Systems Science Friday Noon Seminar Series

Improving Oregon's Geologic Map with Satellite Data and Machine Learning

Files

Download

Download (347.9 MB)

Download Captions (95 KB)

Loading...

Media is loading
 

Format

Video: MP4; File size: 347 MB; Duration: 57:15

Date

2-10-2023

Abstract

The Geologic Map of Oregon needs help! This map is unique among the states in the USA for being composed of the best available data for each area, based on what mapping efforts have taken place, which leaves many areas mapped in low detail. Using a Systems Science approach, both in framing the problem, and in the techniques used, will result in an improved map, in a shorter time frame. Spatial data are mapped at different scales for different purposes. They typically range from high resolution, like 1:24,000 using detailed methods, to regional scale maps in the range of 1:250,000, using less detailed methods. High resolution mapping requires collecting samples for analysis in the lab, as well as occupying an area for some time to do the detailed work, and is often motivated by a broader research question. Mapping Oregon at this scale could take a very long time, at our current rate of mapping. For large portions of the state, the scale of mapping is 1:125,000, or less detailed. In many cases nearby areas, with similar rock types, have been mapped at much higher resolutions, and can be used as training data for an AI approach to mapping the areas that are currently mapped in less detail. Sentinel-2 satellite data are freely available, with 10 bands of usable spectral data, some of which are at a resolution of 20 meters. By training a machine learning model on the spectral signature of areas mapped at high resolution, predictions can be made of where to find outcrops for mapping in the low-resolution areas, which can shorten the time required to map an area in detail. With the level of detail available from Sentinel-2, we are limited to finding outcrops, as opposed to creating actual geologic maps. Future efforts, using higher resolution satellite data, could drastically shorten the time for mapping the rest of the state in the level of detail that would be useful to many stakeholders. The Systems approach has made the connections and relationships in the data and outcomes more evident, and provided the techniques for fulfilling this ambitious vision for improving Oregon’s geologic map.

Biographical Information

Percy is a spatial data scientist in PSU’s Geology department, with strong ties to the Systems Science program. Starting as a hacker in the early 80s he became fascinated with the “exotic computing” of Systems Science, like Cellular Automata, Fractals, and Chaos Theory, eventually taking their Neural Networks class in 1991. Working with databases of neurological data at the time seemed somewhat synergistic, and medical research computing was the primary work activity in the 80s and most of the 90s at Legacy and OHSU. A deep appreciation for Earth Science led to pursuing an education in PSU’s Geology program, which led to employment there as the spatial data manager in 1998, which led to teaching classes there since 1999. An MS degree in Systems Science was completed in 2016, with several presentations on Reconstructability Analysis applied to spatial data resulting, and ongoing research in this topic, anticipated to be published in 2023. With his background in large databases, the geologic map of Oregon was an obvious project to be involved with, and he has served on the standards committee for it, as well as performing the role of framework coordinator for this data set for the state of Oregon since 2019. Pondering the heterogeneity of this database led to the realization that areas mapped at high resolution could be used as training data for an AI to map the rest of the state, and led to the presentation you are going to see. When he's not playing with data, Percy can be found playing music, having played in local bands since 1982, with styles ranging from punk to R & B. Currently in a Christmas band called Pokey Twig and the Yuletide Kindling, a rock outfit called the Lazy Champions, and a side group that performs the Violent Femmes first album, occasionally.

Subjects

Mapping -- Cartography, Maps -- Automation, Remote sensing

Disciplines

Systems Science

Persistent Identifier

https://archives.pdx.edu/ds/psu/39293

Rights

© 2023 David Percy

© Copyright the author(s) IN COPYRIGHT: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DISCLAIMER: The purpose of this statement is to help the public understand how this Item may be used. When there is a (non-standard) License or contract that governs re-use of the associated Item, this statement only summarizes the effects of some of its terms. It is not a License, and should not be used to license your Work. To license your own Work, use a License offered at https://creativecommons.org/

Improving Oregon's Geologic Map with Satellite Data and Machine Learning

Share

COinS