Michael Kenneth Barker
Download SLDEM2015:
http://imbrium.mit.edu/DATA/SLDEM2015/
Global maps (for latitudes between 60 deg South and 60 deg North) can be downloaded from the following links:
FLOAT IMG format:
JPEG2000 format:
512, 256, and 128 pixels per degree
Alternatively, if you are interested in just a few specific regions, then you can download the maps split up into tiles at these links:
FLOAT IMG format:
JPEG2000 format:
Acknowledging this work:
If using this dataset, please cite the following paper:
Barker, M. K., Mazarico, E., Neumann, G. A., Zuber, M. T., Haruyama, J., Smith, D. E. "A new lunar digital elevation model from the Lunar Orbiter Laser Altimeter and SELENE Terrain Camera," Icarus, Volume 273, p. 346-355. http://dx.doi.org/10.1016/j.icarus.2015.07.039
Introduction
The Lunar Orbiter Laser Altimeter (LOLA)aboard the Lunar Reconnaissance Orbiter (LRO) has collected over 6.3 billionmeasurements of surface height with a vertical precision of ~10 cm and anaccuracy of ~1 m. The ability of LOLA to obtain measurements under uniformillumination conditions and in shadowed regions globally provides an advantageover passive stereoscopic imaging, particularly at high latitudes (poleward of60°) where imaging is hindered by low solar incidence angles. This has allowedLOLA to produce the highest resolution and most accurate polar terrain modelsto date. In addition, LOLA provides the necessary geodetic framework with whichlunar stereo imaging-based models must be controlled. However, due to LRO’spolar orbit, gaps in the LOLA coverage still persist especially near theequator where some can be as wide as a few km.
Here we present the results of aneffort to improve the LOLA coverage by incorporating topographic informationfrom the independently-derived and highly complementary SELENE Terrain Camera(TC) dataset (Haruyamaet al., 2012). This dataset, called SLDEM2013, was controlled with an olderversion of the LOLA data geolocated with a gravity field made prior to theGravity Recovery and Interior Laboratory mission (GRAIL; Zuberet al., 2013). In Barker et al. (2015), we co-registered the TC data to thenewer, more accurate GRAIL-based LOLA geodetic framework creating a mergedproduct, called SLDEM2015, which can be downloaded from the Planetary DataSystem LOLA data node. In addition to having many geophysical and explorationapplications, the SLDEM2015 will improve the orthorectification andco-registration of diverse lunar datasets to the latest LRO/LOLA/GRAIL geodeticsystem without the gaps normally present between LOLA groundtracks.
Data
We used the 43,200 1°x1° TC DEM tilescovering latitudes within ±60°. The LOLA data coverage is sufficiently densefor most purposes at latitudes outside this range. During the SELENE missionlifetime (November 2007 to June 2009), the TC acquired stereo imagery for over99% of the surface (Haruyamaet al., 2012). The effective horizontal resolution of the TC DEM dataset isabout 60 - 100 m.
Methods
We followed a two-step approachwhen co-registering the TC tiles to the LOLA data. In step (1), we derived a5-parameter coordinate transformation between every TC tile and the fullresolution LOLA data in that tile (unbinned point cloud with ~100,000points). Thus, the tile-averaged transformation parameters compensate predominantlyfor the Kaguya/TC orbital, pointing, and camera model errors. In step (2), wefit a 3-dimensional (3D) offset to each LOLA profile segment in the transformedTC tile. These offsets reflect primarily the LOLA geolocation errors, whicharise from uncertainties in the LRO orbit and LOLA boresight with a secondarycontribution from Kaguya/TC errors not completely removed by the transformationin step (1) due to the restricted number of degrees of freedom.
Results
The top panel in Fig. 1 shows the spatialdistribution of initial RMS vertical residuals between the LOLA and TC data whilethe bottom panel shows the RMS distribution after applying the tile-averagedtransformation in step (1). Most of the vertical residuals are reduced to 3-4mafter step (1). The fraction of residuals <5m increased from ~50%prior to registration, to ~90% after registration. After step (2), the 90thpercentile further decreases from ~5.0m to 3.4m while themedian RMS residual decreases from 3.2m to 2.6m.
Two areas with particularly high residualsinclude the South Pole Aitken basin (SPA; 45–60°S, 140–210°E) and the westernedge of Orientale basin (45°S–15°N, 230–260°E). The large errors in SPA are dueto lower-resolution Multi-band Imager DEMs included in the TC dataset to fillin areas TC did not observe. In Orientale, reaction wheel troubles on Kaguyaled to degraded orbit reconstruction on those observation dates. Thetile-averaged transformation also significantly improved some areas on the farside, especially between ± 30° latitude and 180–200°E longitude. These areashave initial RMS residuals of ~8 to 18m, and have less vertically-oriented shapesthan the regions mentioned previously. We attribute the initially highresiduals in these far side areas to differences in gravity field models usedby the reference LOLA data; SLDEM2013 was referenced to LOLA data based onpre-GRAIL gravity field LLGM-2 whereas in this work we used the much higherresolution and more accurate GRAIL GRGM900B (see Mazaricoet al. (2013) for discussion).
Fig. 2 compares 3 different DEMs of the region around thelanding site of the Chang’e 3 spacecraft: Fig. 2a shows LOLA data alone, aftercontinuous curvature interpolation between ground tracks. Fig. 2b is our newSLDEM2015 merged product. Fig. 2c shows the GLD100 DEM, which was produced fromLunar Reconnaissance Orbiter Camera (LROC) Wide Angle Camera (WAC) stereoimagery (Scholtenet al., 2012). The SLDEM2015 (Fig. 2b) reveals surface detail not sampledby the LOLA ground-tracks and below the ~1 km resolution limit of the GLD100.
Summary
In this study, we co-registered43,200 TC DEM tiles to the latest GRAIL-based LOLA geodetic framework toproduce a combined topographic map of the Moon at a resolution of 512ppd.The bulk of the co-registered TC tiles have vertical residual with the LOLAdata of 3 to 4m. The co-registered TC data were used to estimate andremove orbital and pointing errors (typically amounting to <10mhorizontally and <1m vertically) from the LOLA altimetric profiles. Bycombining both datasets, gaps in the LOLA data can be filled without the needfor interpolation. Given the high (~1m) absolute vertical accuracyof the LOLA data found in orbit overlap analysis (Mazaricoet al., 2013), we conclude that the typical vertical accuracy of the SLDEM2015is 3 to 4m. SLDEM2015 is available for download from the PDS LOLA data node.This product has many geophysical and cartographic applications in lunar science,as well as exploration and mission design. Studies requiring the high geodeticaccuracy of the LOLA data and the excellent spatial coverage of the TC datawill especially benefit from this merged data product.
References
Barker, M.K. et al., 2016. A new lunardigital elevation model from the lunar orbiter laser altimeter and SELENEterrain camera, Icarus, 273, pp. 346-355. http://dx.doi.org/10.1016/j.icarus.2015.07.039
Haruyama, J. etal., 2012. Lunar global digital terrain model dataset produced from SELENE(Kaguya) terrain camera stereo observations. In: Lunar and Planetary ScienceConference, p. 1200.
Mazarico, E. et al., 2013. Improvedprecision orbit determination of Lunar Orbiters from the GRAIL-derived lunargravity models. In: 23rd AAS/AIAA Space Flight Mechanics Meeting, pp. 13-274.
Scholten, F. et al., 2012. GLD100: The near-global lunar 100 m rasterDTM from LROC WAC stereo image data, J. Geophys. Res. (Planets), 117,CiteID E00H17. http://dx.doi.org/10.1029/2011JE003926
Zuber, M. T. et al., 2013. Gravity field of the Moon from the Gravity Recovery and Interior Laboratory (GRAIL) mission, Science, 339, pp. 668-671. http://dx.doi.org/10.1126/science.1231507
