some edits

openquake/hazardlib/gsim/kuehn_2020.py

@@ -472,10 +472,15 @@ def get_mean_values(C, region, imt, trt, m_b, ctx, a1100=None,
         else:
             usgs_baf = 1.0
 
-        if m9_basin_term:
+        # For KuehnEtAl2020 in US 2023 either the M9 basin term OR the GMM's
+        # basin term is applied (i.e. it is not additive to GMM basin term here
+        # as can be seen in the code - line 457 to 499 of KuehnEtAl_2020.java)
+        if m9_basin_term and imt != PGV:
+            # Apply + np.log(2.0) for long period motions at deep basin sites
             mean = _apply_m9_basin_term(ctx, imt, mean, usgs_baf)
-
-        mean += get_basin_term(C, ctx, region) * usgs_baf
+        else:
+            # Use the GMM's own basin term
+            mean += get_basin_term(C, ctx, region) * usgs_baf
 
     return mean
 

openquake/hazardlib/gsim/mgmpe/m9_basin_term.py

@@ -36,7 +36,8 @@ def _apply_m9_basin_term(ctx, imt, mean, usgs_baf=1.0):
 class M9BasinTerm(GMPE):
     """
     Implements a modified GMPE class that can be used to implement the "M9"
-    US 2023 NSHM basin amplification adjustment.
+    US 2023 NSHM basin amplification adjustment (an additive factor for long
+    period ground-motions in the Seattle Basin region).
      
     This implementation is based on the description of the M9 adjustment 
     within the Moschetti et al. (2024) EQ Spectra article on the conterminous

openquake/hazardlib/gsim/parker_2020.py

@@ -46,7 +46,7 @@
 
 def _get_adjusted_m9_basin_term(C, z2pt5):
     """
-    Return the adjusted version of the m9 basin term as detailed within the 
+    Return the adjusted version of the M9 basin term as detailed within the 
     USGS NSHM java code for the Abrahamson and Gulerce 2020 subduction GMM.
     """
     delta_z2pt5_adj = np.log(z2pt5 * 1000.) - np.log(1179.)
@@ -59,6 +59,7 @@ def _get_adjusted_m9_basin_term(C, z2pt5):
         fb_adj[fb_adj == 0.] = C['C_e3'] * delta_z2pt5_adj
     return np.log(2.0) - fb_adj
 
+
 def _get_sigma_mu_adjustment(sat_region, trt, imt, epi_adjs_table):
     """
     Get the sigma_mu_adjustment (epistemic uncertainty) factor to be applied
@@ -463,19 +464,23 @@ def compute(self, ctx: np.recarray, imts, mean, sig, tau, phi):
                 C, C_PGA, ctx.mag, ctx.rrup, m_b)
             fd = _depth_scaling(trt, C, ctx)
             fd_pga = _depth_scaling(trt, C_PGA, ctx)
-            fb = _get_basin_term(C, ctx, self.region, self.basin) * usgs_baf
+            fb = _get_basin_term(C, ctx, self.region, self.basin)
             flin = _linear_amplification(self.region, C, ctx.vs30)
             fnl = _non_linear_term(C, imt, ctx.vs30, fp_pga, fm_pga, c0_pga,
                                    fd_pga)
 
             # The output is the desired median model prediction in LN units
             # Take the exponential to get PGA, PSA in g or the PGV in cm/s
             pre_baf_mean = fp + fnl + flin + fm + c0 + fd
-            if self.m9_basin_term:
-                m9_adj = _get_adjusted_m9_basin_term(C, ctx.z2pt5)
-                mean[m] = pre_baf_mean + m9_adj
-            else:
-                mean[m] = pre_baf_mean + fb
+
+            # Get the m9 basin adjustment if long period SA for deep basin sites
+            if self.m9_basin_term and imt != PGV:
+                if imt.period >= 1.9:
+                    m9_adj = _get_adjusted_m9_basin_term(C, ctx.z2pt5)
+                    fb[ctx.z2pt5 >= 6.0] += m9_adj[ctx.z2pt5 >= 6.0]
+
+            # Now get the mean with basin term added
+            mean[m] = pre_baf_mean + (fb * usgs_baf)
 
             if self.sigma_mu_epsilon and imt != PGV: # Assume don't apply to PGV
                 # Apply epistemic uncertainty scaling