Abstract
Laser absorption Doppler velocimeters use a crossed-beam configuration to cancel errors due to laser frequency drift and absorption model uncertainty. This configuration complicates the spatial interpretation of the measurement since the two beams sample different volumes of gas. Here, we achieve single-beam velocimetry with a portable dual comb spectrometer (DCS) with high frequency accuracy and stability enabled by GPS-referencing, and a new high-temperature water vapor absorption database. We measure the inlet flow in a supersonic ramjet engine and demonstrate single-beam measurements that are on average within 19 m/s of concurrent crossed-beam measurements. We estimate that the DCS and the new database contribute 1.6 and 13 m/s to this difference respectively.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Velocimetry of gases is essential for many different flow applications such as meteorology [1–3], combustion [4–6], and aerodynamics [7–9]. For many applications, it is desirable to have a velocimetry sensor with a simple experimental setup and high enough spatial resolution to capture gradients in the flow. One powerful velocimetry method is laser absorption spectroscopy (LAS). As an optical method, it does not disturb the flow and additionally does not require significant optical access (in most cases) or tracer particles as with other optical methods such as particle image velocimetry or laser Doppler velocimetry. In LAS, a laser beam propagating through the sample is used to retrieve several path-averaged flow parameters such as pressure, temperature, species mole fraction, and velocity along the laser line-of-sight. Accommodating a single laser beam through the sample requires minimal hardware and optical access. Additionally, the single laser beam can provide spatial resolution corresponding to the length and diameter of the beam. However, velocimetry with LAS traditionally utilizes two crossed, counterpropagating beams [10–14]. A crossed-beam setup increases both hardware and optical access requirements which makes the sensor harder to incorporate into applications with size, weight and material constraints and also creates more failure points for the sensor which is important to avoid in especially harsh environments such as a hypersonic engine. Additionally, adding a second beam increases the spatial scope by representing a plane through the flow as opposed to a single line and thus worsens the spatial resolution of the measurement (particularly for transversely asymmetric flows) potentially losing important detailed information about the flow. In the current work, we demonstrate single-beam velocimetry. It is enabled by a stable, GPS-referenced dual frequency comb spectrometer (DCS) and a new H2O vapor spectroscopic absorption database that produces models with low absorption-transition positional error at high temperatures.
LAS relies on measurements of the absorption of laser light through a sample at frequencies resonant with the quantum transitions of the constituent molecules. If the gas has a bulk velocity ($U$) and the laser beam probes the gas at an angle ($\theta $) with respect to the flow direction (Fig. 1), then the position of the measured absorption feature(s) in static flow (${\nu _o}$) will be shifted to a new position ($\nu $) by the Doppler effect according to Eq. (1) (where c is the speed of light). The Doppler shift is then used to derive velocity as shown in Eq. (2):
The static line position is retrieved from a reference spectroscopic database such as HITRAN2020 [15]. However, an error (or drift) in the spectrometer frequency, ${\epsilon _s}$, or an error in the database line position, ${\epsilon _{db}}$, will impart an error in the derived velocity ($U^{\prime}$) as shown in Eq. (5) (where we replace ${\nu _0} + {\epsilon _{db}}$ from Eq. (4) with ${\nu _0}$ in the denominator since ${\nu _0} \gg {\epsilon _{db}}$). Thus velocity measurements using a single laser beam/path are very susceptible to absolute spectrometer frequency error and spectroscopic database error. For example, because there is no cancellation of laser drift and model error in a single-beam configuration, a < 50 m/s velocity error (<5% for the conditions shown in this paper) in a single-beam measurement at 35 degrees relative to the flow requires <0.0007 cm-1 error in both the relative and absolute frequency axis of the spectrometer. For the relative calibration of a CW laser sensor with an etalon, this would require knowledge and maintenance of the etalon length to ∼100 ppb. For absolute calibration and stabilization of the frequency axis of a CW laser, this would require active locking or tracking to a stabilized optical cavity or perhaps a low pressure reference cell.
The sensitivity to spectrometer and database error led past researchers to utilize a crossed-beam configuration. If a second beam is counter-propagated at an equal but opposite angle to the first (see Fig. 2), the velocity can be determined from the difference in measured line positions between the spectra of the two beams (${\nu _1} - {\nu _2}$) as derived in Eq. (7) and shown in Fig. 2.
In this case, database and spectrometer positional errors effectively cancel when deriving velocity, as demonstrated below in Eq. (8)–(10):
Thus, a crossed-beam measurement minimizes the effect of errors in both the spectrometer measurement and the absorption models (i.e., database error) by comparing the differential shift in laser absorption from the two beams [14]. However, a crossed-beam configuration requires additional experimental hardware (typically multiple sets of optics) and increases the spatial resolution of the measurement within the flow. In a crossed beam setup, it is assumed that both beams experience the same conditions (otherwise pressure shifts are not the same across the two beams and database error does not cancel out) meaning the plane of the flow should be uniform or at least transversely symmetric. Thus, the spatial resolution of a crossed beam setup represents a plane along the flow as opposed to just the line of a single beam and thus can lose some detailed spatial information of a transversely asymmetric flow. An alternative to the traditional crossed-beam setup is using two retroreflectors as demonstrated in Kurtz et al. [16]. Two retroreflectors facing each other create multiple passes across a sample region using a single laser. Each pass is at equal but opposite angles from the previous pass. This technique multiplies the pathlength and thus the SNR of the measurement while only requiring one set of transmit and receive optics. However, this setup also overlays the upstream and downstream Doppler-shifted absorption features on a single spectrum, which can impact the precision of the relative shift measurement if the features overlap (e.g. at lower velocity or higher pressure). The spatial footprint is also increased as the method results in an array of spatially dispersed beams.
Here, we demonstrate the first single-beam LAS velocimetry measurements. This is enabled through extensive work to minimize frequency errors stemming from both the spectrometer and the underlying spectroscopic database. In our implementation, we use a dual frequency comb spectrometer (DCS) that is clocked by a GPS-disciplined oscillator (GPSDO) and employs tight laser locking controls to realize low frequency drift (${\sim} 2 \times {10^{ - 5}}$ cm-1). Additionally, we fit our data using a new near-IR H2O database that derives linecenters and pressure shifts from high-temperature H2O-air laboratory measurements which we demonstrate to contribute a low frequency positional error in the model (${\sim} 2 \times {10^{ - 4}}$ cm-1). We take our single-beam measurements of velocity in a ground-test ramjet isolator located at Wright-Patterson Air Force Base and validate against crossed-beam measurements taken concurrently with the same DCS.
2. Reducing measurement errors
A single-beam velocimetry measurement requires minimizing frequency error in the spectrometer, ${\epsilon _s}$, and frequency error in the absorption model line positions from the spectroscopic database, ${\epsilon _{db}}$. In this study, we reduce ${\epsilon _s}$ by using frequency comb lasers with highly accurate and stable frequency measurement and control. Frequency combs are specialized lasers that emit pulses of light at a fixed repetition rate that is often well known and controlled. In the frequency domain, these pulses comprise a broad spectrum of tightly spaced frequencies of light or “comb teeth” which have a frequency spacing equal to the pulse repetition rate. The comb equation [17] defines the position (${f_n})$ for each comb tooth $(n )$ as shown in Eq. (11):
Thus, comb tooth positions are determined by the repetition rate (${f_{rep}}$) and the comb offset (${f_0})$ i.e., the carrier-envelope offset. To control the comb offset, we use the standard f-2f scheme where a frequency-doubled comb tooth from one end of the comb spectrum is heterodyned against a tooth from the opposite side of the spectrum to determine and stabilize the carrier-envelope offset [18–20]. To control the repetition rate of the laser, we lock one comb tooth to a continuous wave (CW) reference laser [21]. At the high comb tooth numbers for our measurements ($n \approx 1,000,000$), errors in repetition rate have a large effect on comb tooth positions. Thus, to minimize ${\epsilon _s}$, we must measure and control the repetition rate as accurately as possible.
In our system, we measure the repetition frequency with a detector and a digital frequency counter referenced to the GPSDO. The GPSDO has a 25 ppt relative frequency accuracy. At the comb frequencies used in this work (∼7000 cm-1), this frequency measurement accuracy results in a negligible contribution to ${\epsilon _s}$ of $1.75 \times {10^{ - 7}}$ cm-1 ($7000\; \textrm{c}{\textrm{m}^{ - 1}} \times 25 \times {10^{ - 12}}$). While the measurement of the repetition rate is accurate, there can still be drift in the repetition rate if the CW reference laser frequency is not controlled to maintain a stable frep. We use a control loop to minimize this drift by changing the current of the CW reference laser to maintain frep at a specific value. Our current control loop and electronics maintain the frequency of the CW reference laser within $2 \times {10^{ - 5}}$ cm-1. Relative to the CW reference laser operating wavelength (6410 cm-1), this is a frequency accuracy of 3 ppb and results in a value for ${\epsilon _s}$ of $2.2 \times {10^{ - 5}}$ cm-1 at the comb frequencies used in this work (∼7000 cm-1). Thus, the remaining CW laser drift is the main source of spectrometer frequency error.
A value of ${\epsilon _s}$ of $2.2 \times {10^{ - 5}}$ cm-1 results in a velocimetry error of 1.6 m/s (Fig. 3). This ${\epsilon _s}$ is a significant improvement over our previous DCS velocimetry works (Fig. 3). In previous works, we utilized a crossed-beam configuration to measure velocities with angles of 35° [14] and 12.5° [22] in part to overcome the absolute spectrometer drift. In our first work (Yun et al. 2022a [14]), we employed an oscillator that was not GPS-disciplined resulting in a value for ${\epsilon _s}$ of $1.1 \times {10^{ - 2}}$ cm-1 which leads to a velocity error of 750 m/s if a single-beam setup was used. For the second work (Yun et al. 2022b [22]), a GPSDO was employed but the drift of the CW reference laser was not as tightly controlled resulting in a ${\epsilon _s}$ of $1.8 \times {10^{ - 4}}$ cm-1 (velocity error = 35 m/s for a single beam configuration). The current system frequency drift contribution of 1.6 m/s velocity error to the single-beam measurement can be further improved with tighter locking of the CW laser frequency, but there are diminishing returns as other sources of velocity error become dominant.
To minimize ${\epsilon _{db}}$, we employ a new near-IR H2O spectroscopic absorption database which updates database parameters that directly impact line positions, i.e., linecenters and pressure shift. The linecenter parameter corresponds to the absorption frequency of a transition in a zero-pressure and zero-velocity environment. The pressure shift parameter characterizes the change in linecenter from collision-induced pressure shift. Collisions between molecules change the spacing of quantum energy levels of the molecules leading to shifts in the line position. Pressure shift, $\delta ,$ is commonly modeled using either a linear relationship or a power law relationship. The commonly used spectroscopic database HITRAN2020 [15], employs the linear relationship as shown in Eq. (12):
Here, ${\delta _x}$ is the pressure shift coefficient for a collisional partner x in the mixture and ${\delta _x}$’ is the temperature dependence of the shift. The total pressure shift is a summation of pressure shift contributions from each collisional partner, x, based on their respective molefractions, ${\chi _x}$. The linear relationship works best when multiple shift coefficients and temperature dependencies are derived for different temperature ranges [23]. However, currently HITRAN2020 only provides one pressure shift coefficient for each line optimized for 296 K and does not include temperature dependent shift parameters. Thus, we expect significant line position errors when applying HITRAN2020 to high temperatures. The new database used in this work based on Egbert et al. [24,25] utilizes the power law shown in Eq. (13):
The power law has been demonstrated to work well over large temperature ranges for infrared H2O transitions [26,27]. The new database derives pressure shift coefficients, ${\delta _x}$, and pressure shift temperature-dependent exponents, ${n_x}$, for the power law through a series of carefully controlled laboratory experiments measuring H2O-air mixtures with pressures ranging from 0.0007 to 0.79 atm, temperatures of 300 to 1300 K, and water mole fractions of 0.02 to 1.
3. Experiment
We collected dual frequency comb spectroscopy measurements in an axisymmetric, grounded, direct-connect dual-mode ramjet test facility at Wright Patterson Air Force Base. Measurements are taken in the 8.5 cm diameter isolator, which simulates the pre-combustion flow intake of a ramjet using an upstream flow distorter. For these tests, the flow conditions were approximately Mach 2 with temperatures ranging from 450–900 K. In our DCS, two frequency combs with slightly different repetition rates are mixed and the combined light is passed through the flow, after which it is collected onto a fast photodetector. The slight difference in repetition rates results in comb tooth position differences between the two combs that create a heterodyne signal on the photodetector in the radiofrequency (RF) domain [28]. This RF signal can be mapped back to the optical domain (THz region) based on the known optical frequency of each comb tooth (Eq. (11)) and enables a direct conversion of the detected RF frequencies to an optical spectrum.
For this work, we use two erbium-doped fiber combs and combine their light using an optical fiber coupler. A single GPSDO and a single CW reference laser provide the frequency control for both combs (Fig. 4), which creates mutual phase coherence between them. After combining the light from the two combs, we spectrally window the DCS light to span 6600–7400 cm-1 where H2O absorption is strong. We direct the light into the ramjet via quartz windows using a collimator angled at 36.6°. We send a second beam through the ramjet at an equal-but-opposite angle so we can derive crossed-beam values that serve as a reference for our single-beam values. The light traverses across the diameter of the axisymmetric ramjet where it is coupled through a lens and collected into multimode fiber on the other side. The multimode fibers connect to fast photodetectors whose signals are digitized and recorded.
We measure five different flow conditions with velocities ranging from 900–1200 m/s. Spectra are acquired at 625 Hz and averaged for 150s for each condition. For this first demonstration of single-beam velocity measurements, we measure across simple flow conditions with large flat cores, small boundary layers, and radially symmetric flow to simplify the validation with the crossed-beam measurements. Throughout these experiments, we track both the timing error of our clock and the drift in the CW reference laser as shown in Fig. 5.
Timing error is collected from the GPSDO at approximately 30 second intervals and represents the time difference between a clock-generated 1 pulse-per-second source and the GPS signal (Fig. 5(b)). An Allan deviation of the timing error over 6000 s demonstrates that 150 s of averaging amounts to a 25 ppt frequency accuracy (Fig. 5(a)). The Allan deviation increases from 30s to 100s due to the slow ∼100s drift in the timing error (Fig. 5(b)). Though we could get the same clock accuracy at 50 seconds averaging, we choose a longer averaging time to improve SNR in the laser absorption measurements to increase the velocimetry measurement precision.
Drift of the CW reference laser (Fig. 5(c)) is monitored by observing the difference between the repetition rate setpoint of the combs and the actual (measured) repetition rate for the duration of a measurement. The maximum drift of the CW reference laser across any of the measurements was $7 \times {10^{ - 5}}$cm−1 (dashed lines) but the drift stayed within $2 \times {10^{ - 5}}$ cm−1 on average (solid lines). The average drift should correspond to the actual error on the reported velocity measurement, as the spectra themselves are averaged and then fit to determine the velocity-induced Doppler shift. Though we average our measurements for the full 150 s, the plot in Fig. 5(d) shows the instantaneous velocity error contribution from the DCS frequency error. The maximum instantaneous velocity error contribution from the spectrometer is 6 m/s (dashed lines) but stays within 1.5 m/s on average (solid lines).
4. Results
As with our past DCS velocimetry works [14,22], the recorded spectra are fit using modified free induction decay cepstral analysis [29]. Spectra are fit to absorption models derived from the Egbert et al. database [24,25] and we float pressure, H2O mole fraction, temperature, and velocity in the fits. We derive single-beam velocities from both the upstream-propagating beam and the downstream-propagating beam by fitting them separately. An example fit from one of the downstream-propagating spectra is shown in Fig. 6(a-d). In order to validate the single-beam results, we also fit for crossed-beam velocities. In the crossed-beam fits, we input spectra from both beams to the fitting algorithm and constrain the fit to produce a single value for each parameter (pressure, H2O mole fraction, temperature, and velocity) across both spectra as done in Yun et al. [14]. A plot of the single-beam and crossed-beam velocities is shown in Fig. 6(e). For comparison, we also included a similar plot but with results generated using HITRAN2020 as the underlying spectroscopic database in Fig. 6(f). We chose HITRAN for this comparison because it is a commonly used database which has demonstrated more accurate spectra for near-IR H2O when compared to other databases [30].
For results derived with the Egbert database, single-beam values differ from the crossed-beam values by 10–27 m/s (0.9–2.7%) with an average of 19 m/s (1.8%). These correspond to shift errors ranging from $1.6 - 4.0 \times {10^{ - 4}}$ cm−1 with an average of $2.6 \times {10^{ - 4}}$ cm−1. Single-beam values derived with the HITRAN2020 spectroscopic database differ from crossed-beam values by 88–270 m/s (9–23%) with an average of 197 m/s (17.9%) corresponding to shift errors of $1.3 - 4.0 \times {10^{ - 3}}$ cm−1 (average = $2.7 \times {10^{ - 3}}$ cm−1). Most importantly, we can see that the velocity errors in the single-beam upstream and downstream measurements are of similar magnitude and in opposite directions. This suggests the majority of the error in the single-beam measurements derives from the database error (which would create this equal but opposite effect) and that the error from the Egbert et al. database is much lower compared to HITRAN 2020.
We perform an uncertainty analysis to quantify the contributions of the observed differences due to the experiment (i.e. the instrument, the optics setup) between the single-beam and crossed-beam velocity results to better understand the contribution of database error relative to other sources of uncertainty. Results from this analysis for measurement #2 are shown in Table 1.
For this uncertainty analysis, we only consider sources that contribute to the difference between the single-beam and crossed-beam results; thus, we do not include some sources described in our previous DCS velocimetry works [14,22]. For instance, background absorption removal will affect both values equally and thus not contribute to a change in the difference between the single-beam and crossed-beam result, so is not included here. We consider here the spectrometer error, uncertainty in the beam angle, and precision due to noise in the spectral data. Spectrometer frequency error derives from frequency error in the clock and drift in the CW reference laser as discussed in the Sections 2 and 3 and imposes a 1.8 m/s or a corresponding frequency shift error of $2.5 \times {10^{ - 5}}\; \textrm{c}{\textrm{m}^{ - 1}}$ on average for measurement #2. In the experimental setup, we determine the beam angle to within 0.3° uncertainty for all runs, which can affect the velocity retrieval by 0.34% (Eq. (2)) resulting in an uncertainty of 3.6 m/s ($5.4 \times {10^{ - 5}}\textrm{c}{\textrm{m}^{ - 1}}$). Precision due to measurement noise is determined by simulating spectra with noise characteristics similar to the data and deriving the scatter in the retrieved values from iteratively fitting these simulated spectra [31]. Precision is 3.5 m/s ($4.9 \times {10^{ - 5}}\textrm{c}{\textrm{m}^{ - 1}}$) for the single-beam velocity and 2.5 m/s ($3.5 \times {10^{ - 5}}\textrm{c}{\textrm{m}^{ - 1}}$) for the crossed-beam velocity. Precision is lower for the crossed-beam velocity because the crossed-beam measurement fit is based on twice as much spectral data [31]. To calculate the total uncertainty from the experiment, we sum the aforementioned uncertainties in quadrature.
We assume that the database error is the major unaccounted source after removing the total expected experiment uncertainty due to the above other sources from the difference between the single-beam and crossed-beam values and thus can estimate the database error as seen in Table 2. When fitting with Egbert, the remaining error ranges from 5.9 m/s ($8.8 \times {10^{ - 5}}$ cm−1) to 21.2 m/s ($3.2 \times {10^{ - 4}}$ cm−1) and is 14.1 m/s ($2.1 \times {10^{ - 4}}$ cm−1) on average. These errors are in line with the frequency uncertainty of the database itself (∼$2 \times {10^{ - 4}}$ cm−1) [24]. For HITRAN2020, the remaining error ranges from 84.0 m/s ($1.3 \times {10^{ - 3}}$ cm−1) to 263.9 m/s ($4.0 \times {10^{ - 3}}$ cm−1) and is 191.4 m/s ($2.9 \times {10^{ - 3}}$ cm−1) on average. The larger errors from HITRAN2020 are expected since this database uses a linear pressure shift coefficient reference temperature of 296 K which works best over small temperature ranges, while the flows measured here range from 500–900 K.
5. Conclusion
In this work, we demonstrate to our knowledge the first single-beam laser absorption velocimetry. We achieve this by using a mobile DCS system with accurate and stable GPS-based frequency referencing and by employing a new H2O spectroscopic absorption database that is derived from laboratory data taken over a large temperature range. The DCS system is updated from previous works with a GPS-disciplined oscillator and tighter control of the reference laser drift to reduce velocity error contribution by two orders of magnitude. The new H2O database derives pressure shift variables based on the power law to provide better model accuracy over large temperature ranges. Comparing fits with the new database to HITRAN2020, the new database produces single-beam values that improve agreement with the crossed-beam measurements by an order of magnitude. The new database produces single-beam values with an average difference of 19 m/s or $2.7 \times {10^{ - 4}}$ cm−1 with the crossed-beam values. By performing an uncertainty analysis, we determine that 13 m/s or $1.9 \times {10^{ - 4}}$ cm−1 of this difference is due to the database which is on par with the frequency uncertainty of the database itself; whereas the contribution from the spectrometer frequency referencing is only 1.6 m/s or $2.2 \times {10^{ - 5}}$ cm−1 on average.
Funding
Defense Advanced Research Projects Agency (W31P4Q-15-1-0011); Air Force Research Laboratory (FA8650-20-2-2418); Air Force Office of Scientific Research (FA8650-20-2-2418, FA9550-17-1-0224).
Acknowledgments
We would like to thank the test operators at RC-18, Steve Enneking, Andrew Baron, and Justin Stewart, who made sure we had everything we needed to take these measurements. This work has been cleared by the Air Force under case number AFRL-2023-4805.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time as the data contained in the paper has not yet been cleared for public release by the U.S. Air Force.
References
1. M. A. Shapiro, “Meteorological Tower Measurements of a Surface Cold Front,” Mon. Weather Rev. 112(8), 1634–1639 (1984). [CrossRef]
2. A. V. Garmashov, A. A. Kubryakov, M. V. Shokurov, et al., “Comparing satellite and meteorological data on wind velocity over the Black Sea,” Izv. Atmos. Ocean. Phys. 52(3), 309–316 (2016). [CrossRef]
3. V. Wirz, S. Gruber, R. S. Purves, et al., “Short-term velocity variations at three rock glaciers and their relationship with meteorological conditions,” Earth Surface Dynamics 4(1), 103–123 (2016). [CrossRef]
4. M. Shimura, S. Yoshida, K. Osawa, et al., “Micro particle image velocimetry investigation of near-wall behaviors of tumble enhanced flow in an internal combustion engine,” International Journal of Engine Research 20(7), 718–725 (2019). [CrossRef]
5. I. V. Walters, R. M. Gejji, S. D. Heister, et al., “Flow and performance analysis of a natural gas-air rotating detonation engine with high-speed velocimetry,” Combust. Flame 232, 111549 (2021). [CrossRef]
6. R. Rajak, S. R. Chakravarthy, and S. Ganesan, “Measurement of admittance and acoustic augmentation of burning rate of composite solid propellants using Laser Doppler Velocimetry,” Proc. Combust. Inst. 38(3), 4391–4399 (2021). [CrossRef]
7. C. J. Baker, S. J. Dalley, T. Johnson, et al., “The slipstream and wake of a high-speed train,” Int. J. Engine Res. 215(2), 83–99 (2001). [CrossRef]
8. K. E. Garman, K. A. Hill, P. Wyss, et al., “An Airborne and Wind Tunnel Evaluation of a Wind Turbulence Measurement System for Aircraft-Based Flux Measurements,” Journal of Atmospheric and Oceanic Technology 23(12), 1696–1708 (2006). [CrossRef]
9. K. M. Busa, B. E. Rice, J. C. McDaniel, et al., “Scramjet Combustion Efficiency Measurement via Tomographic Absorption Spectroscopy and Particle Image Velocimetry,” AIAA J. 54(8), 2463–2471 (2016). [CrossRef]
10. L. C. Philippe and R. K. Hanson, “Laser-absorption mass flux sensor for high-speed airflows,” Opt. Lett. 16(30), 2002–2004 (1991). [CrossRef]
11. O. Trinks and W. H. Beck, “Application of a diode-laser absorption technique with the D2 transition of atomic Rb for hypersonic flow-field measurements,” Appl. Opt. 37(30), 7070–7075 (1998). [CrossRef]
12. K. H. Lyle, J. B. Jeffries, and R. K. Hanson, “Diode-Laser Sensor for Air-Mass Flux 1: Design and Wind Tunnel Validation,” AIAA J. 45(9), 2204–2212 (2007). [CrossRef]
13. L. S. Chang, C. L. Strand, J. B. Jeffries, et al., “Supersonic Mass-Flux Measurements via Tunable Diode Laser Absorption and Nonuniform Flow Modeling,” AIAA J. 49(12), 2783–2791 (2011). [CrossRef]
14. D. Yun, R. K. Cole, N. A. Malarich, et al., “Spatially resolved mass flux measurements with dual-comb spectroscopy,” Optica 9(9), 1050–1059 (2022). [CrossRef]
15. I. E. Gordon, L. S. Rothman, R. J. Hargreaves, et al., “The HITRAN2020 molecular spectroscopic database,” J. Quant. Spectrosc. Radiat. Transfer 277, 107949 (2022). [CrossRef]
16. J. Kurtz, M. Aizengendler, Y. Krishna, et al., “Rugged, Scramjet Inlet Temperature and Velocity Sensor: Design and Ground Test,” AIAA J. 54(2), 399–407 (2016). [CrossRef]
17. T. Fortier and E. Baumann, “20 years of developments in optical frequency comb technology and applications,” Commun. Phys. 2(1), 153 (2019). [CrossRef]
18. H. R. Telle, G. Steinmeyer, A. E. Dunlop, et al., “Carrier-envelope offset phase control: A novel concept for absolute optical frequency measurement and ultrashort pulse generation,” Appl. Phys. B: Lasers Opt. 69(4), 327–332 (1999). [CrossRef]
19. J. Reichert, R. Holzwarth, T. Udem, et al., “Measuring the frequency of light with mode-locked lasers,” Opt. Commun. 172(1-6), 59–68 (1999). [CrossRef]
20. D. J. Jones, S. A. Diddams, J. K. Ranka, et al., “Carrier-envelope phase control of femtosecond mode-locked lasers and direct optical frequency synthesis,” Science 288(5466), 635–639 (2000). [CrossRef]
21. G.-W. Truong, E. M. Waxman, K. C. Cossel, et al., “Accurate frequency referencing for fieldable dual-comb spectroscopy,” Opt. Express 24(26), 30495–30504 (2016). [CrossRef]
22. D. Yun, N. A. Malarich, R. K. Cole, et al., “Supersonic combustion diagnostics with dual comb spectroscopy,” Proc. Combust. Inst. 39(1), 1299–1306 (2023). [CrossRef]
23. N. Stolarczyk, F. Thibault, H. Cybulski, et al., “Evaluation of different parameterizations of temperature dependences of the line-shape parameters based on ab initio calculations: Case study for the HITRAN database,” J. Quant. Spectrosc. Radiat. Transfer 240, 106676 (2020). [CrossRef]
24. S. C. Egbert, K. Sung, S. C. Coburn, et al., “Water-vapor absorption database using dual comb spectroscopy from 300 to 1300 K part I: Pure H2O, 6600 to 7650 cm-1,” J. Quant. Spectrosc. Radiat. Transfer 318, 108940 (2024). [CrossRef]
25. S. C. Egbert, K. Sung, S. C. Coburn, et al., “Water-Vapor Absorption Database using Dual Comb Spectroscopy from 300-1300 K Part II: Air-Broadened H2O, 6600 to 7650 cm−1,” arXiv, arXiv:2402.18036 (2024). [CrossRef]
26. R. R. Gamache and B. Vispoel, “On the temperature dependence of half-widths and line shifts for molecular transitions in the microwave and infrared regions,” J. Quant. Spectrosc. Radiat. Transfer 217, 440–452 (2018). [CrossRef]
27. X. Liu, J. B. Jeffries, and R. K. Hanson, “Measurements of spectral parameters of water-vapour transitions near 1388 and 1345 nm for accurate simulation of high-pressure absorption spectra,” Meas. Sci. Technol. 18(5), 1185–1194 (2007). [CrossRef]
28. I. Coddington, N. Newbury, and W. Swann, “Dual-comb spectroscopy,” Optica 3(4), 414–426 (2016). [CrossRef]
29. R. K. Cole, A. S. Makowiecki, N. Hoghooghi, et al., “Baseline-free quantitative absorption spectroscopy based on cepstral analysis,” Opt. Express 27(26), 37920 (2019). [CrossRef]
30. P. J. Schroeder, D. J. Pfotenhauer, J. Yang, et al., “High temperature comparison of the HITRAN2012 and HITEMP2010 water vapor absorption databases to frequency comb measurements,” J. Quant. Spectrosc. Radiat. Transfer 203, 194–205 (2017). [CrossRef]
31. D. Yun, W. B. Sabin, S. C. Coburn, et al., “Thermometry and velocimetry in a ramjet using dual comb spectroscopy of the O2 A-band,” Opt. Express 31(25), 42571–42580 (2023). [CrossRef]