1. Introduction

With the ever-increasing worldwide awareness of global warming, there is an intensifying demand for technologies able to assess the current situation as well as the impact of remedies through legislation, policies and technological countermeasures. Among greenhouse gases, carbon dioxide (CO₂) contributes the most to global warming, and monitoring of its concentration dynamics in the atmosphere is essential to predict the environmental impact [1–3].

The atmospheric CO₂-concentration can be mapped by several techniques such as gas correlation imaging [4], scanning Fourier Transform infrared imagers [5], tunable diode laser absorption spectroscopy [6], and differential absorption light detection and ranging (DIAL). DIAL is of particular interest, as it allows for real-time and potentially range resolved monitoring with high accuracy [7,8]. In conventional DIAL the wavelength of laser pulses is alternated between on and off CO₂-resonances while the backscattered echoes are collected and are time-of-flight resolved to get spatial resolution. The ratio of the recorded on- and off-signals, which cancels out the impact of instrumental and atmospheric conditions, is then used to calculate the gas concentration along the light path [9]:

Conventional laser systems for DIAL tend to be complex as both narrow linewidths (to resolve the absorption lines) and short pulses (for spatial resolution) are required. These systems are usually based on a narrow-linewidth laser seeding a tunable OPO and an amplifier for power scaling. More complicated systems based on external-cavity intensity-modulation [10,11], optical parametric generation [12] and optical parametric oscillators [13] have also been demonstrated to achieve tunable pulses with high energy. Furthermore, as the echoes consist of faint short pulses, the detectors must not only be sensitive but also fast, thus further increasing the cost and complexity of the overall system. Other advances reported in the literature employ random modulation of continuous-wave lasers with custom-built single photon detectors [14] and coherent detection of high repetition rate nanosecond lasers [15].

Recently, a simple and efficient scheme enabling continuous-wave DIAL (CW DIAL) was demonstrated and used to map the atmospheric O₂-concentration distribution over a 635-meter distance [9]. CW DIAL has also been used for atmospheric NO₂-monitoring [16] as well as for spatio-temporally resolving the atmospheric structure [17]. In this technique, which is based on the Scheimpflug Principle, echoes along the beam path are imaged onto a CCD line detector where each pixel corresponds to a given distance. This greatly reduces the demands on the detector as well as the source, since tunable CW sources are less complicated than tunable pulsed sources. Therefore, CW DIAL systems can be more compact, stable and cost-effective than conventional DIAL systems.

To access the spectral region required for CO₂-measurements, we have in this work designed a laser/amplifier source for the CO₂ overtone band at 1.57 µm, where the absorption of CO₂ is less interfered by other atmospheric components [18,19]. At this wavelength, reliable and high-quality lasers, amplifiers and detectors, originally developed for the telecommunication industry, are commercially available. The system is based on a narrow linewidth CW distributed feedback (DFB) laser, whose wavelength could be tuned about 2 nm around 1.572 µm by varying its operating current. We boost the output power of the DFB laser in an Erbium/Ytterbium doped fiber amplifier (EYDFA) to levels that allow for long distance, spatially resolved, CW DIAL measurements of the atmospheric CO₂-concentration.

2. Experimental setup

The experimental setup for the CO₂ CW DIAL measurements is shown in Fig. 1(a). The system consisted of the narrow linewidth CW fiber source at 1.57 µm, which was continuously tuned on/off the CO₂ absorption lines, and the control optics/electronics for transmitting, receiving and detecting the signal through the atmosphere.

 

Fig. 1 (a) CW DIAL setup of CO₂ measurement. The laser source is enclosed by the dashed blue line. HPI: high power isolator. LPI: low power isolator. EYDF: Er/Yb co-doped fiber. BC: beam combiner. PP: pump protector. DAC: digital-to-analog converter. DAQ: data acquisition board. PC: personal computer.

Download Full Size | PDF

The seed laser was a fiber pig-tailed single-stripe InGaAsP DFB laser (QDFBLD-1580-20, QPhotonics) with a nominal linewidth of 1 MHz. It could be tuned across the CO₂ absorption lines, between 1571.6 nm and 1573.0 nm, by changing the current from 95 to 158 mA, with the laser temperature controlled at 25 °C. The DFB output was coupled through a mating sleeve to a fiber isolator which in turn was spliced to an 11-meter-long double-cladding Er/Yb co-doped fiber (SM-EYDF-6/125-HE, NUFERN). The Er/Yb fiber worked as a power amplifier and it was counter-pumped by a 975-nm laser diode (LU0975T090, Lumics) through a fiber beam combiner (BC). A pump protector (PP) was used to protect the pump diode from signal radiation. A second isolator was placed at the output of the amplifier to suppress back reflections re-entering the amplifier.

For field measurements the output beam from the fiber amplifier was expanded and transmitted into the atmosphere by a homebuilt telescope, based on a short-wave infrared coated 300 mm focal length lens (Edmund Optics), a lens tube (SM3, Thorlabs) and a monorail focuser (Astroshop). The back scattered light from the atmosphere was received by a F/4 Newtonian telescope (SkyWatcher, Quattro series) with 800-mm focal length, separated from the expander by 814 mm. The angle between the optical axis of the expander and the receiver was 0.6°, and the field-of-view of the receiving telescope was 1.3°. The collected light was sent through a 1550 nm long-pass filter (OD4, Edmund Optics) and then focused onto an InGaAs detecting array (Xenics, Lynx-2048-250um-GigE), with 2048 pixels of 12.5 x 250 µm size, to resolve the CO₂-concentration over a 2 km distance.

3. Results and discussion

3.1 Tunable, high-power narrow-linewidth fiber source

The output power from the seed and the amplifier, with a launched pump power of 4.5 W, within the diode laser tuning range is shown in Fig. 2. As one can see, although the seed power was varied by almost 60% within this 2-nm tuning range, the output power of the amplifier was almost constant. This is attributed to a relatively flat gain over the tuning range, as well as similar accumulated gain for this span of input powers. The slight drop in power (3.7%) above 1572.4 nm was due to reduced gain at longer wavelengths, which could possibly be counteracted by using an even longer active fiber, and thus promoting increased reabsorption for shorter wavelengths. The output power was limited by the available pump power, but could readily be scaled further since the amplifier was not saturated, as indicated in the inset in Fig. 2. The linewidth after the amplifier was approximately 3 MHz over the entire tuning range as measured by a Fabry-Pérot Interferometer with 1 GHz free-spectral range (FPI, Toptica Photonics). As the linewidth of the CO₂ transitions at 1.57 µm in the atmosphere is about 1 GHz at normal pressure above sea level [20], the output from the fiber source was narrow enough for atmospheric CO₂ DIAL measurements.

 

Fig. 2 Output power characteristics of the seed DFB laser (red) and the fiber amplifier (black) at maximum pump power at different wavelengths. The inset shows the relation of output power and pump power at 1572.8 nm.

Download Full Size | PDF

Importantly, the amplifier could be continuously tuned over the entire tuning range while preserving the narrow spectral shape, seen in Fig. 3. However, in our experiment the tuning was step-wise set by the diode driver, which had minimum steps of 0.1 mA corresponding to wavelength steps of 0.65 pm (27 MHz). A wide scan spectrum of the fiber amplifier output is presented in the inset in Fig. 3, and shows that the amplified signal-to-noise ratio (SNR) exceeded 35 dB.

 

Fig. 3 Wavelength tuning by varying the seed’s driving current while keeping the pump at 4.5 W. The inset shows a wide scan of the spectrum with a 35-dB signal-to-noise ratio.

Download Full Size | PDF

To choose a proper pair of on/off wavelengths before in-field measurements, CO₂ absorption lines at 1.57 µm were recorded by launching the output at full power into a 56-cm long gas cell filled with CO₂ at 1 atm, at room temperature and measuring the transmission while sweeping the wavelength. The cell was slightly tilted from the normal incident angle to avoid a Fabry-Pérot effect as the glass windows of the cell were not anti-reflective coated for these wavelengths. The five absorption peaks (R12, R14, R16, R18, R20), that were covered during this sweep, are shown in Fig. 4 together with the CO₂ absorption lines from the HITRAN database [19]. It can be seen that the peaks have an offset of 0.07 nm relative to the HITRAN data which we attribute to a calibration error of the spectrometer. Nevertheless, the 3-dB absorption linewidth of the CO₂ under this condition was measured to be 0.037 nm, which was well resolved by the 0.65-pm tuning resolution of our fiber source.

 

Fig. 4 Measured (black solid) and 0.07 nm red shifted (dashed black) wavelength dependent transmission of the CO₂–filled gas cell by the fiber source. The corresponding position of the absorption peaks from the HITRAN database (red) are also shown for comparison.

Download Full Size | PDF

3.2 CW DIAL measurement

The CW DIAL measurements were carried out by expanding the laser beam into the atmosphere at an inclination of 4.3° over the city of Lund, Sweden (55°42′37.1”N 13°12′17.1”E), for a period of 15 hours between 17:00 and 08:00 from 7th January to 8th January, 2018. The sky was clear and the temperature ranged between −1.3°C to 14°C, as measured by a local weather station and the ambient pressure was around 1 atm. Experimental data was collected over 1000 exposures, 30 ms each, on the InGaAs detector.

In order to validate the performance of the DIAL system, a preliminary data analysis based on a subset of the collected data is presented here. For this feasibility test, we did not invest in a custom made bandpass filter and the signal is retrieved during night time. However, the light cone of the F/4 Newtonian receiver allows band pass filters down to 7 nm FWHM, and from experience with similar laser power we know such filters can sufficiently suppress daylight [21]. Figure 5 shows logarithmic plots of the back-scattered signal P_λ, normalized by a reference signal off the CO₂-resonances P_off, as a function of distance for three different wavelengths in the vicinity of the R14 absorption line. The data presented in Fig. 5 results from averaging measurements carried out over an hour. The black line corresponds to a wavelength off the absorption line, and thus has an essentially constant value close to zero. The red and blue curves correspond to wavelengths at the peak and flank of the absorption line, respectively. Consequently, the linearly fitted red trend line has a steeper slope, −0.155 ± 0.001 km⁻¹, compared to the blue trend line, which has a slope of −0.0588 ± 0.001 km⁻¹. Using these values in Eq. (1), together with cross section data from the HITRAN database [19], results in a CO₂ number density of N = (1.07 ± 0.04)∙10¹⁶ molecules/cm³, which corresponds to a CO₂-concentration of 408 ± 10 ppm (at a temperature of 280 K and normal pressure). This value is in good agreement with the global average CO₂-concentration at the Earth’s surface from 2017, 405.0 ± 0.1 ppm [22], and clearly demonstrates that our fiber source fulfills the necessary requirements for in-field CO₂ measurements with CW DIAL.

 

Fig. 5 DIAL data for three different laser wavelengths in the vicinity of the R14 absorption peak. The black triangles correspond to a signal from a wavelength spectrally off the absorption line, while the blue squares and the red circles correspond to signal from wavelengths at the flank and the top, respectively, of the absorption line. The straight red and blue lines indicate linear fits to the data points, whose slopes are used to estimate the atmospheric CO₂-conectration.

Download Full Size | PDF

4. Conclusions

A compact CW diode laser-fiber amplifier system has been developed and used for spatially resolved atmospheric CO₂ measurements. It is greatly simplified and cost-effective compared to traditional DIAL systems, as it only relies on a single CW laser-amplifier system. It has a linewidth of 3 MHz and a stable output power of 1.3 W over a finely tuned wavelength range of 2 nm, enough to cover 5 CO₂ vibrational-rotational absorption peaks. The results from the presented CW DIAL measurements of atmospheric CO₂-concentration demonstrate that the source meets the high demands to carry out atmospheric CO₂ monitoring with CW DIAL. In future work, we will focus on increasing the power by adding an additional amplifier to increase the possible monitoring distance, as well as to improve the noise suppression and data accuracy. The possibility of implementing range resolved measurements will also be done. In this study the emission wavelength was identified and calibrated by the path-integrated signals and a wavemeter, and it could also be actively stabilized for improved accuracy [23,24]. Furthermore, we plan to evaluate the system under different environmental conditions.