et al., lInt. and AIRBORNE POLARIMETRIC, TWO-COLOR LASER ALTIMETER MEASUREMENTS OF LAKE ICE COVER: A PATHFINDER FOR NASA'S ICESAT-2 SPACEFLIGHT MISSION David Hardingl, Philip Dabne/, Susan Va/ettJ, Anthony yuJ, Aleksey Vasilyev2 and April Kell/ INASA Goddard Space Flight Center, Greenbelt, MD 20771 2Sigma Space Corporation, Lanham, MD, 20706 3National Space Club Scholars Intern, Goddard Space Flight Center, Greenbelt, MD 20771 ABSTRACT Incubator Program [7]. SIMPL is a pathfinder for more efficient, next-generation spaceflight laser altimeters The ICESat-2 mission will continue NASA's spaceflight including ICESat-2. It simultaneously measures surface laser altimeter measurements of ice sheets, sea ice and topography, roughness and slope as well as laser pulse vegetation using a new measurement approach: micropulse, scattering properties to differentiate surface types. The single photon ranging at 532 nm. Differential penetration of SIMPL transmitter is a short-pulse, 1064 nm (near infrared), green laser energy into snow, ice and water could introduce plane-polarized micropulse laser that is frequency doubled errors in sea ice freeboard determination used for estimation to 532 nm (green) and split into four parallel, push-broom of ice thickness. Laser pulse scattering from these surface beams. The receiver employs single-photon detection with types, and resulting range biasing due to pulse broadening, high timing precision at both wavelengths to determine the is assessed using SIMPL airborne data acquired over ice range to a target. The short pulse width and high timing covered Lake Erie. SIMPL acquires polarimetric lidar precision achieves an 8 cm range precision per single measurements at 1064 and 532 nm using the micropulse, detected photon. Upon aggregation of signal photons into a single photon ranging measurement approach. range histogram a measurement with a few cm resolution of pulse width is achieved. The impulse response from a flat, Index Terms- Lidar, sea ice, polarimetry, SIMPL, ICESat non-transparent surface has a full-width half-max pulse width of ~ 15 cm. 1. INTRODUCTION Received pulse broadening is due to surface roughness as NASA's Ice Cloud and land Elevation Satellite (ICESat) well as light transmission into the target and resulting acquired measurements of ice sheet elevations and their volume scattering. Due to limited near infrared (NIR) change, sea ice thickness, land topography and vegetation penetration into snow, ice and water the 1064 nm pulse height from 2003 to 2009 [1-4]. ICESat used a traditional broadening is primarily a measure of surface roughness laser altimeter measurement approach, single-beam, single whereas penetration in the green introduces additional pulse, analog waveform recording in the near infrared. The broadening and a resulting range delay. SIMPL acquires follow-on ICESat-2, scheduled for launch mid-decade, will reflected energy parallel ( II ) and perpendicular ( ) to the use new approach, multi-beam, micropulse, photon counting transmit pulse polarization plane, enabling a depolarization in the green [5,6]. Differential green wavelength measurement at the two wavelengths that provides penetration into snow, ice and water poses a potential source information on surface and volume scattering properties of bias in the measurement of sea ice freeboard used to used in differentiating target types. Increasing scattering estimate ice thickness. Freeboard is the height of the sea ice causes an increasing amount of reflected energy, and an surface relative to that of adjacent water in leads crossing increase in the depolarization ratio I II. The through the ice. The airborne Slope Imaging Multi depolarization signal is particularly useful in differentiating polarization Photon-counting Lidar (SIMPL) is being used water and smooth ice, whose surface is specular, from to assess potential sources of error in freeboard rougher snow and ice. The signal from specular surfaces, determination in preparation for the ICESat-2 mission. consisting of single scattering reflections, contains only II energy. ICESat-2 will have high precision range resolution 2. METHODOLOGY similar to SIMPL but will operate only at 532 nm with no polarization measurement. SIMPL's combination of highly SIMPL is a multi-beam laser altimeter developed through resolved pulse broadening at two wavelengths and the NASA Earth Science Technology Office Instrument polarimetric differentiation of surface types provides a Harding, et al., Proc. 2011 1m. Geoscience Remote Sensing Ju(v 2 2 2 (V")i 'V"i '<"I> z Z Z G:: G:: G:: (;) (') (;) 1 2 1 2 1 2 GRN _L3s GRN _L 3s GRN 1 35 2 2 2 (''") ''"" 'V"i G:: G:: G:: Z Z Z 0 2 0 2 1 2 NIR _L 38 NIR _1_ 3s NIR -'_ 3s Figure 1. Green (top) and near infrared (bottom) parallel versus perpendicular received energy, normalized with respect to a snow reference target, for open water (left), translucent grey ice (middle) and opaque white ice (right). Results are for one of the four SIMPL beams. unique set of measurements ideally suited to address biases surface properties. The nominal flight altitude was 4 km in sea ice freeboard determination that might impact the yielding 8 m spacing between the four beam profiles, ICESat-2 mission. providing a measure of surface slope at 24 m length scales. 3. INSTRUMENTATION 4. RESULTS FOR ICE COVERED LAKE ERIE Instrumentation details are provided in [7J. In summary, the SIMPL was flown aboard NASA's Lear-25 based at the SIMPL instrument consists of a multi-beam laser transmitter Glenn Research Center, acquiring data over ice-covered and receiver mounted on opposite sides of a thermally stable Lake Erie in February, 2009. The ice cover, analogous to optical bench and sharing a common 20 cm aperture young sea ice formation, consisted of four types that, from primary mirror. The laser transmitter is an 11 kHz, 1064 nm youngest to oldest, were fresh skim ice, dark nilas ice, microchip laser, frequency doubled to 532 nm, with a full translucent new grey ice and more opaque new grey-white width half-max pulse width of 1 nsec (0.15 m). The two ice. The lake ice is covered by snow in some places and colors are split into four plane-polarized beams and co broken 9Y open water leads and irregularly shaped polynya. alignment of the two colors is preserved. This co-alignment Surface types were identified in nadir in-flight video and is a key feature, ensuring that the pulse-width measurements correlated with the SIMPL data using GPS time-tagging. are acquired from the same target location. The 16 channel For each of the 16 channels the number of signal photons receiver splits the received energy for each beam into the over 0.1 second segments (- 12 m) were summed and the two colors and each color is split into energy parallel and amplitudes normalized by a snow reference surface. The perpendicular to the transmit polarization plane, thereby relative amplitudes of the parallel and perpendicular providing the measure of backscatter depolarization. The channels at the two wavelengths provide diagnostic solar background count rate is controlled by spatial filtering signatures that aid in differentiating the surface types (e.g., using a pinhole array and by spectral filtering using Figure 1). temperature-controlled narrow bandwidth filters. The receiver is fiber coupled to 16 Single Photon Counting Accumulation of signal photon ranges acquired over 0.5 sec Modules (SPCMs). To avoid range biases due to the long (- 60 m) yields range histograms documenting pulse dead time of these detectors the probability of detection per broadening for the four SIMPL channels. For an opaque laser fire on each channel is controlled to be below 30%, new grey-white ice target green light pulse broadening due using mechanical irises and flight altitude. The laser fire to penetration into the ice is only a few cm greater than that time and photon detection events are tagged with 0.1 nsec for NIR light, indicating the 532 nm range bias is very small precision, yielding a single-photon range precision of - At (Figure 2, top). Perpendicular-polarization received signal typical aircraft flight speeds, each of the 16 channels at both wavelengths is present due to surface and volume acquires a single photon range every 5 to 15 cm along the multiple-scattering. For an open water specular surface four profiles providing a highly sampled measurements of (Figure 2, middle) NIR reflected energy is minimal as Harding, et al., Proc. 2011 Int. Geoscience and Remote Sensing COI~(., Ju~y 24-29, VancoulJer, Canada _'i 1 i r':r/;.~ j 1m i ~ ,'j. "." - t I D6 . ~ _JI o. 1_ ._ __ __ ._ _ ; Q.' ____._ ... __ ~ .. __ ~ ... ~ ~f o ~ 100 1~ 200 0.11 <)2 C-'I oe 08 to Counts Normahzed /'lfnplitude 1m (J./:) 02 0.4 0.6 f.i.e 10 Counts Normat;zed Amptfuce 5m 2$ 00 7~ 100 OJ, 0:( Ci-4 OG <lJt !G Cumulntive P.HI»ni N(}<m;.!i:z(>(j Amplilu~ Figure 2. Range distributions for opaque new grey-white ice (top), water with a wind-roughened surface (middle) and water with a thin cover of skim ice (bottom) showing cumulative signal amplitude (left), number of detected photons (middle) and normalized amplitude (right). The distributions of the four channels are offset in range by small amounts to align the leading edges (at the normalized amplitude 50'h %) in order to emphasize pulse width differences. SIMPL channels: light green = green perpendicular, dark green = green parallel. orange = NIR perpendicular, red = NIR parallel. expected. NIR light that penetrates into the water is fully to the water column optical depth. Depending on the absorbed so the only NIR signal is II energy reflected from method by which the green range is deternlined from the the specular surface, yielding a pulse width slightly broader water column extinction profile, the range bias with respect than the instmment impulse response likely due to wind to the water surface could be 10's of cm for this case. A roughening of the water. The green II signal is also range bias of that magnitude into water relative to adjacent dominated by the specular return from the surface and ice would introduce meter-level errors in sea ice thickness. therefore has a pulse width equal to that of the NIR. The Note also that the peak green II amplitude is lOx weaker than amplitude and width of these returns are a function of the that of the wind-roughened open water case so that the range wind-induced surface roughness that defines the frequency precision will be substantially reduced. Skim ice on open distribution laser pulse incidence angles with respect to the water leads can be a common occurrence in polar region sea water surface [8]. A weak green _L signal is observed with ice. Its presence could introduce range biases and reduced an asymptotic tail to longer ranges due to visible light precision by the mechanism observed in the SIMPL data, penetration and multiple-scattering within the water column. thereby causing errors in ICESat-2 measurements of sea ice freeboard and derived ice thickness. Reflected energy from water covered by thin, smooth, specular skim ice enables fuller assessment of green Several methods (centroid, median, mode, leading edge and penetration and scattering in the water column (Figure 2, Gaussian fit) for determining the range from broadened ice bottom). Because the laser pulse is intercepting the smooth, and water column received pulse distributions are being specular ice at a small off-nadir incidence angle the large investigated as a function of signal-to-noise ratio (Figure 3). majority of the reflected light is directed away from the In this example, NIR and green II range distributions were receiver causing the very weak NIR II return. Similarly, the assessed using data for a block of clear sea ice acquired green II surface return must also be very weak so its using a laboratory breadboard version of SIMPL. For this observed signal is inferred to be due to volume scattering in sample the asymptotic tail of the green volume scattering the water column. This is consistent with the similarity of distribution extends 6 nsec (0.9) beyond that of the NIR. the green II and J_ signal amplitudes and asymptotic tails. Green range biases, relative to the NIR, and precisions were The green II and J_ distributions are therefore interpreted to computed as a function of signal level, from 30 to 1,000 be extinction profiles and the depth of penetration is related counts sampled from the green range distribution, and as a Harding, et al., Proc. 201 I Int. Geoscience and Remote Sensing Conf., July 24-29, Vancouver, 2.5 ~ 2.0 .(s/) c: o 1.5 ~ &. Noise rate = 8 MHz Normalized Amplitude 1.0 10 1000 o 0.6 U Gl .(s/) c: OA o ~ > +i=1~~11 aG l 02 1-532 II 1:1 <1l "0 c: !!l 0.0 (j) 10 100 Total Counts Figure 3. Assessment of range bias (top left) and precision (bottom left) as a function of signal level for 5 range determination methods, using single photon range distributions for parallel NIR and green received energy (bottom right). The position value, the measure of the range bias between green and NIR, is too large by 1.4 nsec due to the method by which the two distributions were referenced. The sample, measured with a laboratory breadboard version ofSIMPL, is a block or clear arctic sea ice provided courtesy ofT. Markus. For this example, an 8 MHz noise rate was applied and a signal threshold equal to 20% of the peak was used for the Gaussian fit method. function of noise rate. This example has an applied noise rate of 8 MHz. In order to reduce the effect of the tai I on the 11. REFERENCES Gaussian fit range bias only the signal above a threshold was used, defined as 20% of the peak signal (10% and 30% [1] H.1. Zwally et aI., "ICESat's laser measurements of polar ice, thresholds have also been tested). atmosphere, ocean, and land," J Geodyn., vol. 34(3-4), pp. 405-445, 2002. [2] B.E. Schutz, H.1. Zwally, CA Shuman, D. Hancock, 1.P. Other than the leading edge method, the range biases are DiMarzio, "Overview of the ICESat Mission," Geophys. Res. relatively insensitive to signal level but they differ between Lett., vol. 32: L21S01, doi:1O.1029/2005GL024009, 2005. the methods. The leading edge is least biased and the [3] R. Kwok, H.1. Zwally and D. Yi, "ICESat observations of centroid is most biased. As expected, the range precision Arctic sea ice: A first look," Geophys. Res. Lett., vol. decreases with increasing signal level for all methods. The 31 :L 1640 I, doi: 10 .1029/2004GL020309, 2004. leading edge method has the worst precision and the median [4] R. Kwok, G. F. Cunningham, M. Wensnahan, 1. Rigor, H. 1. is the best. This kind of assessment, applied to airborne Zwally, and D. Yi, "Thinning and volume loss of the Arctic SIMPL range distributions for a variety of Lake Erie surface Ocean sea ice cover: 2003-2008," 1. Geophys. Res., vol. 114: conditions, will help guide selection of an optimal range C07005, doi: 10.1 029/20091C005312, 2009. [5] W. Abdalati et aI., "The ICESat-2 Laser Altimetry Mission," determination method for ICESat-2. ProG. IEEE, vol. 98(5), pp. 735-751, 2010. [6] D.l. Harding, "Pulsed Laser Altimeter Ranging Techniques and 5. CONCLUSIONS Implications for Terrain Mapping", in Topographic Laser Ranging and Scanning: Principles and Processing, lie Shan SIMPL's polarimetric, high-precision single-photon ranging and Charles Toth, eds., CRC Press, Taylor & Francis Group, at green and NIR wavelengths provides a unique means to pp. 173-194, 2009. assess laser pulse interactions with natural targets. Analysis [7] P. Dabney, D. Harding, 1. Abshire, T. Huss, G. lodor, R. of ice covered Lake Erie data is documenting and Machan, 1. Marzouk, K. Rush, A. Seas, C. Shuman, Xiaoli quantifying pulse-broadening and signal amplitude as a Sun, S. Valett, A. Vasilyev, A. Yu and Y. Zheng, "The Slope Imaging Multi-Polarization Photon-Counting Lidar: an function of ice and water conditions. Resulting effects on Advanced Technology Airborne Laser Altimeter", Proc. Int. 532 nm range bias and precision are being assessed by Geosci. Rem. Sens. Symp., 11686732, DOl evaluating several methods for range determination. These 10.I109I1GARSS.201O.5650862, pp. 253-256, 2010. results will aid in developing data processing and analysis [8] 1.L. Bufton, F.E. Hoge and R.N. Swift, "Airborne methods for ICESat-2 measurements of sea ice freeboard measurements of laser backscatter from the ocean surface," and derived ice thickness. Appl. Opt., vol. 22, pp. 2603-2618, 1983.