$F = [FM] == 'R' ? 1.28 : 1.00;
$A = $F*(0.14*atan([Mw]-6.25)+0.37);
$R2 = 3.67*[Mw]-12.42;
$D2 = -0.125*cos(1.19*([Mw]-6.15))+0.525;
$R5 = 18.04*pow([Mw], 2)-167.9*[Mw]+476.3;
$D3= [Vs30:m/s]>484.5?0.65:0.35;
return exp(log($A)-0.5*log(pow(1-[dr:km]/$R2, 2)+4*pow($D2, 2)*[dr:km]/$R2)-0.5*log(pow(1-sqrt([dr:km]/100), 2)+4*pow($D3, 2)*sqrt([dr:km]/100))-0.24*log([Vs30:m/s]/484.5)-0.5*log(pow(1-sqrt([dr:km]/$R5), 2)+4*pow(0.7, 2)*sqrt([dr:km]/$R5)));

• Douglas, J., "Ground-motion prediction equations 1964-2010", 2011
• Graizer, V.; Kalkan, E.; Lin, K.-W., "Extending and testing Graizer-Kalkan ground motion attenuation model based on Atlas database of shallow crustal events", 2010

See Douglas (2011), page 262.

N.B. We assume to model the basin effect that for rock class(Vs30 > 484.5 m/s) the sediment depth is < 1 km, while for soil class(Vs30 <= 484.5 m/s) the sediment depth is >= 1 km.

Validation: M = 5.0, d = 10 km, Vs30 = 800 m/s, FM = 'S' : PGA = 0.1406083590235920 g M = 6.0, d = 10 km, Vs30 = 800 m/s, FM = 'S' : PGA = 0.34939589252512 g M = 5.0, d = 100 km, Vs30 = 800 m/s, FM = 'S' : PGA = 0.0070300667910307 g