1 Introduction

Remote sensing measurements of trace gases are important to our understanding of planetary atmospheres. On Earth, methane (CH4) is the second most important anthropogenic greenhouse gas with approximately 25 times the radiative forcing of carbon dioxide (CO2) per molecule. Yet, lack of understanding of the processes that control CH4 sources and its potential release from stored carbon reservoirs contributes significant uncertainty to our knowledge of the interaction between the carbon cycle and climate change and challenges our ability to make confident projections of future climate. The importance of this problem is clearly reflected in the last IPCC (Intergovernmental Panel on Climate Change) report [1], the NASA Plan for a Climate-Centric Architecture [2], and the National Research Council Decadal Survey for Earth Science [3]. Currently, the Japanese GOSAT (Greenhouse gases Observing SATellite) mission [4] is making measurements of CO2 and CH4 from space using a passive spectrometer. Passive sensors measure reflected sunlight, and as such are limited to the sunlit areas of the planet. Their sensitivity falls off when measuring at high latitudes, increased cloud cover, and low surface reflectivity. An inherent error source with space-based passive spectrometers is variability in pathlength caused by changes in surface elevation and by optical scattering from aerosols and thin clouds in the illumination or observation paths [5]. Scattering modifies the optical pathlength and can produce large retrieval errors even for thin clouds [6]. Although a laser-based sensor may still be affected by optically thick clouds, it can overcome these limitations and make continuous measurements in darkness, at high latitudes, through thin clouds, and can unambiguously determine the optical pathlength as demonstrated by recent airborne campaigns [7].

A laser transmitter for a greenhouse gas differential-absorption lidar (DIAL) must have high enough pulse energy and a narrow enough spectral linewidth at the wavelength of the target gas absorption line [8–10]. Optical parametric devices are well suited for this application, as they have sufficient wavelength tuning range and energy, even at wavelengths where traditional laser gain media do not work (see, for example, [11] and references therein for a similar high-power OPO that generates Watt level average power for both ground- and space-based applications). At NASA Goddard Space Flight Center, we are developing a laser transmitter system for CH4 detection based on optical parametric devices at 1,651 nm [12]. As a first step to realize a CH4 space lidar mission, we have demonstrated CH4 column measurements from a high-altitude aircraft using an optical parametric amplifier (OPA) that emits 10-μJ pulses at 1,651 nm [13].

In this paper, we report on an optical parametric oscillator (OPO) system that was developed to show a conceptual transmitter design for a future lidar mission to measure CH4 from a low earth orbit. For the higher-energy system, we adopted an OPO instead of an OPA, because the output linewidth of the OPO is narrowed by the optical feedback inside the cavity. Our OPO system differed from other OPO-based lidar systems (see, for example, [14–16]) in its control scheme and in its high (5 kHz) repetition rate. We adopted the high repetition rate for high-precision lidar missions, such as Active Sensing of CO2 Emissions over Nights, Days, and Seasons (ASCENDS) [3, 10], to minimize errors due to the non-overlapping beam spots that can occur in low-repetition rate lidars [e.g., 50 Hz for MEthane Remote sensing LIdar missioN (MERLIN), a CH4 sounder mission [17] proposed by French Centre National d’Etudes Spatiales (CNES) in collaboration with the German Aerospace Centre (DLR) and scheduled for launch in 2016]. Here, we describe our OPO system for a CH4 lidar that does not require dynamic pulse-to-pulse adjustment. This is accomplished by combining an absolute frequency-locking technique, an optical phase-lock loop (OPLL), a phase modulation technique to lock the cavity length, and a fast electro-optical switch. We proved the concept of this scheme, achieving high robustness and narrow linewidth, at an energy level projected to be required for a future methane lidar mission.

2 Experimental setup

2.1 Overview

Figure 1 shows our experimental setup, and the lower right inset of Fig. 1 shows its concept in the frequency domain. The frequency of the master laser is locked to the center of the reference CH4 absorption line at λ 0 = 1,650.960 nm. The OPO cavity length is controlled so that the master laser is resonant within the cavity. The slave laser is offset phase-locked to the master laser, with a frequency offset of integer multiples (6 in our case) of the free spectral range (FSR) of the OPO cavity. The resulting slave wavelength is λ 1 = 1,650.900 nm (60 pm off the CH4 peak). These servo loops ensure that the OPO cavity length is stabilized against the target methane line, while both seed lasers transmit through the OPO cavity. The two lasers are fast-switched by an electro-optical switch and then injected into the OPO cavity as seed lasers. The pump laser generates a 1,064 nm pump pulse, whose timing is synchronized to the optical switching of the OPO seeds.

Fig. 1
figure 1

Experimental setup of the CH4 lidar transmitter based on OPO. Inset on lower right shows the concept of the setup. SG signal generator, QWP quarter waveplate, COL fiber collimator. Other abbreviations are explained in the main text

Full size image

2.2 OPO cavity

The nonlinear crystal is a 35-mm-long, 3-mm-thick magnesium oxide-doped periodically poled lithium niobate (MgO:PPLN). The end surfaces of the crystal were angled at 5.4° and an anti-reflective coating was applied to minimize optical feedback. We used a 31.3-μm poling period at 60.1 °C.

The OPO cavity is resonant at the signal wavelength (1,651 nm), and its length is locked to the master laser. The traveling-wave-type (ring) cavity with three mirrors allows the simplest implementation of locking the cavity length to the master laser, while simultaneously maximizing polarization isolation. The cavity locking beam circles counterclockwise without interacting with the pump beam in the OPO cavity, while the OPO seed beam circles clockwise and interacts with the co-propagating pump beam. The locking is performed by the Pound–Drever–Hall technique [18]. The master laser is phase-modulated at 200 MHz by a fiber-coupled phase modulator, and the demodulated reflected signal is fed back to the piezoelectric transducer (PZT) on the small cavity mirror at the corner. The control bandwidth of this servo loop is ∼10 Hz. The input and output mirrors have 4-m curvature and have 97 and 70 % reflectivities at the signal wavelength (1,651 nm), respectively. They have high transmissions (>97 %) at the pump and idler wavelengths (1,064 and 2,993 nm). The measured finesse of the cavity is 20 at the signal wavelength. The round-trip length of ∼280 mm makes the cavity FSR 1.086 GHz.

2.3 Master laser

The master laser source is a continuous-wave (CW), polarization-maintaining, fiber-coupled, distributed-feedback (DFB) laser diode (LD) at 1,651 nm. The instantaneous linewidth of DFB-LD is about 2 MHz, and its power is about 20 mW. It is frequency-locked to CH4 in the fiber-coupled gas cell using frequency modulation [19]. The reference cell is 5-cm long and contains ∼200-Torr of CH4. The same modulation used for the OPO cavity locking is utilized for absolute frequency locking. The transmitted light through the cell is detected, demodulated at 200 MHz, and then subsequently fed back to the injection current of the master DFB-LD. The control bandwidth of this servo loop is ∼100 kHz. We estimate that the long-term drift of the master laser frequency is less than 1 MHz, based on our experience on a CO2 cell and a similar DFB-LD [20].

2.4 Slave laser

The slave laser source is the same type of DFB-LD as the master laser. It is offset phase-locked to the master laser using the OPLL system demonstrated in our previous CO2 lidar work [20]. The beatnote between the two lasers is detected and frequency-divided by 64. Then, its phase is compared with a reference RF signal at f R = 101.79 MHz, which is generated by a direct digital synthesizer (DDS) inside a commercial IC chip (AD9858 from Analog Devices). A digital phase/frequency detector (PFD) detects the phase difference between the divided beatnote and the RF reference signal, driving a charge pump (CP) to generate an error signal proportional to this phase difference. After being processed by the servo circuit, the signal is fed back to the injection current of the slave DFB-LD, with a control loop bandwidth of ∼100 kHz. This results in a constant frequency offset of 6.515 GHz (=6 FSRs) between the master and slave LDs. The RF reference signal is generated from the 10-MHz rubidium (Rb) clock. The Rb clock and the CH4 reference cell ensure that the offset frequency and the cavity FSR do not drift away from each other. While in lock, the 1-MHz drift in the master laser causes a cavity length change of <2 nm through the OPO cavity length servo. This results in <7 Hz change in the OPO cavity’s FSR. Since this is much smaller than the OPO cavity’s transmission width of ∼50 MHz (the FSR divided by the finesse), there was no need to control the offset frequency actively. For the same reason, we did not adjust the offset frequency between the measurements, where the cavity may have 104 times larger mechanical drift (∼20 μm) and thus 0.07 MHz FSR change.

2.5 Pump laser

The pump laser for this experiment is built around two commercial 50-W diode side-pumped Nd:YAG laser modules. The pump laser is seeded by a single-frequency 1064-nm external cavity diode laser (ECDL) followed by a Ytterbium-doped fiber amplifier (YbFA), which provides 190-mW seed for the pump laser. The linear cavity has a BBO Pockels cell (PC) and a quarter-wave plate to allow active Q-switching of the pump laser. The 65-mm Nd:YAG rods are pumped by CW current of about 20 amps. This CW pumping allows us to use a simple cavity length control for the injection seeding. The end mirror of the linear laser cavity was mechanically modulated at 12 Hz using a PZT. The reflected leakage light from the cavity is picked off from the isolator, demodulated, and then fed back to the PZT, so that the reflected light is minimized on resonance. The output coupler has a reflectivity of 70 %. At the maximum pump current, the laser emits a 60-ns long, ∼1.2-mJ pulse at 5 kHz.

2.6 Other optical systems

The master and slave lasers are switched at 5 kHz by a fiber-coupled optical switch (Nanospeed series from Agiltron). The fast switching is enabled by the electro-optical effect, whereas other mechanical switching devices have slower speed and lower reliability.

Since the two seed lasers have unbalanced and insufficient power as OPO seeds after the optical switch, they are amplified by a semiconductor optical amplifier (SOA). The SOA is used in saturation mode, and it provides about 35-mW of seed power to the OPO cavity at the two wavelengths.

The output beams of the OPO are separated out by a dichroic mirror (DM) and an optical filter. The combined transmission of the dichroic mirror and optical filter at 1,651 nm is approximately 70 %.

3 Experimental results

3.1 Pulse energy

Figure 2 shows the output signal (1,651 nm) energy and the pump (1,064 nm) energy as a function of the injection current for one of the Nd:YAG rods. These were measured by a high rep-rate, pyroelectric energy sensor (J-10MT-10KHZ from Coherent). The maximum output signal energy was 220 μJ (after the 70 % transmission filter) at ∼1.2 mJ of pump energy. Therefore, the optical conversion efficiency from the pump to the signal was ∼20 or ∼29 % if the transmission of the filter and dichroic mirror is taken into account.

Fig. 2
figure 2

Signal and pump pulse energies as a function of the pump current for one of the Nd:YAG rods. Signal energy (left axis 1,651 nm) was measured after the 70 % transmission filter. Pump energy (right axis 1,064 nm) was independently measured before the OPO

Full size image

3.2 Pulse shape

Figure 3 shows the temporal pulse shapes of the generated signal, the pump after the OPO, and the pump before the OPO (measured separately). The temporal full-width half-maximum (FWHM) width of the signal pulse was about 60 ns, which was close to the pump pulse. The distorted Gaussian shape indicates that it is not transform-limited and that its spectral width is larger than 7 MHz (=0.44/60 ns [21]).

Fig. 3
figure 3

Measured temporal profiles for the 1,651-nm signal (black, left axis) and the 1,064-nm pump (gray, right axis). Dotted gray curve, original pump pulse shape before the OPO. Vertical scales are arbitrary

Full size image

3.3 Output linewidth

The linewidth of the OPO output was estimated using a scanning Fabry–Perot (FP) cavity and a test CH4 cell.

Figure 4 shows the transmitted light of the DFB-LD and the OPO signal from the 3-GHz FSR scanning FP. The envelope of each pulse was plotted for the OPO signal output, while the FP is scanned slowly compared to the repetition rate (5 kHz). Limited by the resolution of the scanning FP of ∼300 MHz, they showed essentially equivalent transmission curves.

Fig. 4
figure 4

Linewidth estimation using transmission curves from a 3-GHz FSR scanning Fabry–Perot. The pulsed OPO signal output and the CW DFB-LD were analyzed independently, and plotted on the same scale

Full size image

Figure 5 shows the measured transmission of the test CH4 cell using the pulsed OPO and the CW DFB-LD. For the OPO, we used the master seed laser only without locking it to the reference CH4 cell. This enabled scanning of its wavelength through the CH4 absorption line. The test cell had 30-cm length and contained 19-Torr of CH4. Again, the result of the OPO was very close to that of DFB-LD.

Fig. 5
figure 5

Linewidth estimation using low-pressure CH4 gas cell. The pulsed OPO signal output and the continuous-wave DFB-LD were transmitted through the cell independently. Theoretical transmission curves were also plotted for the cell with 300-MHz laser linewidth and for the atmosphere at an altitude of 400-km

Full size image

From the theoretical calculation (thick gray curve in Fig. 5) and the result from the scanning FP, we set an upper limit of 300-MHz linewidth for the OPO signal output, although it is probably narrower. For comparison purposes, we note that an OPA system based on the same pump and seed lasers gave ∼10 GHz linewidth at the same output signal level. This shows the effectiveness of optical feedback in an OPO, especially when there is no simple way to prepare high (>∼500 mW) seed power, as in our case at 1,651 nm.

3.4 Output beam quality

We measured the output beam quality of the OPO signal beam (M 2) by focusing it with a lens with a focal length of 180 mm. As shown in Fig. 6, the M 2 was measured to be ∼1.3.

Fig. 6
figure 6

M2 measurement of OPO signal output. The output beam was focused down by a lens, and the measured beam radii were compared with the Gaussian beam theory

Full size image

3.5 Open-path demonstration

We performed a measurement demonstration of atmospheric CH4 using the OPO system to demonstrate its capabilities. For this measurement, the OPO transmitter was configured exactly as in Fig. 1 so that the "on" (λ 0 = 1,650.960 nm) and "off" (λ 1 = 1,650.900 nm) wavelengths were used.

Figure 7 shows the experimental setup of the open-path demonstration. The OPO output goes through a beam expander and a wedge, which is set inside the integrating sphere. The scattered light inside the integrating sphere is detected by an InGaAs PIN detector and used as an energy monitor to normalize the return signal. The OPO beam is sent to a target 1.5-km away from the laboratory. The target is a square aluminum plate with sides 1.2-m long and is covered by a retroreflective tape (963 series from 3M). The reflected photons are collected by a 20-cm Cassegrain telescope and sent to another InGaAs PIN detector. The pulse signals from the detectors are integrated by boxcar averagers with 200-ns gate width. The pump laser pulse, the optical switch, and the data acquisition system are triggered by a common signal generator.

Fig. 7
figure 7

Diagram of the open-path measurement demonstration setup. The laser source part is simplified from Fig. 1. F optical filter, BE beam expander, W wedge, IS integrating sphere, DET detector, COL fiber collimator, DM dicroic mirror, ADC analog-to-digital converter

Full size image

Figure 8 shows the CH4 mixing ratio over 50 h measured by the OPO lidar and an in situ gas monitor (Picarro). The lidar data were converted to ppmv in a similar manner to [12]. The in situ monitor measures local CH4 using wavelength-scanned cavity ring down spectroscopy. A 60-s averaging period was used for the lidar and the in situ data. The lidar result had a similar trend to the monitored CH4 concentration. Their deviation can be in part explained by the difference between the lidar’s column measurement and the in situ device’s point measurement. For example, the peak at hour 18 may be from a CH4 puff from a nearby landfill. The lidar could observe it over several hours along the 1.5-km path. In contrast, the in situ monitor could sense it for a short time only when it passed by. The near-constant offset is most likely from the fact that we use the two wavelengths only and the temperature-dependent baseline tilt cannot be removed in the current system. The large scatter that appears occasionally in the lidar data is most likely from speckle noise.

Fig. 8
figure 8

Result of the open-path measurement. Thin black curve and thick gray curve the CH4 concentration measured by the OPO lidar over the 3-km round-trip path and by an in situ monitor set up near the OPO transmitter, respectively

Full size image

There was no loss of lock in the master seed laser, the slave seed laser, and the pump laser servo loops during the 50 h of measurement. The cavity lost the lock several times, possibly due to mechanical and electrical disturbances in the laboratory. It was automatically recovered by the servo system, typically within few seconds. This resulted in a 99.9 % measurement duty cycle. Those unlock points were rejected before averaging to plot Fig. 8.

The 1.5-km ground level path used in the demonstration is not very representative of the expected space measurement. Figure 9 shows the expected transmittance for the ground level path and from a nominal 400-km orbit for a US standard atmosphere. It is clear from Fig. 9 that the ground level path absorbance is a much smaller signal than the expected absorbance from space (∼0.22 and 0.95, respectively,absorbance = 1 − transmittance) and is also pressure broadened. In order to detect a 1 % change in the methane mixing ratio in the ground level path, we would have to detect a 0.0022 change in absorbance versus a 0.0095 change for the space case. Thus, in a sense, the ground level path is a more difficult measurement to make since we are looking for small changes on a smaller signal.

Fig. 9
figure 9

Comparison of the transmittances expected in the ground demonstration and in the spaceborne measurement. The two wavelengths used in the ground demonstration are also indicated

Full size image

4 Discussion

4.1 OPO cavity control scheme

Most other lidar transmitters based on injection-seeded OPOs use a heterodyne detection scheme [22], in order to make each seed wavelength resonant within OPO cavities. In this heterodyne scheme, the frequency difference (beatnote) between the OPO output pulse and the frequency-shifted seed laser is evaluated on a pulse-by-pulse basis, and the information is fed back to the cavity mirror. This scheme has been successfully implemented in low-repetition rate OPO lidar transmitters. However, it may not work with high repetition rate pump lasers, as the control speed can be limited by computation of the central frequency of the beatnote and by the mechanical motion of the cavity mirror. Additionally, it is not easy to know which way to drive the cavity mirror, since there is no phase information in the beatnote frequency.

On the other hand, our control scheme is based on novel phase-locked seed lasers. The seed lasers deterministically resonate within the OPO cavity. In principle, this scheme should work with a higher repetition rate and shorter pulse width, as there is no component that needs adjustment on a pulse-by-pulse basis. The main disadvantage of this scheme is that it requires opto-mechanical components for a counter-propagating locking beam, which adds some complications. OPLL can be implemented on a single circuit board, and it should not add major complications.

4.2 Applicability to space

There are currently no formal, published science and instrument requirements for a NASA space methane mission. However, the French–German MELRIN mission aims for a minimum (“threshold”) relative random error of 36 ppbv in the methane column abundance with a 50-km horizontal resolution. This science requirement would be sufficient “to resolve large wetland fluxes, inter-hemisphere gradients, seasonal and annual budgets on a continental scale” [17, 23, 24]. Our performance calculations for space IPDA lidars [9, 25] are based on similar science requirements as MELRIN and show that the laser energy would have to be scaled to at least 260 μJ in order to achieve a 1 % relative random error, which would be equivalent to 17.8 ppbv, assuming an evenly distributed methane column and a 1,780 ppbv mixing ratio. Our calculations use the typical lidar parameters (e.g., 400-km altitude, 0.5-m receiver telescope) and a HgCdTe electron Avalanche Photo Diode (e-APD) detector that has 70 % quantum efficiency and was developed in collaboration with DRS Technologies [26]. We demonstrated ∼300 μJ energy before the 70 % transmission filter. Therefore, our OPO concept was proved to work at an energy level required for a space mission.

The measured linewidth of <300 MHz is narrow enough compared to the expected absorption lineshape expected from orbit (thick dotted curve in Fig. 5). We believe the measured M 2 of 1.3 should not have any adverse impact on the projected measurement precision of 1 %. The candidate laser source for MERLIN has M 2∼2 [24]. Other successful space lidars flown by NASA, such as the Geoscience Laser Altimeter System (GLAS) [27], have an M 2 of <2. A higher M 2 would, among other things, make co-alignment of the transmitter path and the receiver field of view, very difficult. Since the seed lasers are locked to absolute references with wide bandwidths, they are virtually free from the errors due to wavelength uncertainty. Overall, we have successfully shown the basic specifications anticipated to be required for high repetition rate gas-sensing lidar transmitter from space.

4.3 Upgrade of seed laser

In principle, only two wavelengths (“on” and “off” the line) are needed to determine the transmittance through the atmospheric column, as demonstrated here. However, we have been proposing to use several wavelengths to sample the line at different wavelengths [13, 10, 28] in order to adequately sample instrumental and systematic errors, such as etalon fringes and baseline structures that may affect the measurement precision and averaging time [29]. Having multiple slave seed lasers should be a straightforward upgrade to the present system, as we have demonstrated 6 seed laser system at 1,572 nm previously [20]. For the gas lines that allow us to use fast widely tunable seed lasers at the telecom C-band or L-band, one could also apply the fast step-locking scheme [30] using a single laser instead of multiple lasers.

The ∼6.5 GHz frequency offset between the two seed lasers is too small to optimize the measurement of atmospheric CH4, and it is desirable to have >10 GHz offset to determine a more accurate baseline. The offset frequency here is limited by the bandwidth (∼7 GHz) of the detector to monitor the beatnote and the frequency divider. Using components with wider bandwidths, we have shown that ∼20 GHz offset locking is possible [30]. We hope to apply a similar technique for the CH4 seeder in the future for the “off” line points.

4.4 Upgrade of pump laser

The pump laser used in this experiment is not suitable for space use because of its low efficiency and low technology readiness level (TRL). As a replacement, we are looking into a Nd:YAG laser that is being deployed on the international space station as a part of the CATS (Cloud Aerosol Transport System) instrument [31]. The laser seems to satisfy our requirements for an OPO pump laser, such as single-frequency output at 1,064 nm, active Q-switch at 10 kHz, >1 mJ pulse, and high TRL for spaceflight.

5 Summary

Using the absolute frequency locking and the optical phase-lock techniques, we have demonstrated an OPO lidar transmitter that switches its wavelength on a pulse-by-pulse basis at a 5-kHz rate. The OPO output properties met tentative requirements for a space CH4 lidar mission. The system showed high robustness, as key systems are locked to absolute references and remain stable during dynamic switching. We are improving the pump laser and the seed laser systems to achieve higher TRL and measurement precision.