Open Access
Add to BibSonomyAbstract
A hyperspectral remote sensing instrument employing a novel near-infrared supercontinuum light source has been developed for active illumination and identification of targets. The supercontinuum is generated in a standard normal dispersion multi-mode fiber and has 16 W total optical output power covering 1000 nm to 2300 nm spectral range. A commercial 256-channel infrared spectrometer was used for broadband infrared detection. The feasibility of the presented hyperspectral measurement approach was investigated both indoors and in the field. Reflection spectra from several diffusive targets were successfully measured and a measurement range of 1.5 km was demonstrated.
© 2014 Optical Society of America
1. Introduction
Active hyperspectral detection in the infrared (IR) spectral region is attractive for remote identification of various targets. Applications range from mineral exploration and identification of vegetation for diversity studies to the detection of objects and atmospheric constituents [1–3]. Active hyperspectral detection of diffusive targets with measuring ranges of a few hundred meters, employing rather expensive instrumentation for generating and detecting the IR radiation, have been reported [4–6]. Modern supercontinuum (SC) light sources [7,8] employing nonlinear optical fibers for the SC generation are attractive for remote hyperspectral sensing applications due to their unique combination of laser-like directionality, broad wavelength coverage and spectral intensity. However, typical commercial SC sources are limited in optical power due to the small core size of the highly nonlinear optical fibers used for SC generation. The optical power is one of the major factors determining the achievable measurement range in lidar (light detection and ranging) studies. However, as the theory and nonlinear optical phenomena behind the SC generation are well known [9], it is possible to tailor SC light sources to reach higher output power levels, required for long range hyperspectral lidar. In this work, a broadband SC light source having 16 W total optical output power in the near infrared spectral range was built from commercially available components.
Active hyperspectral lidars are typically applied to the measurement of solid targets. Natural solid targets exhibit two types of reflection: angle insensitive diffuse reflection, following Lambert’s cosine law, and strongly angle-dependent specular reflection, having a maximum at equal incident and reflection angles [10]. The balance between these reflection types is characteristic to the target and experimental geometry. In many applications as was also the case in this work, the specular reflection is avoided by measuring tilted diffusive targets. The transceiver geometry for active hyperspectral lidar can either be coaxial, having the transmitter and the receiver on the same optical axis, or biaxial, having the transmitter transversally separated from the receiver. For simplicity of the experimental setup, the coaxial arrangement is preferred in many cases.
The objective of the present study was to develop and test the feasibility of an affordable active hyperspectral instrument for the measurement of IR reflection spectra of diffusive targets over a measurement range of one kilometer. To achieve the goal, an instrument employing a high-power SC source and a coaxial transceiver design was developed.
2. Instrumentation
In this work, SC light in the near-IR spectral region was generated in a standard low-cost multimode graded-index fiber having 50 μm core diameter (Corning, InfiniCor), pumped with a 20 W fiber laser operating at 1064 nm (Nufern, NuQ-1064). SC generation in the fiber is caused by the interaction of several nonlinear optical effects such as stimulated Raman scattering and intensity dependent refractive index (optical Kerr effect) [9]. The nonlinear effects are strongly intensity dependent and thus the nonlinear processes are more efficient in the high intensity center part of the multi-mode fiber core, yielding a good SC output beam quality despite the fact that the fiber can support multiple modes [11–13]. The spectrally averaged beam quality parameter M2 measured in this work was about 2, whereas that of the pump laser is specified to be below 1.5. The fiber length of 100 m was found sufficient to allow for efficient SC generation yet avoiding excessive absorption losses in the fiber. The observed SC spectra at pump power levels between 3 W and 20 W are shown in Fig. 1.The SC starts to evolve from the pump laser wavelength towards longer wavelengths in the IR as the pump power increases. The spectral power density of the developed SC source is about 10 mW/nm.
Fig. 1 Normalized SC emission as a function of the pump laser power. The pump pulses were 100 ns in duration and the repetition rate was 25 kHz. The intensity dip at 1400 nm is caused by the hydroxyl group absorption in the silica fiber. The longest wavelengths are limited by the fiber’s material absorption.
The active hyperspectral measurement setup, realized in this work, is shown in Fig. 2. The setup consists of the high power SC source, transmitter optics employing either achromatic lenses or an off-axis parabolic (OAP) mirror, a receiving mirror telescope (Celestron, C5) and a commercial near-IR spectrometer (Ocean Optics, NIRQuest) for detection. The spectrometer can record 900 nm −2500 nm spectral range at 6.3 nm spectral resolution with the minimum integration time of 1 ms. Background spectrum with the SC switched off was automatically measured and subtracted from each individual recorded spectrum. The spectra presented in this work have been averaged over 16 spectral points corresponding to a spectral resolution of about 100 nm. The averaging is justified by the fact that no sharp spectral features are expected in the near-IR spectral region for the solid targets. The coaxial optical design, i.e., transmitter placed in front of the receiver on the same optical axis, was chosen for simplicity as no transversal realignment of the optics was required whenever the measurement distance was changed. An external rangefinder (Newcon Optik, LRF Module) was integrated and co-aligned with the transceiver to provide target’s range information. Focal lengths of the transmitter and the receiver were automatically adjusted using translation stages to optimize the instrument’s performance at the determined distance.
Fig. 2 Hyperspectral setup. The SC is generated in a graded-index multimode fiber and expanded by the OAP-based transmitter optics. The receiver telescope collects the reflected light into the spectrometer.
3. Results and discussion
The instrument was first tested at a short distance of 33 m in an indoor measurement range. Selected test targets were illuminated with the SC beam having a 20 mm diameter at the target. Two achromatic lenses were used to shape and transmit the SC beam in these tests. The receiver telescope was aligned with the illuminated point. The reflection spectra measured from a cotton fabric, zinc plate, sheet of white printer paper, transparent (in visible spectral region) polycarbonate plate, green leaf (Rubus odoratus) and varnished birch wood are shown in Fig. 3.The spectra are corrected for the response function of the spectrometer. Two normalization routines were applied: (a) in Fig. 3 shows normalization by the intensity value at 1100 nm which was observed to peak for most spectra, whereas (b) in Fig. 3 is normalized by the total integral and a white reference (Spectralon SRT-99-050) reflection spectrum, measured at the same distance and same ambient conditions. Normalization by the maximum intensity reveals spectral differences of the selected targets. However, the spectra are dominated by spectral features of the transmitted SC radiation as well as atmospheric water absorption bands close to 1400 nm and 1900 nm. The reflection spectrum of the white reference reproduces the transmitted SC radiation (data not shown). Therefore, normalization by the white reference eliminates the effect of the atmospheric absorption and the shape of the transmitted spectrum. All the presented near-IR spectra show apparent differences and are distinguishable from each other.
Fig. 3 Reflection spectra of cotton fabric, zinc plate, sheet of white paper, polycarbonate plate, leaf and varnished wood measured at 33 m distance using 2.5 s integration time. The reflection spectra are normalized by (a) the value corresponding to 1100 nm wavelength and (b) by the total integral and Spectralon white reference reflection spectrum.
The achromatic lenses used in these tests are specified between 1050 nm and 1620 nm. The instrument was tested for chromatic aberrations by slightly misaligning the receiver. However, no spectral distortions were observed in the spectra at this distance.
Two outdoor measurement campaigns were arranged to study instrument’s performance at ambient conditions over long target distances up to 1500 m. In the first campaign, the test field allowed for measurement distances up to about 500 m. The achromatic lens-based transmitter was used in this campaign. The white reference target was measured to compare the instrument’s performance at 150 m and 508 m target distances. The measured reflection spectra, normalized by the signal at 1268 nm wavelength (small atmospheric absorption), and the corresponding atmospheric transmissions (bright daytime conditions; 14.8°C, 88% RH, 1.017 bar, and 19.7°C, 55% RH, 1.017 bar in the afternoon for 150 m and 508 m measurements, respectively) are shown in Fig. 4.Atmospheric absorption is not taken into account in the presented spectra owing to low transmission at the wavelengths corresponding to water absorption bands. The spectral features of the white reference target are reproduced within the wavelength band specified for the achromatic lenses. However, outside the specification band, i.e. above 1620 nm, long wavelengths start to diverge and, hence, the signal decreases at longer distances. This effect of chromatic aberration was verified by simulating the transmitter optics’ performance at various distances using Zemax (Radiant Zemax, LLC) ray-tracing software. The beam spot size at the target distance of 508 meters was estimated to be below 20 cm within the 1050 nm - 1620 nm band, specified for the lenses, whereas the beam at 2200 nm wavelength covered an area of about 1.6 m in diameter. Based on the simulations and the measurements, it was therefore obvious that the wavelength bandwidth of the instrument was limited to 1050 nm - 1620 nm with the lens-optic transmitter.
Fig. 4 Reflection spectra of white reference, measured at 150 m and 508 m distances with 1 s and 4 s integration times, respectively (left axis). The reflection spectra are normalized by the value corresponding to 1268 nm wavelength. Simulated atmospheric transmission spectra corresponding to 150 m and 508 m measurement distances are also shown (right axis).
The results of the 508 m measurement range clearly showed the limitation of the lens-based design and, therefore, for the second field measurement campaign the lens-optical transmitter was replaced by a design employing an off-axis parabolic (OAP) mirror. Due to the differences in the surface quality between the lenses and the OAP, divergence of the beam was expected to be higher for the mirror-optical transmitter. Zemax simulations showed over three times larger spot diameter for the OAP transmitter at the distance of 500 m. However, the OAP eliminated the chromatic aberration of the transmitter and, therefore, the full SC emission bandwidth became useful even at longer distances. The higher divergence of the beam corresponds to about one order of magnitude decrease in the signal-to-noise ratio (snr). It was therefore necessary to increase the integration time of the detector to compensate for this signal loss.
The results from the long-range field campaign are shown in Fig. 5.The spectra, recorded from targets at 297 m and 408 m distances are nearly identical, indicating the consistency of the measurements. The spectrum, recorded from 1468 m distance, resembles the same spectral shape as for the shorter distances, however with lower snr (1 s, 1σ snr is 112, 109 and 21 for 297 m, 408 m and 1468 m, respectively). The atmospheric transmission is considerably lower for the longest range measurement, as the total light path is nearly 3 km. The atmospheric transmission at the wavelengths corresponding to strong water absorption bands at 1400 nm and 1900 nm is practically zero.
Fig. 5 Reflection spectra of a black & white target measured at 297 m, 408 m and 1468 m distances with 5 s integration time (left axis). The reflection spectra are normalized by the intensity value corresponding to 1545 nm wavelength. Simulated atmospheric transmission spectrum corresponding to a 1468 m measurement distance is also shown (right axis).
Atmospheric absorption was numerically simulated for both the indoor and the outdoor measurements using data from the HITRAN 2012 database [14] and the measured ambient temperature, pressure and relative humidity. Water vapor absorbs strongly at the 1.4 μm and 1.9 μm bands as can be seen in Fig. 4 and Fig. 5. Thus, depending on the target distance, the hyperspectral data could be used to estimate the total water content along the optical path and/or of the target.
The effect of ambient light was readily avoided by automated background subtraction. Without the background subtraction at daytime illumination conditions, signal is dominated by the ambient light and alignment of the optics onto the target becomes impossible for distances exceeding a few hundred meters. The coaxial optical geometry used in this work has advantages over the biaxial arrangements in measurements at varying target distances; the transmitter and the receiver are overlapped by definition and only focusing at the same target point is necessary for optimizing the signal strength at considerably different distances.
4. Conclusions
In this feasibility study, an active hyperspectral setup employing a supercontinuum light source for active illumination has been developed for remote measurements. The SC light source, implemented using a commercial fiber laser and commercial low-cost graded index fiber, shows a total optical output power of about 16 W in the wavelength range from 1000 nm to 2300 nm. The presented SC source design is straightforward to realize and is cost-effective for generating high-power near-IR continua. The study shows that the instrument is capable of spectral measurements of diffuse targets at distances up to 1.5 km.
Refractive turbulences due to temperature gradients and varying wind, humidity and aerosol content also affect the performance of hyperspectral remote measurements. At the kilometer-range measurement distances, atmospheric refractive turbulences play significant role in the signal stability, especially if the target size is close to the detector’s field of view, as the beam is randomly moving around the target. At shorter distances, spectral absorption could also be exploited to provide information about the ambient conditions. For example, atmospheric humidity could be estimated by exploiting the strong absorption bands of molecular water. In addition, weather conditions and aerosols such as fog, rain, snow and smoke could be estimated from the spectra.
The chromatic aberrations caused by the lens-based transmitter optics were overcome by using an off-axis parabolic mirror. Remaining aberrations can be further minimized by replacing the Schmidt-Cassegrain telescope with a Ritchey–Chrétien or similar. By slightly modifying the optics, the same telescope could be used both to transmit and to receive the light. In addition to simplicity of the hardware, using the telescope to transmit light both ways would maintain correct transverse alignment. The drawback of using a single telescope is the fact that some light is lost due to the secondary mirror placed on the optical path and the beamsplitter separating the input and output beams. In addition, the instrument could be further improved by employing a more sensitive detector array or a set of single detectors. Also integration of the presented setup with a scanning platform into a small footprint and robust instrument to allow for 3D measurements is planned. Further development would also require algorithm development for atmospheric absorption compensation and database comparison for target identification.
Acknowledgments
The authors would like to acknowledge the Scientific Advisory Board for Defence (MATINE) for funding the project. This research was supported by a Marie Curie European Reintegration Grant within the 7th European Community Framework Programme. TL acknowledges the financial support from the Academy of Finland. Authors thank Isto Nironen and Jyrki Pikkarainen from Millog Ltd. for providing the 1.5 km measurement range and for their assistance during the long-range measurements.
References and links
1. M. L. Nischan, R. M. Joseph, J. C. Libby, and J. P. Kerekes, “Active spectral imaging,” Lincoln Lab. J. 14(1), 131–144 (2003).
2. F. D. van der Meer, H. M. van der Werff, F. J. van Ruitenbeek, C. A. Hecker, W. H. Bakker, M. F. Noomen, M. van der Meijde, E. J. M. Carranza, J. B. D. Smeth, and T. Woldai, “Multi- and hyperspectral geologic remote sensing: A review,” Int. J. Appl. Earth Obs. 14, 112–128 (2012).
3. D. M. Brown, Z. Liu, and C. R. Philbrick, “Supercontinuum lidar applications for measurements of atmospheric constituents,” Proc. SPIE 6950, 69500B (2008). [CrossRef]
4. R. A. Lamb, “A review of ultra-short pulse lasers for military remote sensing and rangefinding,” Proc. SPIE 7483, 748308 (2009). [CrossRef]
5. T. Hakala, J. Suomalainen, S. Kaasalainen, and Y. Chen, “Full waveform hyperspectral LiDAR for terrestrial laser scanning,” Opt. Express 20(7), 7119–7127 (2012). [CrossRef] [PubMed]
6. D. A. Orchard, A. J. Turner, L. Michaille, and K. R. Ridley, “White light lasers for remote sensing,” Proc. SPIE 7115, 711506 (2008). [CrossRef]
7. J. M. Dudley and J. R. Taylor, Supercontinuum Generation in Optical Fibers (Cambridge University, 2010).
8. C. Kaminski, R. Watt, A. Elder, J. Frank, and J. Hult, “Supercontinuum radiation for applications in chemical sensing and microscopy,” Appl. Phys. B 92(3), 367–378 (2008). [CrossRef]
9. J. M. Dudley, G. Genty, and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. 78(4), 1135–1184 (2006). [CrossRef]
10. H. Hasegawa, “Evaluations of LIDAR reflectance amplitude sensitivity towards land cover conditions,” Bull. Geogr. Surv. Inst. 53, 43–50 (2006).
11. S. C. Buchter, “Broadband high power light source,” Patent Application Publication US2012/0099340 A1 (2012), http://patentscope.wipo.int/search/en/WO2010115432.
12. S. H. Baek and W. B. Roh, “Single-mode Raman fiber laser based on a multimode fiber,” Opt. Lett. 29(2), 153–155 (2004). [CrossRef] [PubMed]
13. A. Polley and S. E. Ralph, “Raman amplification in multimode fiber,” IEEE Photonics Technol. Lett. 19(4), 218–220 (2007).
14. L. Rothman, I. E. Gordon, Y. Babikov, A. Barbe, D. Benner, P. F. Bernath, M. Birk, L. Bizzocchi, V. Boudon, L. R. Brown, A. Campargue, K. Chance, E. A. Cohen, L. H. Coudert, V. M. Devi, B. J. Drouin, A. Fayt, J.-M. Flaud, R. R. Gamache, J. J. Harrison, J.-M. Hartmann, C. Hill, J. T. Hodges, D. Jacquemart, A. Jolly, J. Lamouroux, R. J. LeRoy, G. Li, D. A. Long, O. M. Lyulin, C. J. Mackie, S. T. Massie, S. Mikhailenko, H. S. P. Müller, O. V. Naumenko, A. V. Nikitin, J. Orphal, V. Perevalov, A. Perrin, E. R. Polovtseva, C. Richard, M. A. H. Smith, E. Starikova, K. Sung, S. Tashkun, J. Tennyson, G. C. Toon, Vl. G. Tyuterev, and G. Wagner, “The HITRAN2012 molecular spectroscopic database,” J. Quant. Spectrosc. Radiat. Transfer 130, 4–50 (2013).
