The Global CTX Mosaic of Mars
The Bruce Murray Laboratory for Planetary Visualization has completed a 5.7 terapixel mosaic of the surface of Mars rendered at 5.0 m/px. Each pixel in the mosaic is about the size of a typical parking space, providing unprecedented resolution of the martian surface at the global scale.

The mosaic covers 99.5% of Mars from 88°S to 88°N. The pixels that make up the mosaic can all be mapped back to their source data, providing full traceability for the entire mosaic. The mosaic is available to stream over the internet and to download, as described below.

All data in the mosaic come from the Context Camera (CTX) onboard the Mars Reconnaissance Orbiter (MRO).

Below is the entire mosaic within a 3D viewer. Click "See the Mosaic in 3D," or click here to see it in a new window.



Introduction
The V01 release of the Global CTX Mosaic is comprised of 86,571 separate images acquired between 2006 and 2020. Overlapping images were registered to each other (semi-controlled) before blending. We provide details of the full data processing pipeline below.

We are currently preparing a manuscript for peer-review. Until then, the most recent description of this product is from the 2023 Lunar and Planetary Science Conference.

We emphasize transparency both in how the mosaic was generated and for users to understand where data in the mosaic come from. We have developed a Python-based pipeline that incorporates non-destructive image processing techniques that preserve all information about the original data that comprise the mosaic and map all seams. This connects the mosaic directly with its original data, ensures that blending artifacts are not mistaken for landforms and geologic contacts, and provides instant access to the raw data that comprise the mosaic.

This version of the mosaic is the successor to the Beta01 version, which is still available here.

When using these data in a publication, presentation or poster, please acknowledge the hard work of the scientists and engineers at Malin Space Science Systems and the Jet Propulsion Laboratory. For scientific and technical overviews of the CTX instrument, please see Malin et al. (2007) and Bell et al. (2013).
Description of the V01 Release

The entire global CTX mosaic. Gaps in coverage (0.5% of the planet) are shown as white strips.

The Global CTX Mosaic was constructed using all CTX data through MRO release 49, CTX mission phase K11 (December 1, 2018). Sporadic images subsequent to release 49 (up to mission phase N08, July 2020) were used to fill unusually large gaps in the mosaic. The mosaic is comprised of 86,571 separate CTX images out of 104,647 available. Orbits not included were either (1) low signal/noise (increased atmospheric opacity), or (2) redundant coverage (stereo targets, change detection, etc.). The mosaic covers 99.5% of Mars.

The mosaic spans from 88°S to 88°N. All data are resampled to 5.0 m/px in an equirectangular projection. The planet was divided into 3,960 4°x4° tiles and blended independently, then blended together to complete the mosaic. As we describe below, every pixel can be traced to its host orbit through vectorized seam-maps that are provided with every tile. These seam-maps provide pixel-for-pixel spatial documentation of image seams to prevent misinterpretation of seams as possible landforms and to provide instant access to original data. All images are registered to each other, with tie points and residuals included as point shape files with all tiles. Data were also registered to controlled basemaps and these data will be released in the future. Absolute errors are all under 200 meters and were greater at higher latitudes than at the equator.
Where the Data Come From
The Global CTX Mosaic was constructed in four main stages:

1. Image-to-Image registration.
2. Tile-to-tile registration.
3. 4°x4° tile generation.
4. Tile blending.

1. Image-to-Image registration.

Within 64° of the equator, we divided Mars into 4°x4° tiles, then included a 0.25° buffer on all sides to create overlap with neighboring tiles (for steps 3 and 4 below). All images within a tile were projected and placed using SPICE data from the MRO spacecraft and CTX camera. Python code using area-based matching (Arcpy’s Register Raster tool) was used to attempt to align each image to all of its overlapping images. Once a successful match was made (at least three tie points with sub-pixel residuals), the image’s shifted position was recorded and then the image was used as a target for successive overlapping data. This process iterated until the entire tile was covered or no more successful matches could be made. If the tile was not complete, manual registration (two tie points distributed as far away from each other as possible) was performed on images that could not be automatically matched.

This process was performed on high-latitude and polar regions, as well, but tiles were grouped together to increase the physical area available. From 64° to 84° in each hemisphere, images were grouped in bins of 20° of latitude and 40° of longitude. From 84° to 88° degrees in each hemisphere, images were grouped in bins of 4° of latitude and 88° or 92° of longitude (creating 4 groups at each pole).

2. Tile-to-tile registration.

Using updated image placement derived from the co-registration process (step 1), all images within a tile were projected and merged into one layer to be used for tile-to-tile registration.

Alternating tiles were manually registered to existing basemaps. Within 60° of the equator, the THEMIS IR controlled mosaic was used as a target, while MOLA gridded shaded relief was used at higher latitudes. For each tile, 8 well-distributed tie-points were generated and a first order (affine) adjustment was applied. All remaining tiles were then automatically registered to their four neighboring tiles that had been registered to a basemap. A first order (affine) adjustment was applied based upon the automatically generated control network of tile-to-tile registrations. Thus, each tile had a control network that could be applied to individual orbits (step 3, below).

3. 4°x4° tile generation

The following video provides a brief overview of how CTX data go from raw to mosaicked within a 4.0°x4.0° tile.



All images that were successfully co-registered in step 1 were processed through the following pipeline:

- Ingestion into ISIS.
- Collection of SPICE data.
- Radiometric calibration.
- Column-based normalization.
- Projection.
- 8σ linear stretch (chosen to keep histogram balanced for tone-matching).
- Linear shift to co-registered location (from step 1).
- Non-linear histogram curve adjustment.
- Affine adjustment (using control network generated in step 2).
- Projection onto a 4.5°x4.5° canvas (4.0°x4.0° tile with 0.25° of buffer).

At this stage, all images within a tile must have image seams removed. The CTX mosaic has served as an experiment at implementing "non-destructive" image processing into a traditional image processing pipeline. After testing an array of blending options, by far the most efficient and high-quality platform was Adobe Photoshop, which is built to preserve information for rapid-iteration during workflows.

Non-destructive blending is achieved by determining the path of least contrast between overlapping orbits (a seamline). For the CTX mosaic, this typically occurs in featureless plains where disparities in imaging and illumination conditions are subtle. Thus, features of more geologic interest are more likely to be captured within one orbit, reducing artifacts. Once seamlines are generated, masks are applied to shield data that are not used in the final composition. The inverse of each orbit's mask represents the portion that is retained and is used to generate the vectorized seam map (see below). After masks are applied, contrast is adjusted radially away from the seamline (“feathering”) so that the images match but are adjusted less and less away from the seamline.



Masks that shield data for each orbit during the blending process (above) are exported as rasters with values of 0 where data are used in the mosaic and 255 where data are not used. Each raster can then be vectorized by using gdal_polygonize.py. This vectorizes the entire tile and attaches the DN value (0 or 255) as an attribute to each feature. This field is used to remove all DN=255 features, which represent regions not used by that specific orbit. If an orbit was not used at all during the non-destructive blending stage, it is removed entirely at this step. Features that do survive this stage have their DN attribute removed and a PRODUCT_ID field added, with the orbit number assigned in this field.

Once all features have been vectorized, they are appended to a centralized shapefile for that tile. This produces one shapefile with all features with their PRODUCT_ID field assigned. This field is used to apply a join to the original CUMINDEX table from the PDS, which appends all spacecraft and imaging conditions metadata to each feature. Finally, we add attribute fields called SESE_LINK and PDS_IMG to provide links to ASU’s data page for each orbit, and the raw PDS image file.

4. Tile blending.

3,960 4.0°x4.0° tiles must be blended together to create the fully blended global CTX mosaic. To do this, tiles were binned into 3x3 grids (12°x12°) and blended together, except for data from 80°N to 88°N, which were binned into 2x3 grids (8°x12°). This blending process typically leads to a tone shift that inhibits the subsequent blending of each 3x3 grid with its neighbors. Therefore, after blending, a uniform gamma shift was applied to all 9 tiles within a 3x3 grid to set the mean value across the grid to 128, tone-balancing all grids.

These 3x3 grids must then be blended with their neighbors, but this is challenging since if grid A is blended with its neighbor to the east (grid B), then blended with its neighbor to the west (grid C), adjustments made during the A-C blending will undo the successful blending of grids A and B. We developed a layer-and-gradient approach that solved this problem while retaining all traceability of the mosaic.

As shown in the example below, we blended the three easternmost tiles of grid A with the three westernmost tiles of grid B (1). We preserved the unblended versions of these 6 tiles as a layer beneath the blended version (2). We then applied a mask that only allows portions of the blended layer to appear and allowed the unblended layer to show through elsewhere. The blended version is resolved at 100% opacity along the seam between the two grids, then gradually fades to 0% opacity in either direction towards the center of each grid (3). Thus the final composition grades from the fully blended grids along the seams to the pre-blended grids away from the seams.



This process is repeated across all vertical and horizontal seams between 3x3 grids across the entire planet.

For all blending procedures, masks are preserved and converted into vectorized polygons. These are then used to clip the individual orbit features generated in step 3 above, such that pixel heritage is maintained across tile boundaries. This does mean that images originally blended in one tile (but within the 0.25° margin surrounding the 4.0°x4.0° tile) may be rendered in the neighboring tile in the final composition. This information is recorded in the "MOS_QUAD" attribute for all features.
Access the Data
The Global CTX Mosaic is available in a variety of ways, accounting for the range of applications that we anticipate. We describe these avenues below. If you are developing a platform that uses the mosaic, please contact us to have it listed in this section.

Esri Sceneview
Explore the mosaic in full resolution and draped over topography in a web browser. Use the embedded scene above or click here to open it in a new browser.

JMARS
The V01 release of the global CTX mosaic is available through JMARS. In the "Main" listing of layers, click the plus button and search for "CTX global v01". Alternatively, browse by instrument, CTX, then choose "Global Mosaic" and choose the V01 layer.

ArcGIS Pro
In the Catalog pane in ArcGIS Pro, select "Portal" and search for "Mars CTX V01" and add the layer to your map.

ArcMap 10.x
Choose the menu "File", then "Add Data", then "Add Data From ArcGIS Online." Search for "Mars CTX V01" and click "Add" to place the mosaic in your map. Users have noticed that some segments of the mosaic do not render at full resolution in ArcMap when using the streaming version. We encourage you to switch to ArcGIS Pro, where this does not happen, or download the data that you need (below).

Mars Quickmap
ACT has added the V01 version of the CTX mosaic to the Mars Quickmap interface. Scroll down the list of layers to "CTX mosaic" to activate it.

Google Mars
Download this KML file and load it into Google Earth Pro. We broke the entire mosaic up into 198 separate tiles to make it work on Google Mars, which is why seams do still show up. The seams become smaller and smaller as you zoom in.

GIS Tile Maps
A map of the 3,960 4.0°x4.0° tiles that comprise the full mosaic can be found in shapefile (for GIS applications) and KML (for Google Mars) formats. These coverage maps provide links to the data for each tile, as well as thumbnails and data sheets.

Download the Data
For scientific use, we recommend downloading the invdividual tiles for your study site, or downloading the entire mosaic. The entire uncompressed mosaic can be found here.

The mosaic images are stored in zip files that reduce the download size. The entire mosaic (compressed) is 5.6 TB, including seam maps but not including overviews for the GeoTiff files. Unzipped and uncompressed, the entire mosaic is 11.484 TB, including overviews. Assuming a 5 mb/s download speed, the entire mosaic will take 13 days to download. All tiffs are uncompressed and all vector files (seam maps and tie points) are provided as shapefiles. We encourage you to download the entire mosaic if you want but ask that you only download one tile at a time to preserve bandwidth for other users.

Tiles are named based upon the location of the lower-left pixel. Therefore, E040_N16 covers 40°-44° east longitude and 16°-20° north latitude.

Each zip file contains the following files:
- MurrayLab_CTX_V01_${tile}_Mosaic.tif (Uncompressed GeoTiff of the mosaic for this tile).
- MurrayLab_CTX_V01_${tile}_SeamMap.shp (Polygon map of image seams and source data).
- MurrayLab_CTX_V01_${tile}_TiePoints.shp (Point map of image-to-image registration points).
- MurrayLab_CTX_V01_${tile}_DataSheet.pdf (Rendered version of the mosaic to show how data should appear).
- MurrayLab_CTX_V01_${tile}_ReadMe.txt (Text file with detailed information about all contents).

If you are streaming the image mosaic so only need to download seam maps, they can be found here. Likewise, maps of image-to-image registration points can be found here.
Engaging with the Mosaic
If you download individual tiles of the mosaic, there are some techniques that help to make the data easier to use. Below, we describe some of these methods.

VRT Files
All 4°x4° tiles in the mosaic are seamlessly blended with their neighboring tiles. If you load multiple individual tiles into GIS software, however, the tiles will be treated separately and will not behave as one seamless image. An efficient solution to this for a group of tiles is to create a "virtual" file (.vrt). This is a text file that references the individual tiles and is interpreted by GIS software as one layer.

VRT files are easy to make and instructions can be found here. In short:

To build a virtual (.vrt) file that links all GeoTiffs within a directory:

     gdalbuildvrt output.vrt *.tif

To build a virtual file of all GeoTiffs within all subdirectories:

     gdalbuildvrt output.vrt */*.tif

In QGIS, a virtual raster can be built with a GUI by choosing Raster -> Miscellaneous -> Build Virtual Raster (Catalog).

Mosaic Dataset
Depending on your hardware resources, VRT files may not scale up to the whole planet if you want to access the entire mosaic locally within GIS software. For this, one option is to generate a Mosaic Dataset within ArcGIS. This does scale up very well, but cannot be used in QGIS or other GIS software.

Building a Mosaic Dataset requires three steps:

1. Create the Mosaic Dataset within a Geodatabase.
2. Add rasters to that Mosaic Dataset.
3. Build pyramids for the Mosaic Dataset. (Overviews can be built in step 2 abut we have found it more reliable to build them in a separate command)

This file will behave as one image within an ArcGIS Pro map.

Interacting with Seam Maps

Demo of accessing a processed orbit from ASU through the CTX Mosaic seam maps (ArcGIS Pro).

Every pixel of the Global CTX Mosaic can be traced back to its original orbit instantaneously by using our vectorized polygon seam maps. This is important for validating observations within the mosaic and accessing either raw or less-processed versions of the data. Each polygon has been joined to all data provided for that specific orbit in the CTX CUMINDEX.TAB file provided via the PDS. Thus, all metadata for each pixel is contained in the attribute table for that tile's shapefile. In addition, we have provided a direct link to that orbit's page on ASU's Mars Image Explorer. This page contains a zoomable version of the orbit, and a pyramidized TIFF file that, once downloaded, will render quickly within your GIS project. This efficiently connects the mosaic with the data that were used to construct it.
Acknowledgements

When using these data, please acknowledge the scientists and engineers
who built CTX and have operated it for over a decade at
Malin Space Science Systems and the Jet Propulsion Laboratory.

The CTX Mosaic project has been led by Jay Dickson (Caltech).
Significant help has been provided by Bethany Ehlmann (Caltech/JPL), Laura Kerber (JPL), Caleb Fassett (APL), Daven Quinn (U. Wisconsin), and Trent Hare (USGS).

We are thankful for the effort of Lucian Plesea of ESRI to make this version of the mosaic available for streaming, and to Scott Dungan and Ken Ou for IT infrastructure expertise.

This project went through an extensive beta process within the planetary science community and substantially benefited from feedback we have received from the scientists listed below.
Jenny Whitten, Tulane Colin Dundas, USGS Christopher Edwards, NAU
Marisa Palucis, Dartmouth Lulu Pan, Univ. Lyon Debra Needham, MSFC
Ellen Leask, Caltech Erica Jawin, Smithsonian Mark Salvatore, NAU
Jim Head, Brown Vivian Sun, JPL Mike Bramble, Brown
Brad Thomson, UTK Nathan Williams, JPL Keenan Golder, UTK
Jack Mustard, Brown Adeene Denton, Brown Abby Fraeman, JPL
Ingrid Daubar, JPL Chris Kremer, Brown Serina Diniega, JPL
Becky Williams, PSI Ashley Palumbo, Brown Kaitlyn Stacey, UT Dallas
Tim Goudge, UT Austin Edwin Kite, U. Chicago Jacob Widmer, UMD
Devon Burr, UTK Claire Mondro, UTK Chris Yen, Brown
Jim Skinner, USGS Susan Conway, U. Nantes The JMARS team, ASU

We are grateful for funding for this project from Foster & Coco Stanback for the Beta01 release and to the NASA PDART program for the V01 release.


jdickson@caltech.edu


All data used in the construction of the CTX global mosaic have been publicly released and are freely available via the NASA Planetary Data System. The CTX camera was built by Malin Space Science Systems, the Mars Reconnaissance Orbiter was built by the Jet Propulsion Laboratory/NASA. The Murray Lab/Caltech grants free use of the V01 version of the mosaic for all purposes. Data credit should be to: NASA/JPL/MSSS/The Murray Lab.