KBCG20 epistemic uncertainty only read from lowest distance bin of table

The distance lookup values in the KBCG20 GroundMotionTable are not log10 so epistemic values are only ever pulled from the lowest (10km) distance. Magnitude and spectral period are ok.

If we look at figure 14 in the Kuehn model EQ Spectra paper, the Cascadia and Alaska regionalized models are fairly stable across magnitudes and distance, so there will be little change to the Conterminous U.S NSHM (interface and intraslab) and Alaska NSHM (intraslab) when a fix is implemented. In Alaska we use the global models for interface, and what is troubling is the huge increase (values up to 1.0 natural log units) that occurs at distances >400-500 km. The Parker and Abrahamson models never have epistemic uncertainty higher than ~0.4 natural log units, and the paper states that the steep slope of the Kuehn model at larger distances is more the result of a lack of data than any physical basis.

Consider limiting Kuehn epistemic to 0.4 or 0.45 natural log units. This fix could be included in an update to the model ID publication years.