found bug with lon_ix in WOA interpolation

was causing incorrect weights (-ve, > 1)

bgcArgo/interp.py

@@ -174,22 +174,26 @@ def interp_woa_data(track, woa_track, data, verbose=False):
         xwt[1].append(lat_wt)
 
         lon_ix1 = np.where(lon < track[i,2])[0]
+        print(lon[lon_ix1])
         if lon_ix1.shape[0] == 0:
             lon_ix1 = 0
         else:
-            lon_ix1 = lon_ix1[0]
+            lon_ix1 = lon_ix1[-1]
         lon_ix2 = lon_ix1 + 1
 
         if lon_ix2 == lon.shape[0]:
-            print('NOTE: longitude is unbounded, giving all averaging weight to nearest observation')
-            lon_ix1 = lon_ix1 - 1
-            lon_ix2 = lon_ix2 - 1
+            if verbose:
+                sys.stdout.write('NOTE: longitude is unbounded, giving all averaging weight to nearest observation\n')
+            lon_ix1 -= 1
+            lon_ix2 -= 1
             lon_wt = 0
         else:
             dx1 = lon[lon_ix2] - lon[lon_ix1]
             dx2 = track[i,2] - lon[lon_ix1]
             lon_wt = (dx1 - dx2) / dx1
 
+            print([lon_wt, dx1, dx2, lon[lon_ix1], lon[lon_ix2], track[i,2]])
+
         xwt[2].append(lon_wt)
 
         # ---------------------------------------------------------------------