Data for: Fast and flexible gamma dose rate calculation for three-dimensional fields and arbitrarily distributed points with building shielding effects

In the BBFMM runs, the target accuracy was typically ε = 10⁻⁵, while the cubic domain size, number of Chebyshev nodes, and tree depth were case-dependent (e.g., about L ≈ 3200 m, n ≈ 7, level ≈ 3 for the multi-building cases, and about L ≈ 12000 m, n ≈ 12, level ≈ 4 for the RIMPUFF cases).Spatially, the data are defined in a local simulation coordinate system rather than as long-term georeferenced observational grids.

The multi-building cases cover a horizontal domain of roughly 0–2000 m × 0–2000 m and source heights of about 0–800 m, with a horizontal source spacing of about 5 m and vertically stretched layers (near-surface steps of about 5–10 m, coarsening to tens or hundreds of meters aloft). Monitoring points on horizontal planes are typically sampled at 20 m intervals. Dose fields are provided on fixed-height planes (e.g., z = 1 m and z = 100 m) as approximately 100 × 100 matrices, and on vertical sections (e.g., x = 1300 m and y = 1150 m) as approximately 42 (vertical layers) × 100 (horizontal points) matrices.

For the RIMPUFF-related cases, sources span about x, y ∈ [−2500, 2500] m and z ∈ [0, 700] m at roughly 10 m spacing (about 1.78 × 10⁷ source points). Near-surface monitoring planes (z = 1 m) use a 50 m spacing, yielding about 101 × 101 dose fields, together with the concentration field in nongdu_z=1.mat (about 501 × 501). Temporally, the data represent selected simulation instants or steady/instantaneous integral samples (file timestamps mainly from 2024) rather than continuous long-term time series.

The timing table reports algorithmic runtimes for different mesh sizes only, in seconds. The DCF cases correspond to idealized uniformly contaminated spherical/disk geometries for method verification at the radionuclide level, with spatial extent controlled by an equivalent integration radius and a near-ground receptor (e.g., z = 1 m).The Data_submit/DCFs directory contains DCF verification tables. DCF_air_multieng_results_newxp1017_fgr12_ba137m_noskin.xlsx lists immersion (air) DCF comparisons for 24 radionuclides, with columns for index, radionuclide name, reference DCF (labeled FGR15; unit Sv/s per Bq/s), three-dimensional direct integration (3D integral), three-dimensional BBFMM result (BBFMM_3D), and relative error (dimensionless).

DCF_dep_results_newxp1029_fgr12_ba137m_noskin2.xlsx lists ground-deposition DCF comparisons for 12 radionuclides (reference values labeled IAEA; same unit Sv/s per Bq/s), together with two-dimensional integration (2D integral), BBFMM_2D, and relative error. For most radionuclides the relative errors are within a few percent, whereas soft-gamma/low-energy nuclides (e.g., Xe-131m, Kr-83m) can show substantially larger relative deviations, mainly due to low-energy photon attenuation, kernel approximation, discretization, and multipole truncation.The Data_submit/muti-buildings directory corresponds to multi-building shielding and building-resolving concentration scenarios. shuju_x=1.25km_qes.xlsx and shuju_y=1.14km_qes.xlsx provide dose comparisons along profiles near x ≈ 1.25 km and y ≈ 1.14 km.

Sheet1 contains about 21 valid records: the first column is a coordinate (0–2 with a step of about 0.1; depending on the case, this may be interpreted as an energy or normalized parameter scan), followed by FMM and 3D integral results with buildings, and the corresponding FMM and 3D integral results without buildings. Sheets 2–4 each contain about 50 records of fmm, 3D integral, and semi-infinite cloud results along different profile coordinates (Sheets 2/3: about 1–491 with a step of about 10; Sheet 4: about −2500–2400 m with a step of about 100 m). bizhi_x=1.25km.xlsx and bizhi_y=1.14km.xlsx each contain about 21 valid records with columns for coordinate, FMM, and 3D integral, representing the ratio of doses with buildings to those without buildings (a dimensionless shielding-type factor).

Planar and sectional dose fields are stored as MATLAB .mat files: z=1_* and z=100_* are approximately 100 × 100 horizontal-plane matrices; x=1300_* and y=1150_* are approximately 42 × 100 vertical-section matrices. In the filenames, fmm, direct, and semi denote BBFMM, direct integration, and the semi-infinite cloud approximation, respectively. Files such as dangedian_2000m_*.mat and wangge_2000m_nest.mat provide about 101 × 101 single-point/grid comparison results over the 2000 m domain.

Dose magnitudes are typically from about 10⁻⁵ to 10⁻¹², consistent with activity normalization and the kernel units used in the solver, and should be interpreted as external dose rates under that same calibration.The Data_submit/Rimpuff directory corresponds to plume-dispersion-coupled dose calculations. nongdu_z=1.mat stores a two-dimensional concentration (or activity-density) field near z = 1 m (about 501 × 501); z=1_fmm_rimpuff.mat, z=1_semi_rimpuff.mat, and z=1_semi_dose.mat store the corresponding near-surface dose fields (about 101 × 101).

The profile tables shuju_x=300_rimpuff.xlsx, shuju_x=600_rimpuff.xlsx, shuju_y=-140_rimpuff.xlsx, and shuju_y=410_rimpuff.xlsx each contain about 50 records with columns for coordinate, FMM, 3D integral, and Semi_infinite cloud; the profile coordinates range over about ±2.5 with a step of about 0.1 (kilometer scale, corresponding to about ±2500 m). shuju_time.xlsx contains 6 valid records with columns for mesh size, fmm_time, and direct_time, reporting BBFMM versus direct-method runtimes (in seconds) for different mesh sizes to illustrate the acceleration.