Miscellaneous Changes

* Improved data-coverage function in FederatedASDFDataSet
* Improved handling of location codes in cross-correlation routines

Author: geojunky

seismic/ASDFdatabase/FederatedASDFDataSet.py

@@ -138,13 +138,15 @@ def get_global_time_range(self, network, station=None, location=None, channel=No
 
     # end func
 
-    def get_nslc_list(self):
+    def get_nslc_coverage(self):
         """
-        Get a list of all net, sta, loc, cha combinations featured in the database
+        Get a structured numpy array with named columns
+        'net', 'sta', 'loc', 'cha', 'min_st', 'max_et'
+        representing contents of the database
         @return:
         """
 
-        results = self.fds.get_nslc_list()
+        results = self.fds.get_nslc_coverage()
         return results
     # end if
 

seismic/ASDFdatabase/_FederatedASDFDataSetImpl.py

@@ -490,11 +490,17 @@ def get_global_time_range(self, network, station=None, location=None, channel=No
         return min, max
     # end func
 
-    def get_nslc_list(self):
-        query = "select net, sta, loc, cha from nslc"
+    def get_nslc_coverage(self):
+        query = "select net, sta, loc, cha, st, et from nslc"
         rows = self.conn.execute(query).fetchall()
 
-        return rows
+        fields = {'names': ['net', 'sta', 'loc', 'cha', 'min_st', 'max_et'],
+                  'formats': ['U10', 'U10', 'U10', 'U10', 'f8', 'f8']}
+        result = np.zeros(len(rows), dtype=fields)
+
+        for i, row in enumerate(rows): result[i] = row
+
+        return result
     # end if
 
     def get_stations(self, starttime, endtime, network=None, station=None, location=None, channel=None):

seismic/ASDFdatabase/viewer/FederatedASDFViewer.py

@@ -482,9 +482,9 @@ def addWidget(emitter):
         # populate net, sta, loc, cha dict
         self.nslc_dict = defaultdict(lambda: defaultdict(lambda: defaultdict(list)))
 
-        nslc_list = self.fds.get_nslc_list()
-        for row in nslc_list:
-            net, sta, loc, cha = row
+        nslc_coverage = self.fds.get_nslc_coverage()
+        for row in nslc_coverage:
+            net, sta, loc, cha, _, _ = row
             self.nslc_dict[net][sta][loc].append(cha)
         # end for
 

seismic/xcorqc/correlator.py

@@ -193,7 +193,6 @@ def cull_pairs(pairs, keep_list_fn):
 
     startTime = UTCDateTime(start_time)
     endTime = UTCDateTime(end_time)
-    stationsCache = defaultdict(list)
     for pair in proc_stations[rank]:
         netsta1, netsta2 = pair
 
@@ -212,7 +211,9 @@ def get_loccha(cha1, cha2):
                      cha1 and cha2 -- e.g. ['.SHZ', '00.BHZ'], ['01.HHZ']
             """
             result = [[], []]
-            for chidx, (netsta, cha, ds) in enumerate(zip((netsta1, netsta2), (cha1, cha2), (ds1, ds2))):
+            for chidx, (netsta, cha, ds) in enumerate(zip((netsta1, netsta2),
+                                                          (cha1, cha2),
+                                                          (ds1, ds2))):
                 if('*' in cha1):
                     cha = cha.replace('*', '.*')  # hack to capture simple regex comparisons
                 # end if
@@ -225,12 +226,14 @@ def get_loccha(cha1, cha2):
 
                 net, sta = netsta.split('.')
 
-                if((start_time, end_time, net, sta) in stationsCache):
-                    stations = stationsCache[(start_time, end_time, net, sta)]
-                else:
-                    stations = ds.fds.get_stations(start_time, end_time, net, sta)
-                    stationsCache[(start_time, end_time, net, sta)] = stations
-                # end if
+                # find a list of entries where network and station names match and
+                # start- and end-times overlap with data coverage. Note that this is
+                # an approximate estimate and an actual cross-correlation may not be
+                # computed due to gaps in data
+                stations = ds.nslc_coverage[(ds.nslc_coverage['net'] == net) & \
+                                            (ds.nslc_coverage['sta'] == sta) & \
+                                            (ds.nslc_coverage['max_et'] >= startTime.timestamp) & \
+                                            (ds.nslc_coverage['min_st'] <= endTime.timestamp)]
 
                 loc_pref = location_preferences_dict[netsta]
                 ulocs = set()

seismic/xcorqc/utils.py

@@ -20,6 +20,8 @@ def __init__(self, asdf_file_name, netsta_list='*'):
         self._earth_radius = 6371  # km
 
         self.fds = FederatedASDFDataSet(asdf_file_name)
+        self.nslc_coverage = self.fds.get_nslc_coverage()
+
         # Gather station metadata
         netsta_list_subset = set(netsta_list.split(' ')) if netsta_list != '*' else netsta_list
         self.netsta_list = []