Open Access
Add to BibSonomy
Abstract
Accurate estimation of ground elevation on a large scale is essential and worthwhile in topography, geomorphology, and ecology. The Ice, Cloud and Land Elevation Satellite-2 (ICESat-2) mission, launched in September 2018, offers an opportunity to obtain global elevation data over the earth’s surface. This paper aimed to evaluate the performance of ICESat-2 data for ground elevation retrieval. To fulfill this objective, our study first tested the availability of existing noise removal and ground photon identification algorithms on ICESat-2 data. Second, the accuracy of ground elevation data retrieved from ICESat-2 data was validated using airborne LiDAR data. Finally, we explored the influence of various factors (e.g., the signal-to-noise ratio (SNR), slope, vegetation height and vegetation cover) on the estimation accuracy of ground elevation over forest, tundra and bare land areas in interior Alaska. The results indicate that the existing noise removal and ground photon identification algorithms for simulated ICESat-2 data also work well for ICESat-2 data. The overall mean difference and RMSE values between the ground elevations retrieved from the ICESat-2 data and the airborne LiDAR-derived ground elevations are −0.61 m and 1.96 m, respectively. In forest, tundra and bare land scenarios, the mean differences are −0.64 m, −0.61 m and −0.59 m, with RMSE values of 1.89 m, 2.05 m, and 1.76 m, respectively. By analyzing the influence of four error factors on the elevation accuracy, we found that the slope is the most important factor affecting the accuracy of ICESat-2 elevation data. The elevation errors increase rapidly with increasing slope, especially when the slope is greater than 20°. The elevation errors decrease with increasing SNR, but this decrease varies little once the SNR is greater than 10. In forest and tundra areas, the errors in the ground elevation also increase with increasing vegetation height and the amount of vegetation cover.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
A new space-based laser altimetry mission called the Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) was launched by the National Aeronautics and Space Administration (NASA) in September 2018. ICESat-2 is a spaceborne LiDAR system to obtain global data with a spatial coverage between 88° N and 88° S. The sole instrument onboard ICESat-2 is the Advanced Topographic Laser Altimeter System (ATLAS) [1–3]. The repetition rate of ATLAS is 10 kHz, which is 250 times that of the Geoscience Laser Altimeter System (GLAS) onboard ICESat. GLAS generates footprints with a diameter of 70 m, spaced at 170 m intervals. With a higher repetition rate, ATLAS generates overlapping footprints on the earth’s surface with a diameter of 14 m, spaced at 0.7 m in the ground track [4,5]. The extensive spatial coverage and dense sampling of ATLAS is conducive to measuring global heights including glaciers, sea ice, land, and forests, which will allow scientists to monitor sea-level changes [6–8], estimate forest structures and biomass [9–12], improve global digital terrain models [13,14], and investigate the global impacts of these measurements [15].
Unlike other LiDAR systems, such as discrete return or full waveform LiDAR [16–18], ATLAS employs a micropulse photon-counting system that is sensitive at the single-photon level. The detector of ATLAS can receive any return photons from the earth’s surface or solar background [19–21]. Therefore, ATLAS introduces a large amount of noise photons, especially in daytime data. Many noise removal and signal photon classification algorithms have been proposed for airborne-simulated ATLAS data, such as Multiple Altimeter Beam Experimental Lidar (MABEL) data and MATLAS data [22–27]. ATLAS data are now available for download. Considering the difference between ATLAS data and airborne-simulated ATLAS data, it is urgent to verify whether the existing algorithms are also suitable for ATLAS data and to verify the accuracy of ground elevation retrieved from ATLAS data. Neuenschwander et al. [28] demonstrated a first look at ATLAS data and verified the ground elevation accuracy of ATLAS data. The RMSE value of the ground elevation retrieved from ATLAS data is 0.85 m in Finland, which is lower than that from other existing ground elevation products such as the Shuttle Radar Topography Mission (SRTM) [18,29]. However, the study area, containing only approximately 1,000 samples, was too small to cover various types of land cover and terrain. In addition, no studies have verified the availability of existing algorithms for removing noise photons and retrieving ground elevation from ATLAS data and further analyzed the effects of various factors on ground elevation errors.
To assess the performance of existing algorithms on ICESat-2 data and verify the accuracies of ground elevations extracted from ATLAS data in large areas, our research was conducted in interior Alaska. This study area was chosen not only because large-scale airborne LiDAR data are available through NASA's Goddard Space Flight Center but also because of the complex terrain and various land cover types. The overall goal of this paper is to evaluate the performance of ICESat-2 data in ground elevation retrieval. The main aims of this study are (1) to test the performance of existing noise removal and ground photon identification algorithms on ATLAS data; (2) to validate the accuracy of ground elevations retrieved from ATLAS data over different land cover types; (3) to analyze the effects of the slope, signal-to-noise ratio (SNR), vegetation height and vegetation cover on the accuracy of ground elevations extracted from ATLAS data.
2. Materials
2.1 Study area
The study area is located in interior Alaska (61.53°N-65.76°N, 141.25°W-154.34°W), covering an area of approximately 550,000 km2 (Fig. 1). The topography is complex, and elevation values range from 50 m to 1950 m. According to GlobeLand30 data downloaded from the National Catalogue Service for Geographic Information (http://www.webmap.cn/mapDataAction.do?method=globalLandCover) [30], our study area is mainly covered by forest, tundra and bare land areas. Tundra is dominated by lichens, moss, hardy perennial herbs and shrubs.
Fig. 1. Study area and ATLAS ground tracks within interior Alaska. Counties, cities and land types are displayed.
2.2 ATLAS data
The ICESat-2 mission produces three levels of data products (ATL01 to ATL21). ATL01 and ATL02 are Level-1 products of ICESat-2. ATL03 is a Level-2 product containing time, latitude, longitude, and height above the World Geodetic System 1984 (WGS84) ellipsoid for each photon [31]. It provides all the photon information for Level-3A data products such as land-ice height (ATL06), sea-ice elevation (ATL07), and land and vegetation elevation (ATL08). Level-3B data products are gridded data such as the gridded land and vegetation height (ATL18) and the gridded monthly sea surface height on the open ocean (ATL19). In this paper, we used ATL03 data to conduct our research. ATL03 data from October 2018 to April 2019 were downloaded from the National Snow & Ice Data Center (NSIDC) website (https://nsidc.org/data/ATL03/versions/1) [32]. Multiple geophysical corrections have been applied to the ATL03 data, including atmospheric effects, tides and solid earth deformations, to provide corrected heights. The geographic coordinates of ATL03 are referenced to the WGS84 ellipsoid, while the vertical datum of airborne LiDAR data used in our study is at the mean sea level. To be consistent with airborne LiDAR data, the heights of the ATL03 data in our study have subtracted the geoid heights which were recorded in the ATL03 data.
There are three pairs of ground tracks in the ATL03 data (gt1l-gt3r). The two tracks in each pair, referred to as strong and weak beams, have an energy ratio of approximately 4:1 and are approximately 90 m apart across the ground tracks and 2.5 km apart along the ground tracks. Pair tracks are approximately 3.3 km apart across the ground tracks. Three ground tracks from strong beams were utilized in our study to obtain sufficient experimental data.
2.3 Airborne LiDAR data
Airborne LiDAR data and its corresponding products, such as the digital terrain model (DTM) and the canopy height model (CHM), were downloaded directly from Goddard’s LiDAR, Hyperspectral, and Thermal Imager (G-LiHT) website (http://gliht.gsfc.nasa.gov/)[33,34]. The airborne LiDAR data used in our study were collected in 2014 with vertical and horizontal accuracies of 0.1 m and 0.3 m, respectively. The resolutions of the DTM and CHM are 1 m. Slope maps were generated from the resampled DTMs with 30 m resolution. Mean vegetation height products and vegetation cover products with 30 m resolutions were generated from the CHMs. The projection system of all products generated from airborne LiDAR data was the Universal Transverse Mercator (UTM) WGS84 datum.
3. Methods
3.1 Ground photon extraction
To verify the performance of existing noise removal and ground photon identification algorithms on the ATL03 data, algorithms proposed by Zhu et al. [27] were utilized in our study to filter out noise photons and extract ground photons. The noise removal algorithm mainly includes two steps. First, an elevation frequency histogram was built to remove apparent noise photons. Second, a local photon density distribution histogram was established to eliminate the remaining noise photons. The ground photon identification algorithm consists of four key steps. To summarize, the first step is to build a local elevation frequency histogram and extract the photon at the lowest elevation peak with the maximum density as the initial ground photon. The second step is to extract accurate ground photons using an empirical mode decomposition (EMD) algorithm. The third step is to increase the number of ground photons based on progressive densification. The last step is to extract all ground photons utilizing cubic spline interpolation. However, these algorithms may classify noise photons as ground photons in areas with dense clouds. Considering that the clouds are far from the earth’s surface, our study utilized the elevation information from the SRTM DEM to remove incorrectly classified ground photons caused by clouds. If the absolute difference between the elevation of ground photons and the corresponding elevation of the SRTM DEM is smaller than 50 m (determined by a trial-and-error approach), this photon was classified as a true ground photon.
3.2 Accuracy validation by airborne LiDAR data
DTM data derived from airborne LiDAR data were utilized to assess the accuracy of the ground elevation retrieved from the ATLAS data. The ground elevation errors of the ATLAS data mentioned in our study are the elevation values retrieved from the ATLAS data minus the corresponding elevation values from the DTMs. Four statistical variables, the coefficient of determination (R2), mean (μ), standard deviation (σ), and root-mean-squared error (RMSE) values between the ground elevations and the corresponding DTM values were calculated to quantitatively evaluate the accuracy of the ground elevations retrieved from the ATLAS data.
3.3 Error factor analysis
Our research also explored the effect of the SNR, slope, vegetation height and vegetation cover on the elevation errors of the ATL03 data. The slope, vegetation height and vegetation cover products were generated from the DTM and CHM. The SNR values were calculated according to [31].
In this paper, random forest (RF) models were established to investigate the influence of the slope, SNR, vegetation height and vegetation cover on the elevation errors of the ATL03 data. The RF algorithm is suitable for analyzing complex nonlinear relationships and provides a parameter named the percent increase in the mean squared error (%IncMSE), to determine the importance levels of the error factors. The larger the %IncMSE value is, the more important the variable. Through the %IncMSE values, we can find the key factor affecting the ground elevation errors, thus providing guidance for the selection and application of ATL03 data.
4. Results and discussion
4.1 Verification by LiDAR data
Figure 2 shows the initial ATL03 data collected during the daytime over forest areas in Alaska and the results of the noise removal and ground photon identification algorithms. Most noise photons were removed by the noise removal algorithm, and ground photons were successfully extracted by the ground photon identification algorithm, which indicates that the existing noise removal and ground photon identification algorithms for the simulated ATLAS data are appropriate for the ATLAS data.
Fig. 2. (a) Raw ATL03 data over forest areas in Alaska; (b) Filtered ATL03 data over forest areas in Alaska; (c) Classified ground photons over forest areas in Alaska.
The quantitative results of the elevation errors in the ATL03 data are shown in Fig. 3. The large R2 values illustrate that the ground elevations retrieved from the ATL03 data are strongly consistent with the reference ground elevations. The mean difference and RMSE values of all ground photons are −0.61 m and 1.96 m, respectively. For forest, tundra and bare land areas, the mean difference values are −0.64 m, −0.62 m, and −0.59 m, and the RMSE values are 1.90 m, 2.06 m and 1.77 m, respectively. Both the mean difference and RMSE values for bare land are the smallest among the three land types, which indicates that ground objects such as vegetation have a negative impact on the accuracy of the ground elevation retrieved from the ATL03 data. In addition, the mean difference values for all land types are negative. The reason may be that there is multiple scattering in the surfaces, which leads to a widening of the surface return, so the ground photons identified by our algorithms are located somewhere below the land surface instead of being located at the land surface. Further research is needed to decrease the influence of multiple scattering in the ATLAS data and to select the true ground photons effectively.
Fig. 3. The scatterplots of the ATLAS-derived ground elevations versus the airborne LiDAR-derived ground elevations over (a) all the data, (c) forests, (e) tundra, and (g) bare land, and the frequency histograms of the elevation differences between the ATLAS-derived ground elevations and the airborne LiDAR-derived ground elevations over (b) all the data (d) forests, (f) tundra, and (h) bare land.
4.2 Error factor analysis
The importance levels of the slope, SNR, vegetation height and vegetation cover to the elevation errors are demonstrated in Fig. 4. Only the slope and SNR are considered in bare land areas. Overall, slope is the most important factor affecting the elevation accuracy of the ATL03 data, followed by the SNR, regardless of the land cover type. For all the data, the forest and tundra data, vegetation height and vegetation cover also affect the accuracy of the ground elevations extracted from the ATL03 data. However, the effects of the vegetation height and vegetation cover are relatively small compared to those of the slope and SNR.
Fig. 4. Importance levels of the error factors (measured as the %IncMSE) for all the data, the forest, tundra and bare land data. VC means the vegetation cover, and VH means the vegetation height.
4.2.1 Slope
To further explore the influence of slope on the elevation errors, we divided the slope values into 5 groups. The quantitative results of the elevation errors classified by slope are summarized in Table 1. For all the data, the mean difference and RMSE values for the elevation errors are −0.31 m and 1.62 m, respectively, when the slope is less than 5. The absolute mean difference and RMSE values increase dramatically with an increase in the terrain slope, especially when the slope is greater than 20°. For forest, tundra and bare land areas, the mean difference and RMSE values for the elevation errors are consistent with the overall data as the slope increases. In addition, the absolute mean difference and RMSE values of forest and tundra areas are larger than those of bare land areas at all slope levels. A possible reason is that the number of ground photons in the areas covered by vegetation is less than that in bare land, resulting in large elevation errors. Figure 5 shows the boxplots of the elevation errors between the ATLAS-derived ground elevations and the airborne LiDAR-derived ground elevations. The distribution of the elevation errors tends to be scattered with the increasing slope for all land types.
Fig. 5. Boxplots of the ATLAS-derived ground elevation minus the airborne LiDAR-derived ground elevation. Outliers are not shown in the boxplots.
Table 1. Statistical results for the elevation errors in the ATL03 data, classified by slope.
The negative effect of slope on the elevation errors may be caused by the geolocation error in the ATL03 data. The geographic accuracy of the ICESat-2 mission requirement is 6.5 m, and Neuenschwander et al.[28] verified that the geolocation error in the ICESat-2 data was −5 m in the along-track direction over Finland. In flat areas, the ground elevation values vary little, so the mean difference and RMSE values for elevation errors caused by the geolocation error are small. However, ground elevation values change greatly in steep terrains, so the absolute mean difference and RMSE values increase obviously with increasing slope. Therefore, future studies should focus on the verification and correction of the geolocation error in ATL03 data to eliminate the effect of slope.
4.2.2 SNR
In our study, the SNR values were divided into 6 groups. As shown in Table 2, the absolute mean values and RMSE values for the ground elevation errors decrease with increasing SNR values for all the data. However, the RMSE values, which are close to the overall RMSE value, vary little when the SNR values are greater than 10, as do the absolute mean values. As the SNR increases, the decreasing trend in the elevation errors for the three land types is consistent with the overall data. This result can be explained as follows: (1) when the SNR is small, the signal photons are difficult to separate from noise photons, resulting in high elevation errors. (2) When the SNR is greater than 10, most signal photons can be separated from noise photons; only a few near-ground photons are misclassified as ground photons, so the elevation errors vary little.
Table 2. Statistical results for the elevation errors in the ATL03 data, classified by SNR values.
4.2.3 Vegetation height
As shown in Table 3, the vegetation height values were divided into 5 groups with 3 m intervals for forest areas and 0.5 m intervals for tundra areas, as the vegetation heights in tundra areas are relatively low. The absolute mean difference and RMSE values increase with increasing vegetation height levels for forest and tundra areas. The reason may be that when the vegetation height increases, the probability of a photon passing through the vegetation and hitting the ground surface decreases, resulting in large elevation errors. However, our study only explored two vegetation types (forest and tundra), and the vegetation heights of forest areas are relatively low; only 7.3% of vegetation heights are larger than 12 m. Future studies with multiple vegetation types and vegetation height levels are needed to analyze exhaustively the effect of vegetation heights on elevation errors.
Table 3. Statistical results for the elevation errors in the ATL03 data, classified by vegetation heights.
4.2.4 Vegetation cover
Vegetation cover values were divided into 5 groups with intervals of 0.2. The statistical results for the elevation errors are summarized in Table 4. The absolute mean difference and RMSE values increase gradually with increasing vegetation cover for all the data and the forest and tundra areas (Fig. 6). The possible reason is that photons have difficulty penetrating the vegetation and reaching the ground surfaces when the vegetation cover is extensive, resulting in a small number of ground photons and subsequent large elevation errors. From Table 4, we can also see that the RMSE values for tundra areas are higher than those for forest areas at all vegetation cover levels. This result may be caused by the low vegetation in the tundra areas. Low vegetation photons and ground photons in tundra areas are difficult to separate, as these photons are mixed together. To reduce the negative impact of vegetation cover, further research is needed to explore how to distinguish low vegetation photons and ground photons effectively.
Table 4. Statistical results for the elevation errors in the ATL03 data, classified by vegetation cover level.
5. Conclusion
In this paper, we quantitatively assessed the performance of existing noise removal and ground photon identification algorithms on ICESat-2 data, validated the accuracy of ground elevations retrieved from ICESat-2 data and explored the influence of various factors on the ground elevation errors over forest, tundra and bare land areas in interior Alaska. These results led to the following conclusions: (1) The existing noise removal and ground photon identification algorithms for simulated ATLAS data proposed by Zhu et al. [27] are suitable for ATLAS data. (2) The overall mean difference and RMSE values between the ground elevations retrieved from the ATL03 data and the corresponding airborne LiDAR DTM data are −0.61 m and 1.96 m, respectively, highlighting the ability of the ATL03 data to retrieve ground elevations. (3) Among the error factors of slope, SNR, vegetation height and vegetation cover, slope is the most important factor affecting the elevation accuracy of the ATL03 data, followed by the SNR. (4) The elevation errors increase rapidly with increasing slope, especially when the slope values are greater than 20°. The elevation errors decrease with increasing SNR, while the decrease varies little when the SNR is greater than 10. In forest and tundra areas, the elevation errors also increase with increasing vegetation height and amount of vegetation cover.
Overall, this study provided a preliminary look at ATL03 data in ground elevation extraction and explored the influence of various factors on ground elevation errors, which will provide a basis for better estimating the ground elevation from ATL03 data. However, there are some limitations in this study. First, these preliminary conclusions do not involve comprehensive calibration. Forthcoming research will focus on further validation and analysis of ground elevations retrieved from ATLAS data for multiple study areas. Second, our study have analyzed the influence of various factors in ground elevation errors, but we didn’t propose a correction model for ground elevation errors in this study. Our next plan will also focus on the development of correction models to retrieve more accurate ground elevation.
Funding
National Natural Science Foundation of China (41901289, 41671434); Youth Innovation Promotion Association of the Chinese Academy of Sciences (2019130); the Open Fund Project of HIST Hengyang Base.
Acknowledgments
We thank the National Snow & Ice Data Center for providing the ATLAS data and NASA's Goddard Space Flight Center for providing the airborne LiDAR data used in this study. We also thank the editor and anonymous reviewers for reviewing our paper.
Disclosures
The authors declare no conflicts of interest.
References
1. S. Li, Z. Zhang, Y. Ma, H. Zeng, P. Zhao, and W. Zhang, “Ranging performance models based on negative-binomial (NB) distribution for photon-counting lidars,” Opt. Express 27(12), A861–A877 (2019). [CrossRef]
2. T. Evans, “Optical Development System life cycle for the ICESat-2 ATLAS instrument,” in2014 IEEE Aerospace Conference (IEEE, 2014), pp. 1–12.
3. T. Evans, “Integration and alignment of ATLAS instrument engineering model components in Optical Development System Lab,” in Optical System Alignment, Tolerancing, and Verification VII (International Society for Optics and Photonics, 2013), paper 884408.
4. U. C. Herzfeld, B. W. Mcdonald, B. F. Wallin, T. A. Neumann, T. Markus, A. Brenner, and C. Field, “Algorithm for detection of ground and canopy cover in micropulse photon-counting lidar altimeter data in preparation for the ICESat-2 mission,” IEEE Trans. Geosci. Remote Sens. 52(4), 2109–2125 (2014). [CrossRef]
5. L. A. Magruder and K. M. Brunt, “Performance analysis of airborne photon-counting lidar data in preparation for the ICESat-2 mission,” IEEE Trans. Geosci. Remote Sens. 56(5), 2911–2918 (2018). [CrossRef]
6. Y. Ma, R. Liu, S. Li, W. Zhang, F. Yang, and D. Su, “Detecting the ocean surface from the raw data of the MABEL photon-counting lidar,” Opt. Express 26(19), 24752–24762 (2018). [CrossRef]
7. G. Moholdt, C. Nuth, J. O. Hagen, and J. Kohler, “Recent elevation changes of Svalbard glaciers derived from ICESat laser altimetry,” Remote Sens. Environ. 114(11), 2756–2767 (2010). [CrossRef]
8. R. Kwok, G. F. Cunningham, J. Hoffmann, and T. Markus, “Testing the ice-water discrimination and freeboard retrieval algorithms for the ICESat-2 mission,” Remote Sens. Environ. 183, 13–25 (2016). [CrossRef]
9. L. L. Narine, S. C. Popescu, A. Neuenschwander, T. Zhou, S. Srinivasan, and K. Harbeck, “Estimating aboveground biomass and forest canopy cover with simulated ICESat-2 data,” Remote Sens. Environ. 224, 1–11 (2019). [CrossRef]
10. L. I. Duncanson, K. O. Niemann, and M. A. Wulder, “Integration of GLAS and Landsat TM data for aboveground biomass estimation,” Can. J. Remote Sens. 36(2), 129–141 (2010). [CrossRef]
11. S. Luo, C. Wang, G. Li, and X. Xi, “Retrieving leaf area index using ICESat/GLAS full-waveform data,” Remote Sens. Lett. 4(8), 745–753 (2013). [CrossRef]
12. S. Nie, C. Wang, X. Xi, S. Luo, G. Li, and F. Cheng, “A continuous wavelet transform based method for ground elevation estimation over mountainous vegetated areas using satellite laser altimetry,” IEEE J. Sel. Top. Appl. Earth Observ. 11(8), 2945–2956 (2018). [CrossRef]
13. L. Yue, H. Shen, L. Zhang, X. Zheng, Z. Fan, and Q. Yuan, “High-quality seamless DEM generation blending SRTM-1, ASTER GDEM v2 and ICESat/GLAS observations,” ISPRS J. Photogramm. Remote Sens. 123, 20–34 (2017). [CrossRef]
14. Y. Su and Q. Guo, “A practical method for SRTM DEM correction over vegetated mountain areas,” ISPRS J. Photogramm. Remote Sens. 87, 216–228 (2014). [CrossRef]
15. P. J. DeMott, A. J. Prenni, X. Liu, S. M. Kreidenweis, M. D. Petters, C. H. Twohy, M. S. Richardson, T. Eidhammer, and D. C. Rogers, “Predicting global atmospheric ice nuclei distributions and their impacts on climate,” Proc. Natl. Acad. Sci. U. S. A. 107(25), 11217–11222 (2010). [CrossRef]
16. C. Mallet and F. Bretar, “Full-waveform topographic lidar: State-of-the-art,” ISPRS J. Photogramm. Remote Sens. 64(1), 1–16 (2009). [CrossRef]
17. B. E. Schutz, H. J. Zwally, C. A. Shuman, D. Hancock, and J. P. DiMarzio, “Overview of the ICESat mission,” Geophys. Res. Lett. 32(21), L21S01 (2005). [CrossRef]
18. E. Naesset, “Predicting forest stand characteristics with airborne scanning laser using a practical two-stage procedure and field data,” Remote Sens. Environ. 80(1), 88–99 (2002). [CrossRef]
19. A. Neuenschwander and L. Magruder, “The potential impact of vertical sampling uncertainty on ICESat-2/ATLAS terrain and canopy height retrievals for multiple ecosystems,” Remote Sens. 8(12), 1039 (2016). [CrossRef]
20. L. A. Magruder, M. E. Wharton III, K. D. Stout, and A. L. Neuenschwander, “Noise filtering techniques for photon-counting ladar data,” in Laser Radar Technology and Applications XVII (International Society for Optics and Photonics, 2012), paper 83790Q.
21. M. S. Moussavi, W. Abdalati, T. Scambos, and A. Neuenschwander, “Applicability of an automatic surface detection approach to micro-pulse photon-counting lidar altimetry data: implications for canopy height retrieval from future ICESat-2 data,” Int. J. Remote Sens. 35(13), 5263–5279 (2014). [CrossRef]
22. D. Gwenzi, M. A. Lefsky, V. P. Suchdeo, and D. J. Harding, “Prospects of the ICESat-2 laser altimetry mission for savanna ecosystem structural studies based on airborne simulation data,” ISPRS J. Photogramm. Remote Sens. 118, 68–82 (2016). [CrossRef]
23. S. C. Popescu, T. Zhou, R. Nelson, A. Neuenschwander, R. Sheridan, L. Narine, and K. M. Walsh, “Photon counting LiDAR: An adaptive ground and canopy height retrieval algorithm for ICESat-2 data,” Remote Sens. Environ. 208, 154–170 (2018). [CrossRef]
24. X. Wang, Z. Pan, and C. Glennie, “A novel noise filtering model for photon-counting laser altimeter data,” IEEE Geosci. Remote Sens. Lett. 13(7), 947–951 (2016). [CrossRef]
25. J. Zhang, J. Kerekes, B. Csatho, T. Schenk, and R. Wheelwright, “A clustering approach for detection of ground in micropulse photon-counting lidar altimeter data,” in2014 IEEE Geoscience and Remote Sensing Symposium (IEEE, 2014), pp. 177–180.
26. S. Nie, C. Wang, X. Xi, S. Luo, G. Li, J. Tian, and H. Wang, “Estimating the vegetation canopy height using micro-pulse photon-counting LiDAR data,” Opt. Express 26(10), A520–A540 (2018). [CrossRef]
27. X. Zhu, S. Nie, C. Wang, X. Xi, and Z. Hu, “A ground elevation and vegetation height retrieval algorithm using micro-pulse photon-counting lidar data,” Remote Sens. 10(12), 1962 (2018). [CrossRef]
28. A. L. Neuenschwander and L. A. Magruder, “Canopy and terrain height retrievals with ICESat-2: A first look,” Remote Sens. 11(14), 1721 (2019). [CrossRef]
29. K. J. Bhang, F. W. Schwartz, and A. Braun, “Verification of the vertical error in C-band SRTM DEM using ICESat and Landsat-7, Otter Tail County, MN,” IEEE Trans. Geosci. Remote Sens 45(1), 36–44 (2007). [CrossRef]
30. J. Chen, X. Cao, S. Peng, and H. Ren, “Analysis and applications of GlobeLand30: A review,” ISPRS Int. J. Geo-inf. 6(8), 230 (2017). [CrossRef]
31. T. A. Neumann, A. Brenner, D. Hancock, J. Robbins, J. Saba, and K. Harbeck, “Ice, Cloud, and Land Elevation Satellite -2 (ICESat-2) Project Algorithm Theoretical Basis Document (ATBD) for global geolocated photons ATL03,” Available online: https://icesat-2.gsfc.nasa.gov/science/data-products (accessed on 25 August 2018).
32. T. A. Neumann, A. Brenner, D. Hancock, J. Robbins, S. B. Luthcke, K. Harbeck, J. Lee, A. Gibbons, J. Saba, and K. Brunt, “ATLAS/ICESat-2 L2A global geolocated photon data, Version 1,” 2019. [Indicate subset used]. Boulder, Colorado USA. NSIDC: National Snow and Ice Data Center. doi: https://doi.org/10.5067/ATLAS/ATL03.001. [Date Accessed].
33. B. D. Cook, J. McCorkel, and E. M. Middleton, “Data products of NASA Goddard’s LiDAR, hyperspectral, and thermal airborne imager (G-LiHT),” in Next-Generation Spectroscopic Technologies VIII (International Society for Optics and Photonics, 2015), pp. 94821D.
34. B. Cook, L. Corp, R. Nelson, E. Middleton, D. Morton, J. McCorkel, J. Masek, K. Ranson, V. Ly, and P. Montesano, “NASA Goddard’s LiDAR, Hyperspectral and Thermal (G-LiHT) airborne imager,” Remote Sens. 5(8), 4045–4066 (2013). [CrossRef]
