SimPEG-EM1D: Gradient-Based 1D Inversion Software for Large-Scale Airborne Electromagnetic Data
Abstract
Water scarcity is becoming a prevalent issue for many countries around the world. To address this problem, groundwater experts are showing increased interest in the use of airborne electromagnetic (AEM) surveys as an effective tool for mapping hydrogeological units (e.g. aquifers, aquitards, and clays). Modern AEM systems can detect resistivity contrasts in the subsurface (e.g. resistive aquifer and conductive clay units) and quickly cover large areas. The same technology is used extensively in mineral exploration for the characterization of regional geology.
1D inversions are a standard tool used to interpret AEM data and produce a layered Earth model. However, most of the existing codes that perform this function are proprietary black-boxes. This limits accessibility, reproducibility, and advancements in inversion methodologies. Our aim with SimPEG-EM1D is two-fold: to increase the accessibility and usefulness of AEM inversions for the hydrogeophysical community and to effectively handle large-scale AEM inversion problems. SimPEG-EM1D is open source, written in Python, and connected to the larger SimPEG community. To achieve our goals, we address both numerical and modeling challenges. First from an efficiency standpoint, we parallelize the forward solves and sensitivity computations at each sounding. Solving the semi-analytic 1D EM problem across multiple processors allows us to easily scale our algorithm to large data sets. Secondly, we improve our modeling capabilities by leveraging functionalities already developed within the broader SimPEG ecosystem. Various geological assumptions can easily be incorporated to constrain the solution, such as the ability to apply compact norms or employ spatial constraints between soundings. Joint inversion of different types of AEM and ground-based EM data are also possible. In this presentation, we provide an overview of the capabilities of SimPEG-EM1D and demonstrate its use with synthetic and field examples.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMNS53A0557K
- Keywords:
-
- 0520 Data analysis: algorithms and implementation;
- COMPUTATIONAL GEOPHYSICSDE: 0545 Modeling;
- COMPUTATIONAL GEOPHYSICSDE: 0599 General or miscellaneous;
- COMPUTATIONAL GEOPHYSICSDE: 1999 General or miscellaneous;
- INFORMATICS