Hazard Comparisons with NSHM18
This report compares hazard between the NSHM23 and NSHM18 WUS ERFs, attempting to tease apart the contributions from the fault and gridded seismicity models to overall hazard changes.
All comparisons are done with a branch-averaged mean ERF model (as is used in the final NSHM), and a single ground motion model, ASK (2014), using uniform and default site conditions. All maps are for 1s SA, 2% in 50 year. The Cascadia subduction zone implementation is excluded in these comparisons.
Table Of Contents
- Fault-Based Hazard Comparisons
- Gridded Seismicity Model Hazard Changes
- Complete Model Hazard Changes
- Full Model Rate and Moment Rate Comparisons
- Smoothed Moment Rate Hazard vs Model Hazard Change Comparisons
- Smoothed M5 Rate Hazard vs Model Hazard Change Comparisons
Fault-Based Hazard Comparisons
This section compares NSHM23 with NSHM18, but holding the gridded seismicity model constant in order to focus on fault-based changes (due both to ingredient and methodological changes) in areas where hazard is dominated by modeled faults.
| NSHM23 | NSHM18 w/ NSHM23 Gridded |
|---|---|
 |  |
Ratio | Difference |
 |  |
Hazard Comparison Statistics:
Within 1% | Within 5% | Within 10% | % Range | Average % Change |
|---|---|---|---|---|
| 20.47% | 59.11% | 76.49% | [-55.27%, 408.32%] | 5.74% |
Within 0.05g | Within 0.1g | Within 0.2g | Diff Range (g) | Average Diff (g) |
| 95.20% | 98.44% | 99.78% | [-0.27, 0.60] | 0.01 |
Fault and Deformation Model Moment Changes
This section shows how fault-based deformation model moment changed between NSHM23 and NSHM18. We will use this map to mask hazard changes, highlighting only areas where changes are primarily due to methdological differences. In the ratio map, areas where moment exists in NSHM23 but not in NSHM18 are shown in yellow (i.e., a fault was added), and those with moment in NSHM18 but not in NSHM23 are shown in green (i.e., a fault was removed).
To smooth out minor geometry changes between the two models, moment is distributed around faults using a linear ramp in (3-D) up to 12 km; this is the same ramp used to carve out gridded seismicity near faults when constructing the NSHM23 gridded seismicity model.
| Ratio | Difference |
|---|---|
 |  |
Fault-Based Hazard Comparisons, Moment Chage Masked
This section masks hazard changes, only showing them near faults where NSHM23 fault-moment is within a given factor of NSHM18 fault-moment. Hazard changes in areas that remain may still be largely due to moment changes, but this helps to narrow the search space for identification of areas where methodological changes may dominate hazard changes.
Various thresholds are used, starting with one that only includes areas with faults in both models, and then the threshold decreases to only show areas where moment is similar in the two models. Areas where the sign of the hazard change differs from the sign of moment change are shown regardless of the threshold value.
| Ratio | Difference |
|---|---|
 |  |
 |  |
 |  |
 |  |
Near-Fault Hazard Comparison Statistics:
| Threshold | % Locs Within Moment Threshold | Within 1% | Within 5% | Within 10% | % Range | Average % Change |
|---|---|---|---|---|---|---|
| Areas w/ Moment in Both Models | 22.23% | 7.41% | 37.80% | 63.85% | [-55.27%, 408.32%] | 2.19% |
| <100% Moment Change Threshold | 19.05% | 8.38% | 40.61% | 66.56% | [-55.27%, 85.72%] | -0.96% |
| <50% Moment Change Threshold | 15.73% | 9.15% | 42.88% | 68.62% | [-43.85%, 78.42%] | -0.32% |
| <20% Moment Change Threshold | 10.89% | 10.52% | 46.54% | 71.43% | [-43.85%, 78.42%] | 1.32% |
| Threshold | % Locs Within Moment Threshold | Within 0.05g | Within 0.1g | Within 0.2g | Diff Range (g) | Average Diff (g) |
| Areas w/ Moment in Both Models | 22.23% | 81.46% | 93.73% | 99.05% | [-0.27, 0.60] | 0.01 |
| <100% Moment Change Threshold | 19.05% | 81.36% | 94.04% | 99.52% | [-0.27, 0.44] | 0.00 |
| <50% Moment Change Threshold | 15.73% | 81.99% | 94.32% | 99.55% | [-0.27, 0.44] | 0.00 |
| <20% Moment Change Threshold | 10.89% | 82.08% | 94.15% | 99.66% | [-0.21, 0.44] | 0.01 |
NSHM18 Special Cases
NSHM18 includes a number of special cases that are not carried forward. These also affect hazard comparisons beyond the moment changes shown above, and are plotted below.
NSHM18 Ingredient Runs vs NSHM18
We also ran inversions using NSHM18 deformation models as input, but using updated methodologies. This helps to isolate regions where methodological changes (the inversion, as well as elmination of the special cases above) dominate hazard change. All comparisons hold the NSHM23 gridded seismicity model constant.
We ran this test both using updated scaling relationships, as well as using Wells and Coppersmith (94); for the latter, results are masked in California as UCERF3 used their own scaling relationships. We also show results using the classic segmentation branches only, which should be most similar to NSHM18 outside of California.
NSHM18 Ingredients, NSHM23 Methodology (new scaling) | |
|---|---|
 |  |
NSHM18 Ingredients, NSHM23 Methodology (W-C '94) | |
 |  |
NSHM18 Ingredients, NSHM23 Methodology (new scaling, classic only) | |
 |  |
NSHM18 Ingredients, NSHM23 Methodology (W-C '94, classic only | |
 |  |
Gridded Seismicity Model Hazard Changes
These plots compare the gridded seismicity components of the models. For the ratio and difference maps, the contributions of NSHM23 fault sources are held constant (added to both models).
| NSHM23, Gridded Only | NSHM18, Gridded Only |
|---|---|
 |  |
Ratio | Difference |
 |  |
Hazard Comparison Statistics:
Within 1% | Within 5% | Within 10% | % Range | Average % Change |
|---|---|---|---|---|
| 3.38% | 16.28% | 33.01% | [-57.32%, 422.19%] | 11.88% |
Within 0.05g | Within 0.1g | Within 0.2g | Diff Range (g) | Average Diff (g) |
| 90.80% | 97.45% | 99.60% | [-0.25, 0.60] | 0.01 |
Complete Model Hazard Changes
These plots compare the full models, including both fault and gridded seismicity changes (both ingredients and methodology).
| Full NSHM23 | Full NSHM18 |
|---|---|
|  |
Full Ratio | Full Difference |
 |  |
Fault-Only Ratio | Fault-Only Difference |
| |
Gridded-Only Ratio | Gridded-Only Difference |
| |
Hazard Comparison Statistics:
Within 1% | Within 5% | Within 10% | % Range | Average % Change | |
|---|---|---|---|---|---|
| Full Models | 3.38% | 16.28% | 33.01% | [-57.32%, 422.19%] | 11.88% |
| Faults Only | 20.47% | 59.11% | 76.49% | [-55.27%, 408.32%] | 5.74% |
| Gridded Only | 3.38% | 16.28% | 33.01% | [-57.32%, 422.19%] | 11.88% |
Within 0.05g | Within 0.1g | Within 0.2g | Diff Range (g) | Average Diff (g) | |
| Full Models | 90.80% | 97.45% | 99.60% | [-0.25, 0.60] | 0.01 |
| Faults Only | 95.20% | 98.44% | 99.78% | [-0.27, 0.60] | 0.01 |
| Gridded Only | 90.80% | 97.45% | 99.60% | [-0.25, 0.60] | 0.01 |
Fault vs Gridded, Hazard Change Attribution
This section shows hazard changes, colored by the primary contributor to hazard change (faults or gridded seismicity) rather than the sign of that change. Small changes are masked out.
| Ratio Attribution | Difference Attribution |
|---|---|
 |  |
| Attributed to Faults | Attributed to Gridded | Sum | |
|---|---|---|---|
| Full Map | 20.26% | 79.74% | 100.00% |
| > 10% Change | 11.38% | 55.61% | 66.99% |
| > 0.05g Change | 4.71% | 4.49% | 9.20% |
Full Model Rate and Moment Rate Comparisons
Full Model Rate Comparisons
NSHM23, M≥5 | NSHM18, M≥5 |
|---|---|
 |  |
Ratio | Difference |
 |  |
NSHM23, M≥6.5 | NSHM18, M≥6.5 |
 |  |
Ratio | Difference |
 |  |
NSHM23, M≥7.5 | NSHM18, M≥7.5 |
 |  |
Ratio | Difference |
 |  |
Full Model Moment Rate Comparison
NSHM23 | NSHM18 |
|---|---|
 |  |
Ratio | Difference |
 |  |
Smoothed Moment Rate Hazard vs Model Hazard Change Comparisons
Here, we calculate simple hazard maps by placing point sources at every grid location with a G-R b=1 MFD with magnitudes in the range [6,7.5] that satisfies each cell's moment rate. We then compare hazard changes from this simple model to the actual model hazard change. Areas where simplified model hazard change is similar to full model hazard change are likely dominated by ingredient (moment rate) changes, and those that differ may be affected by methodological changes (although this simplified comparison is not definitive).
Note that these comparisons use final model moment rate maps, as opposed to the prior moment comparisons that used deformation model moment rates directly. Thus, any slip rate misfits will be incorporated into these comparisons.
| Simplified Full Model Moment Rate Hazard Ratio | Full Model Hazard Ratio |
|---|---|
 |  |
Scatter Plot | Hazard Change Comparison |
 |  |
| Simplified Fault Moment Rate Hazard Ratio | Fault Hazard Ratio |
|---|---|
 |  |
Scatter Plot | Hazard Change Comparison |
 |  |
| Simplified Gridded Moment Rate Hazard Ratio | Gridded Hazard Ratio |
|---|---|
 |  |
Scatter Plot | Hazard Change Comparison |
 |  |
Smoothed M5 Rate Hazard vs Model Hazard Change Comparisons
Here, we calculate simple hazard maps by placing point sources at every grid location with a G-R b=1 MFD with magnitudes in the range [5,7.5] that matches the total nucleation M5 rate in each cell. We then compare hazard changes from this simple model to the actual model hazard change. Areas where simplified model hazard change is similar to full model hazard change are likely dominated by ingredient (total rate) changes, and those that differ may be affected by methodological changes (although this simplified comparison is not definitive).
| Simplified Full Model M5 Rate Hazard Ratio | Full Model Hazard Ratio |
|---|---|
 |  |
Scatter Plot | Hazard Change Comparison |
 |  |
| Simplified Gridded M5 Rate Hazard Ratio | Gridded Hazard Ratio |
|---|---|
 |  |
Scatter Plot | Hazard Change Comparison |
 |  |