FRISK88M^{TM} is our most sophisticated software package for studying the probabilistic seismic hazard associated with a single site or for seismic hazard mapping. It operates on multiple-weighted assumptions and produces a consolidated risk evaluation. This package uniquely accounts for both randomness and uncertainty.

Randomness (sometime called "aleatory uncertainty") is variability that cannot be reduced with more or better observations. The only way to reduce randomness is to completely change the model (e.g. using some sort of tectonic physical model). Uncertainty (sometimes called "epistemic uncertainty") results from inaccurate or incomplete information and can be reduced or eliminated given better models or additional observations.

A common seismic hazard plot indicates spectral response amplitude (or ground motion amplitude) on the x-axis and annual probability of exceedence (APE) on the y-axis. A single curve of this plot reflects the randomness of the hazard.

We may recognize that several assumptions are uncertain within the model, e.g. suppose that the fault slip rate may be higher or lower than originally estimated. A new hazard curve can be computed for each new assumption. Each curve has a weight associated with it derived from logic tree analysis and user-supplied weights. These curves can then be reduced to a set of fractile curves and a mean curve.

A profile of the uncertainties can be very useful when choosing design or retrofit criteria. Also, analysts get a sense of how sensitive a study is to various input interpretations.

FRISK88M^{TM} allows multiple, weighted assumptions on the input parameters. The following is a list of models and parameters that can be treated as uncertain:

1. Ground motion equation

2. Source locations (Trace or area source boundary)

3. Source profiles (fault dip angles, depths, etc.)

4. Seismicity recurrence model (e.g. characteristic, exponential)

5. Maximum magnitude

6. Recurrence rate

7. Slip rate

8. Recurrence slope (b-, beta-values)

Other input items include:

1. List of amplitudes to study (acceleration, velocity or displacement)

2. Magnitude integration step size (faults)

3. Vertical and horizontal rupture integration step size (faults)

4. Number of integration steps (areas)

Using the above weighted assumptions, a logic tree can be built that looks like the following (simplified):

1. Source 2. Max Mag 3. Rate-Beta 4. Depth Pair

The program then efficiently executes each branch of the tree and determines an individual hazard curve. Given that every hazard curve has a weight, mean and fractile annual probabilities can be calculated at each amplitude level. These fractiles quantify the epistemic uncertainty of the hazard, resulting from epistemic uncertainties in the input parameters (see the above list).

FRISK88M^{TM} requires you to construct a text input file. This file is thoroughly described in the User’s Manual. The program produces text output files for the hazard, uniform hazard spectra and deaggregation of hazard by magnitude, distance and ground motion epsilon. No plotting functions are provided with the program.