NASA Logo, National Aeronautics and Space Administration
FROST Dataset Title superimposed on panorama of ice-like terrain

[Attribution]  [Acknowledgements]  [Support]  [Download Primary Dataset]  [Download Extras]


Summary

We present an analog dataset that provides examples of possible terrain features, geometry, and appearance at the 1-10cm scale on ocean worlds/icy moons such as Europa, Enceladus, and Pluto. The motivation for collecting this dataset was a lack of available high-resolution digital models suitable for development of surface missions to these bodies (circa 2019), including use for simulation of mechanics, sampling, and imaging. We took advantage of NASA field opportunities to Death Valley, California and the Atacama Desert, Chile in order to observe and record analog sites.

Data was collected at the Devil's Golf Course site in Death Valley and the Yungay site in the Atacama Desert. Areas of interest were preselected from satellite views with final terrain selections made in person. A high-resolution laser scanner (with built-in color camera) was used to survey each area with a multi-view scan pattern to minimize occlusions. The highest resolution centroid of each survey's point cloud was then cropped to create 5m x 5m digital elevation models (DEMs) of terrain at 2.5mm resolution. A total of 11 scenes were collected across 4 unique sites. Patches collected span a range of salt, soil, and rock feature distributions.


Three pictures of field experimentation. Left image is of a man giving thumbs up in death valley. Middle is laser scan of planetary terrain. Right is closeup of terrain detail.

Keywords: Dataset, ocean world, icy moon, 3D, LIDAR, DEM, point cloud, terrain, lander, Europa, simulation, robotics

Changelog:

Dec 11, 2019 - Come see our poster/abstract at AGU!
Dec 01, 2019 - Initial Release
 

Attribution

The FROST Dataset was produced through the work of the OceanWATERS simulation team in the Intelligent Systems Division at the NASA Ames Research Center.

Citation:

Uland Wong, Orkan Umurhan, Antoine Tardy, Terry Welsh, Mark Allan, Larry Edwards. FROST Dataset: Features Relevant to Ocean Worlds Surface Terrain. NASA Ames Research Center, Dec 2019.

Acknowledgements

This work was funded by the Planetary Exploration Science and Technology Office in the NASA Science Mission Directorate as a part of the development of the OceanWATERS simulator.

We would also like to thank Thomas Stucky and the rest of the OW team for advice and support.

Support and Documentation

Download the complete dataset documentation [.pdf].

Each terrain contains a DEM (TIFF format), a registered orthographic color image (PNG format), and the raw point cloud (Matlab mat file). The DEMs and orthoimages are at 2048 x 2048 pixel resolution covering a 5m x 5m area, giving a spatial resolution of 2.5mm between postings. This resolution was chosen because it is a convenient value that is similar to the LIDAR's intrinsic range error (3mm). We provide the point clouds due to the known limitations of orthographic projection in creating a 2.5D DEM from a 3D point cloud, such as artifacts from overhangs and occlusions. You can use this point cloud to create your own 3D voxel or mesh representation, for example. 

Preview of datatypes in dataset including digital elevation map, color orthoimage, and point cloud.

For questions about the dataset, contact: Dr. Uland Wong <uland dot wong at nasa dot gov> or Dr. Orkan Umurhan <orkan dot m dot umurhan at nasa dot gov>.


Primary Dataset

Download the entire dataset with one link [.zip, 17.8 GB], or download individual parts below.


Death Valley Location 1 - "Parking Lot"

Terrain D1A
 		Terrain D1B
Rendering of Terrain D1A
 Rendering of Terrain D1B
Description: Satellite view of this location indicates a surface supporting three different color textures. Within this broad marker we selected two regions, “Terrain D1A” and “Terrain D1B”, the former of which is representative of the grey colored terrain of the larger (10 km2) surrounding region. Terrain D1B was chosen due to its relatively darkened surface color compared to A. Note both locations are on the valley floor (-85m).

Download: 709 MB [.zip]


Death Valley Location 2 - "Salacious Crumb"

Terrain D2A
 		Terrain D2B
Rendering of Terrain D2A
 Rendering of Terrain D2B
Terrain D2C Terrain D2D
Rendering of Terrain D2C Rendering of Terrain D2D
Description: This region is characterized by flat, greyish-colored flat dry ponding zone with no relief, which we call “Salacious Planitia” (SP). The region was selected because the presence of the road (Westside Road, “WR”), located on its southern border, influences the manner in which freshly emplaced rain runoff ponds and, subsequently, the character of emergent salt efflorescence to develop on either side of WR. Four zones were selected surrounding SP maximizing variety of textural appearance from this satellite imaging. [Map data: Google, Maxar Technologies]

Download: 845 MB [.zip]


Death Valley Location 3 - "Badwater Basin"

Terrain D3A
 		Terrain D3B
Rendering of Terrain D3A
 Rendering of Terrain D3B
Description: The Badwater Basin region is located 15km south of locations 1 and 2 and is found at the end of Salt Road. The locations to study (A+B) were chosen based on analysis of satellite images. One relatively darkened zone (A) and one relatively light zone selected (B). Satellite imagery indicates B supports hexagonal structures.

Download: 458 MB [.zip]


Atacama Desert Location 1

Terrain Y1A
 		Terrain Y1B
Rendering of Terrain Y1A
 Rendering of Terrain Y1B
Terrain Y1C
Rendering of Terrain Y1C
Description: Data was collected in the Atacama desert south of Yungay. Three samples were taken constituting rough salt growth (A), smooth growth (B) and no growth(C).

Download: 501 MB [.zip]


Dataset Extras

Uncropped Scans of Terrain

Description: The raw point clouds from the laser scans cover a large area (>30m from each scan point), 5 or more scans from each location are registered to create a unified point cloud, of which a central 5m x 5m square is cropped. This download link provides the all of the uncropped scan data in an as-is manner. Because of the measurement density fall-off with range, a spatial resolution of 2.5mm is not guaranteed.  

Download: 15.3 GB [.zip]