Feb 17, 20215 min read

Digital Outcrop Modelling - The Crash Course I Received

Hello everyone,

I haven't got much to update you on in terms of PhD research. For past two weeks, I have mainly been working on manuscript drafts and trying to get two papers ready for publication. I have one almost ready to be re-submitted. I just need to remove 916 characters (spaces included) from the document, update my response to reviewers document, re-write my cover letter to the journal and then finally resubmit! The other paper, well. I'm still waiting on edits from a co-author, which is delaying me from sending the draft to other co-authors. It is a bit frustrating, and right now I am debating whether to send a reminder email to that co-author. I probably should but I always feel like I am being pushy everything I send a reminder email. Anyway, that has pretty much been it in terms of research. I plan to resume working on my introduction chapter this week and aim to have a first draft completed by the start of next week. By first draft, I do not mean a draft constructed well enough to send to my research committee. What I mean by first draft is the first milestone in preparing the introduction chapter for my research committee. If I stay on track, I will be able to send them a rough copy before the start of April.

Europlanet Society Geology and Planetary Winter Mapping School

What I wanted to show you in this post is what I was working on February 15th, during family day (yes, I know it is a holiday, but I barely got any work completed on Friday...). A couple of weeks ago, I was registered in the Europlanet Society Geology and Planetary Winter Mapping School, where European scientists organized workshops to provide basic and intermediate skill set lessons on how to map geologic features on planetary surfaces using spacecraft orbiter and rover imagery and data. Unfortunately, I could not attend any of the lectures or practical's live since the time slots were set for Central European Time (CET 6 hours ahead of me) and they started at 3:15 am local time. Instead, I watched the recorded lectures and practical's I had signed up for when I registered for the school. We had to complete the tutorials during the practical sessions and upload screenshots of our final products. There is no grade for the tutorial, you just get a certificate for participation and completion of the Planetary Mapping School.

The practical I signed up for was the Extraction of Geological Features and Creation of 3D Surfaces. I read the description of this practical and it said I would learn how to create digital outcrop models and 3D surfaces from Mars Reconnaissance Orbiter SHARAD data. I was immediately intrigued because (1) I love learning about radar, especially synthetic aperture radar, and (2) I have never worked with digital outcrop models directly before so I was excited to learn how you can create them using freely available software such as CloudCompare.

Now, going through the tutorial step by step is a bit challenging since it is very difficult to explain the programs and data through text. Interactive visuals and illustrations are the best methods for showing all of you what I learned and how I learned it. What I can explain in this post is the difference between digital outcrop models (DOMs) and digital elevation models (DEMs).

For anyone reading in the earth and planetary sciences, you will know that DEMs are used to study the elevation and topography of planetary surfaces. DEMs are produced using stereo-photogrammetry (Image 1), where 3D coordinate and other information (e.g., height and relief) points of an object are measured in two or more orthomosaic photographic images (Image 2) taken from different positions (Image 1). The X and Y coordinates in a DEM only have one Z value so therefore no vertical planes are produced during this processes. In planetary science, scientists and engineers use images acquired from low to high resolution cameras on spacecrafts to produce DEMs that will provide topographic and elevation data of planetary geologic features (e.g., lava domes, impact craters, river deltas, floodplains, etc). Now, when you have a DEM, you are not true 3D. You are closer to 2.5 D (if that exists at all!) since a DEM does not reproduce the volume of an area.

Image 1: Illustration of stereo-photogrammetry. Two orthomosaic images are collected at two different points during the orbit of a satellite. This provides a different perspective of the imaged site, allowing relief and height to be calculated from the stereo pair of images.

Image 2: Examples of orthomosaic images of the same site taken at different angles. The satellite views the site at a different incidence angle, exposing and covering more areas and observing objects at a different perspective. Credit: SlideShare, Kumar (2011).

A DOM, is a bit different to a DEM. Unlike a DEM, the X and Y coordinates in a DOM have multiple Z values assigned to them, allowing a reproduction of the volume of a site. DOMs are able to reconstruct the actual 3D morphology and shape of a geologic structure, but only if a sufficient number of images are taken of the structure at different angles. An advantage of a DOM over a DEM is you can get a better 3D perspective of the geologic structure in study. This is vital when studying geologic processes that require the analysis of structures in 3D (e.g., cross-bedding in ancient river channels on Mars). To generate a DOM, you can either use dense point cloud LiDAR data or photogrammetry. A LiDAR dataset will create a coloured point cloud while photogrammetry will create a coloured/textured mesh. As mentioned earlier, a DEM is produced when the relief of two orthomosaic images is calculated. For a DOM, two images is not sufficient to create a 3D model. A different photogrammetric technique is required, called Structure-from-Motion photogrammetry (Image 3); uses multiple photos to recreate the 3D structure and texture of an object, sometimes requiring 100s to 1000s of photos.

Image 3: Illustration of how images are collected and then stitched together using the Structure-from-Motion photogrammetry technique. Credit: M.A thesis, Riel (2016).

Below (Image 4) is an illustration of how DOM 3D mesh images are generated from numerous images using the Structure-from-Motion technique. You start with ensuring you have images that overlap and are taken at different angles of an object. The more images you have, the denser the point cloud will be. You want a dense point cloud for a high resolution DOM. From the dense point cloud, complex algorithms used by the Structure-from-Motion technique assigns the vertical value and geospatial coordinates to each point. This allows the DOM to be projected and positioned correctly. After values are assigned to each point, a mesh network is created, connecting each point in the dense point cloud. The denser the mesh network, the greater the resolution the DOM will be. The mesh comprises of polygons which are coloured grey to remove the mesh lines in the object. When the mesh is complete and is projected in 3D, the coloured images of the object cover the 3D mesh.

Image 4: Method of generating a DOM 3D mesh using the Structure-from-Motion technique. Credit: Europlanet Society, The Geology and Planetary Winter Mapping School, February, 2021.

For my lab friends, I will be able to show you on my computer the data I got to work with. For everyone else reading this, I would recommend watching a YouTube video and visiting Sketchfab-PlanMap that will explain how you can work with digital outcrop model data, photogrammetry data, download 3D models of geologic outcrops on Mars, and create 3D surfaces using SHARAD radar data.

I have added the images I submitted to the Europlanet Society Geology and Planetary Winter Mapping School.