rather than the center of the line. By using the side of the absorption line, the optical depth of the DIAL measurement can be tailored for optimal performance. The side line reference  is made by locking one seed laser onto line center and referencing a second laser to the center-line laser by monitoring the heterodyne beat signal between the two. A feedback loop, with results shown in Figure. 1, has been implemented to lock the side-line laser to the center-line laser.  Electronic control holds an offset from center-line locked laser. Offset can be electronically programmed and laboratory tests have assessed quality of offset lock set upto for 2.8 GHz (37.3 pm). Atmospheric tests were conducted in a zenith-pointing mode at NASA Langley during the summer of 2006 using the heterodyne detection system to test the ability to operate in the sideline mode. Figure 2 shows the results of measurements of differential optical thickness as a function of altitude. The side-line was used as on-line and 3 different sideline positions were used that were offset by 24, 31, and 38 pm from line center. These results indicate ability to operate the laser in the sideline mode to optimize performance by tuning the sideline to desired absorption. More quantitative CO 2 measurements in the atmosphere were made in March, 2007 in the boundary layer using the Ho:Tm:LuLF laser and the existing heterodyne detection system.
These DIAL measurements were compared with in situ gas analyzer (LI-COR 6252, [10]) and initial results indicate that the two sensors show the same trend and occurrence of CO 2 perturbations and DIAL data show excellent precision.

DETECTOR SYSTEM INTEGRATION AND ATMOSPHERIC TESTS
There are no detectors available commercially that can meet the requirements of the DIAL system under development. An ideal detector would have high quantum efficiency (~70%), high gain (~100) with low noise equivalent power (~2E-14 W/Hz 1/2 ) and low excess noise factor of < 2.0, high bandwidth (> 1 MHz), and fast settling time (1-3 micro seconds to reach 1/100 signal level). A newly developed phototransistor (InGaAsSb/AlGaAsSb; AstroPower) with a 200 μm sensitive area diameter was used as the detector of choice [11]. The advantages of the phototransistor are its high gain (up to 3000), lower noise equivalent power (NEP), and higher quantum efficiency (~70%) compared to the traditional extended wavelength PIN photodiodes. These detectors are sensitive over the wavelength region 1.5 to 2.3 microns with peak performance near 2.05 micron. Characterization of these detectors showed the capability to provide high gain (~3000) with lower bandwidth and longer recovery times. To capture rapid variations of signals in the lower troposphere, a low gain setting for the phototransistor will be required for the near field and a high gain setting for the far field. Post detector electronics circuit was developed that consists of analog and digital electronic circuit elements. Figure 3 shows the configuration of the detection system electronics. The features of detection system electronics are: • Computer controlled detector bias and temperature control electronics. • Trans-impedance amplifier with dark current compensation. • Voltage amplifier with offset and gain adjustments. • State-of-the-art waveform 24-bit digitizer (National Instruments PXI-5922).
The capability of the detector system and its applicability to this program could not be demonstrated at Langley (while the receiver system was still under development  [12] of National Center for Atmospheric Research (NCAR) at Boulder, CO, was used for atmospheric testing by integrating newly developed phototransistors (HPT) into the REAL system. The REAL system has a 16" telescope, two 200-micron APD detector channels, and operates at 1.543-micron wavelength. Atmospheric tests of the detector system were conducted at NCAR initially in June 2006 and later in December 2006 to test the HPT. The 200-micron HPT was used in one of the APD channels and the other APD was used as a reference in the other channel. While the HPT detector is not optimum at 1.543-micron, still, the first atmospheric tests in June 2006 indicated that the HPT has sensitivity to detect atmospheric features (cloud and aerosol layers) to altitudes > 5 km. This was the first time a HPT was used in a lidar system [13]. These tests showed that the HPT could resolve atmospheric features ~100 m in size (even for a 3 V high-gain setting at 20 o C). However, the HPT detection system showed signal overshoot effects during recovery from a high signal that obscured measurements up to 3 km. Since the HPT in itself had not shown an overshoot before, therefore, amplifier circuit following the HPT was a suspect. Figure. 3. Block diagram of detector system electronics.
Electronic trouble-shooting was performed on the TIA circuit at NASA LaRC. This investigation indicated that the detector's bias node required a large capacitance to stabilize the bias junction. An improved circuit with a stabilizing capacitor was used in the circuit. Results of overshoot (blue) and the improved performance (pink) using the new circuit are shown in Figure. 4. When a 1 mW optical pulse was applied to the phototransistor without the proper capacitance, a long undershoot occurs (70 μsec), which in a lidar system will cause masking of measurements in the near field (up to about 3 km). Therefore, near field data for the first trip at NCAR was compromised. The problem was corrected for the second trip at NCAR. Figure 5 shows comparison of measurements with the test detector (HPT) with the standard InGaAs APD at NCAR at 1.543-micron wavelength. Clearly the near field overshoot problem is eliminated and boundary layer features are recovered compared to the measurements during the period of June 6, 2006. However, lower bandwidth and slower recovery problems cause systematic effects in data from the HPT system compared to the APD system as illustrated in data from a far field thin cirrus cloud at 10-11 km range shown in Figure 6.   To minimize systematic effects due to the bandwidth and slow recovery and to retrieve atmospheric lidar data from the HPT channel further lab characterization was conducted at NCAR including impulse response tests using short (0.1 μs) pulse laser and dynamic linearity tests using simulated lidar signal profiles [14]. Data from the impulse response tests were used to deconvolve measurements from the HPT channel. An iterative convolution technique was employed that is a modification of an earlier technique [15]. Results of deconvolution using this procedure are shown in Figures  7 and 8 for both HPT and APD channels.   Figure 7. Comparison of data from HPT and APD channels after deconvolution of the near field data. Figure 8. Comparison of data from HPT and APD channels after deconvolution of the far field data.
The deconvolution process eliminated recovery, increased resolution, and minimized phase delay between HPT and APD data channels as shown in Figure 8 [14]. The deconvolution procedure was fast and was implemented for the whole series of measurements with consistent results [14]. Some degradation of signal-tonoise (S/N) in the low S/N regions was observed.

RECEIVER SYSTEM
Two receiver channels are planned to capture the full dynamic range of signals in the near (within the boundary layer 0.5 to 2.0 km range) and far field (above the boundary layer >1 km). The optical receiver uses a 16-inch diameter F/2.2 all aluminum telescope to minimize influences of thermal effects. The optical scheme of the receiver is shown in Figure 9. The incident light is focused by the two mirrors of Cassegrain telescope through the pinhole and coupled into a fiber optic. The receiver optics includes a collimating lens, narrowband interference filter, focusing lenses, and protective window. A beam splitter is used to divide into two optical paths leading to detectors in the two channels. The optical design includes focusing the optical signal onto the 200-micron diameter detector.
Finally, the collimating and focusing lenses are of a triplet (3 lenses) design and fabricated from SF11 optical glass. These are custom coated for use at 2-micron wavelength. Off-the-shelf interference filters and beam splitters are used in the design. A custom designed all aluminum telescope (F/2.2) with a 400 mm diameter primary was manufactured by Welch Mechanical Design, Baltimore, MD. One of the specifications for this telescope was to have a small (45 micron diameter) focus area (blur spot size). The laser beam is transmitted into the atmosphere co-axially after a 20X beam expansion to limit the transmitted field of view to 85 micro-radians. The receiver FOV is set to 350 microradians using a pinhole at the focus of the telescope. Light from the telescope focus spot is coupled to the collimating lens of the aft.-optics system by a multimode optical fiber. The telescope, aft.-optics, detectors, and other components are placed inside an all-Aluminum enclosure box to limit stay light from the laser. This box also provides optical baffling and overall structural support. Figure 9. Schematic diagram of the optical design of the receiver system.