New Mexico Geological Society Annual Spring Meeting — Abstracts

Mapping geothermal resources using AI/ML

Tracy Kliphuis1, Hope Jasperson1 and Velimir Vesselinov1

1EnviTrace LLC, 1048 Mansion Ridge Road,, Santa Fe, NM, 87501, United States,

[view as PDF]

GeoTGo dashboard for AI/ML analysis, mapping, interpretation, and prediction of geothermal resources.

Geothermal energy is a vital renewable resource. However, exploring and producing geothermal resources can be challenging, expensive, and risky. One key factor contributing to these challenges is the need for accurate characterization and mapping of geothermal resources. Without a comprehensive understanding of the location, extent, and properties of geothermal reservoirs, it is difficult to plan and execute successful drilling operations.

Geothermal reservoirs are typically deep underground and hidden without explicit surface manifestation. Additionally, the geological formations associated with geothermal resources can be complex and spatially variable. To overcome these challenges, various geophysical and geological techniques are employed to characterize and map geothermal resources. These techniques include seismic surveys, gravity surveys, magnetotelluric surveys, and geochemical analyses. By combining data from multiple sources, we can create detailed models of geothermal reservoirs, which are essential for planning drilling operations and optimizing resource extraction.

We have developed cloud-based interactive software and a user-friendly interface for geothermal exploration and utilization. It is called GeoTGo (https://envitrace/geotgo). GeoTGo applies supervised, unsupervised (self-supervised), and physics-informed machine learning methods. The methods are designed to operate and solve real-world problems critical for our customers with minimal user input. GeotTGo allows for the joint processing of diverse datasets with different sizes, pedigrees (geological, geochemical, geophysical, etc.), attribute types (qualitative and quantitative), accuracy, and support scales. Large and complex datasets may require extensive use of cloud-computing resources. The software is designed to operate on alternative cloud computing services (Google Cloud, Microsoft Azure, AWS, etc.). Our company provides tiered licensing and commercial support. Machine learning and artificial intelligence methods in GeoTGo are based on our existing open-source algorithms (SmartTensors,, and MADS, However, the developed software is proprietary. It includes a frontend for data management, control, and visualization.

GeoTGo is preloaded with predeveloped ML models and datasets related to the Great Basin in the Southwestern U.S. It also includes New Mexico geothermal datasets. GeoTGo is designed to provide information for local communities in these regions to understand and develop their geothermal resources. Our work bridges the gap between technology advancements and community needs by facilitating interactions between the geothermal industry, regulators, stakeholders, and end-users. GeoTGo aims to provide equitable and sustainable solutions. GeoTGo will help accelerate the development of geothermal energy and contribute to the achievement of net-zero carbon emissions.


machine learning, data analyses, artificial intelligence, community-based geothermal development

pp. 45-46

2024 New Mexico Geological Society Annual Spring Meeting
April 19, 2024, Macey Center, Socorro, NM
Online ISSN: 2834-5800