Solar System Treks Mosaic Pipeline (SSTMP)

Contents

1. Description
2. Requirements
3. Setup
4. Creating a mosaic

Description

SSTMP creates high-resolution lunar DEMs and orthoimage mosaics of Earth's moon for requested areas. Given a bounding box, it will select, fetch, ingest, stereo reconstruct, and merge raw imagery from the Lunar Reconnaissance Orbiter's (LRO) Narrow Angle Camera (NAC) to make your mosaic. It can be configured to use any number of real or virtual computers for processing. NAC Mosaic Pipeline uses laser altimeter data from LRO's Lunar Orbiter Laser Altimeter (LOLA) for alignment before merging the mosaics.

SSTMP includes:

• An Argo workflow that supervises all of the processing for a stereo (nac-mos-stereo.yaml) or mono (nac-mos-mono.yaml) mosaic.
• A Dockerfile defining a container which can do any of the processing steps (Dockerfile)
• A python package with scripts needed for processing (nacpl/)
• A conda environment needed for processing (nacpl_env.yml)
• A Skaffold configuration for setting everything up (skaffold.yaml)

Development goals of the project are:

• In the near term, SSTMP should:
  • Be easy to set up and run on any computer or cluster
  • Produce a reasonable mosaic from no user input other than requested bounding box, if sufficient data exists
  • Be flexible for cases where defaults don't result in good output
• In the medium term:
  • NAC Mosaic Pipeline should provide rigorous error analyses

Requirements

• Any Kubernetes cluster. This can be minikube running on a single machine / node, a cloud service such as Amazon EKS, or your own custom cluster. k3s is recommended as a quick way to set up Kubernetes.
• A Skaffold installation that talks to your Kubernetes cluster
• Kustomize installed in your PATH. Unfortunately this needs to be kustomize 3.9.3 or an earlier version of kustomize 3, because of this regression in kustomize 4.
• A container image registry to which you have write access. You can set up a private registry using docker run -d -p 5000:5000 --restart=always --name sstmp-reg registry:2 and then specify it using --default-repo=localhost:5000 when the installation step using skaffold below

Internally, SSTMP uses a slew of free and open source programs, including:

• USGS ISIS
• Ames stereo pipeline (ASP)
• Geopandas
• GDAL
• Orfeo toolbox

Setup

Once you have the requirements,

1. Clone this repository
2. Change to the directory containing skaffold.yaml : cd src
3. Configure your storage settings. Open volumes-example.yaml in an editor, follow instructions in the comments, and save it as volumes.yaml.
4. Configure your ingress settings and tls. Open volumes-example.yaml in an editor, follow instructions in the comments, and save it as volumes.yaml.The included ingress-example.yaml assumes a Traefik 2 ingress controller, which comes bundled with k3s. If you are using another ingress controller such as nginx, you will need to edit this file.
5. Install SSTMP: run skaffold run. If you're using a private registry, specify that with --default-repo, for example skaffold dev --default-repo=localhost:5000. You may also be interested in trying skaffold dev or skaffold run --tail.

Creating a lunar mosaic

Using the SSTMP web interface

Please see this video:

Using the command line interface

Let's say you want a DEM and orthoimage mosaic of a bounding box which goes from 25.8 degrees easting to 25 degrees easting, and from 45 degrees north to 45.8 degrees north.

After following setup above, Run argo submit NAC_pl_workflow.yaml -p east=25 -p west=25.8 -p south=45 -p north=45.8 .

You can monitor the progress of the mosaic creation using the map interface which will be running at http://yourcluster/moon, or the Argo interface at http://yourcluster/workflows, where yourcluster is the ip address or hostname of your cluster master node.

History

SSTMP grew out of work at JPL primarily by Marshall Trautman, Charles Nainan, Natalie Gallegos. The Trek Team maintained an internal repo called "TrekDataPrep" which was a precursor of SSTMP. Aaron Curtis joined the project in late 2019, re-imagined and re-wrote TrekDataPrep as a system of Argo Workflows, and obtained permission from JPL to open source the project. For file size and security reasons, repository history before 2020-05-06 was removed before the first release of SSTMP. Although 98 of the 400 commits in the repo were preserved, the commit hashes were changed.