Validation data of parallel 3D surface-borehole electromagnetic forward modeling

Forward modeling of geophysical electromagnetic fields over large three-dimensional volumes is a heavy computational task that demands effective accelerating strategies. As a solution to this computational challenge, a hybrid parallel computing algorithm with multiple meshes has been previously proposed for 3D forward modeling of time-domain electromagnetic (TEM) fields by Liu et al. (2019). MPI and OpenMP were used for parallel computing and multiple meshes for optimizing the design of the geological model and frequencies used in forward modeling. The data presented in this paper offers complementary information on the calculation of the different components, such as the model discretization with regular or multiple meshes, the parallel computing with even or uneven modes, and an example of 3D TEM forward modeling through the proposed algorithm.

All data are organized in one folder, named "ModSBTEM", with six subfolders (Fig. 1), including all the source codes and the input/output data used in the forward modeling. Detailed information can be found in the "readme" file. In the subfolder "ModSBTEM_3D_LbyL.v1", regular meshes are used with even mode parallelization, i.e., the tasks are assigned to the processors evenly. In the subfolder "ModSBTEM_3D_LbyL.v2", regular meshes are also used with uneven mode parallelization, i.e., the tasks are assigned to the processors randomly. In the subfolder "ModSBTEM_3D_LbyL.v3", the forward modeling is parallelized with uneven mode, and the 3D volume is discretized by multiple meshes. In the subfolder "ModSBTEM_3D_LbyL_Hybrid.v2.2", the forward modeling is parallelized with hybrid MPI and OpenMP.
All codes are compiled with the programming language ANSI Standard Fortran 95, running with the Visual Studio and Intel Parallel Studio XE. The file 'user manual.txt' describes the main information and data format used in the codes. Also, each subfolder contains a note file indicating the main structures of the codes.
The forward modeling process is shown through an example. Fig. 2 describes the survey configuration, including a transmitter loop with an array of TEM receivers in five boreholes within a 3D earth model. The main transmitter waveform is trapezoidal with a ramp (Fig. 3). The geology and physical properties of lithologies are based on a typical geological environment in the Abitibi greenstone belt of Canada (Table 1). The surface-borehole TEM responses generated by the parallelized code in five boreholes are illustrated in Fig. 4.

Design
It is well known that the number of physical processors limits the speedup of parallel computing [3]. A fine mesh in the early time and a coarser mesh in the late time is proved as an efficient strategy to Specifications Table   Subject Geophysics Specific subject area Parallel computing in TEM forward modeling Type of data Value of the data Forward modeling helps to test different geological scenarios and to study the significance of geophysical exploration methods for specific geological settings.
Geophysicists who are looking for an example of TEM field survey design with high performance and quick TEM forward modeling, benefit from data presented here. The present data can be used as a benchmark for future validations and comparisons. The present data and methods can be used for further inverse modeling developments.     enhance the performance of parallel computing. We assign each frequency its effective region and mesh strategy in the frequency-domain simulation, and then the frequency-domain response is transformed into the time domain. In order to illustrate how to integrate multiple meshes into 3D modeling, we deployed this strategy in an example in the next section.

Illustration of the method
This example aims to simulate the surface-borehole TEM response of the ore bodies (lithologies #11, #12, and #13) hidden in a complex geological environment (Fig. 2). The template geological model was constructed from 45 cross-sections inside a 3D volume of 3.6 km (NeS) by 1 km (E-W) by 1.8 km (depth). Since the TEM response results from the coupling of the emitted electromagnetic field and the geological model, Fig. 3 (Table 1). Fig. 4 illustrates the TEM forward modeling responses from the 3D parallelized code.