fixes to squirrel persistence // adding local magnitude argetina

src/qseek/magnitudes/local_magnitude_model.py

@@ -378,4 +378,15 @@ def get_amp_attenuation(dist_hypo_km: float, dist_epi_km: float) -> float:
         return 0.89 * np.log10(dist_epi_km / 100) + 0.00256 * (dist_epi_km - 100) + 3
 
 
+class ArgentinaVolcanoes(WoodAnderson, LocalMagnitudeModel):
+    author = "Montenegro et al. (2021)"
+
+    epicentral_range = Range(0.0 * KM, 100.0 * KM)  # Bounds are not clear
+    component = "north-east-separate"
+
+    @staticmethod
+    def get_amp_attenuation(dist_hypo_km: float, dist_epi_km: float) -> float:
+        return 2.76 * np.log10(dist_epi_km) - 2.48
+
+
 ModelName = Literal[LocalMagnitudeModel.model_names()]

src/qseek/octree.py

@@ -306,7 +306,7 @@ class Octree(BaseModel):
         description="The reference location of the octree.",
     )
     root_node_size: PositiveFloat = Field(
-        default=2 * KM,
+        default=1 * KM,
         description="Initial size of the root octree node at level 0 in meters.",
     )
     n_levels: int = Field(

src/qseek/pre_processing/downsample.py

@@ -16,7 +16,7 @@ class Downsample(BatchPreProcessing):
 
     process: Literal["downsample"] = "downsample"
     sampling_frequency: PositiveFloat = Field(
-        100.0,
+        default=100.0,
         description="The new sampling frequency in Hz.",
     )
 

src/qseek/search.py

@@ -31,6 +31,7 @@
 from qseek.models.detection_uncertainty import DetectionUncertainty
 from qseek.models.semblance import Semblance, SemblanceCache
 from qseek.octree import NodeSplitError, Octree
+from qseek.pre_processing.frequency_filters import Bandpass
 from qseek.pre_processing.module import Downsample, PreProcessing
 from qseek.signals import Signal
 from qseek.station_weights import StationWeights
@@ -215,11 +216,11 @@ class Search(BaseModel):
         description="Station inventory from StationXML or Pyrocko Station YAML.",
     )
     data_provider: WaveformProviderType = Field(
-        default=PyrockoSquirrel(),
+        default_factory=PyrockoSquirrel.model_construct,
         description="Data provider for waveform data.",
     )
     pre_processing: PreProcessing = Field(
-        default=PreProcessing(root=[Downsample(sampling_frequency=100.0)]),
+        default=PreProcessing(root=[Downsample(), Bandpass()]),
         description="Pre-processing steps for waveform data.",
     )
 
@@ -379,7 +380,7 @@ def write_config(self, path: Path | None = None) -> None:
         logger.debug("writing search config to %s", path)
         path.write_text(self.model_dump_json(indent=2, exclude_unset=True))
 
-        logger.debug("dumping stations...")
+        logger.debug("dumping stations")
         self.stations.export_pyrocko_stations(rundir / "pyrocko_stations.yaml")
 
         csv_dir = rundir / "csv"
@@ -458,7 +459,7 @@ async def prepare(self) -> None:
         Returns:
             None
         """
-        logger.info("preparing search...")
+        logger.info("preparing search components")
         self.data_provider.prepare(self.stations)
         await self.pre_processing.prepare()
 
@@ -489,7 +490,7 @@ async def start(self, force_rundir: bool = False) -> None:
 
         await self.prepare()
 
-        logger.info("starting search...")
+        logger.info("starting search")
         stats = self._stats
         stats.reset_start_time()
 

src/qseek/waveforms/squirrel.py

@@ -14,7 +14,6 @@
     PositiveInt,
     PrivateAttr,
     computed_field,
-    field_validator,
     model_validator,
 )
 from pyrocko.squirrel import Squirrel
@@ -113,10 +112,14 @@ class PyrockoSquirrel(WaveformProvider):
 
     provider: Literal["PyrockoSquirrel"] = "PyrockoSquirrel"
 
-    environment: DirectoryPath = Field(
-        default=Path("."),
+    environment: DirectoryPath | None = Field(
+        default=None,
         description="Path to a Squirrel environment.",
     )
+    persistent: str | None = Field(
+        default=None,
+        description="Name of the persistent collection for faster loading.",
+    )
     waveform_dirs: list[Path] = Field(
         default=[],
         description="List of directories holding the waveform files.",
@@ -151,18 +154,20 @@ class PyrockoSquirrel(WaveformProvider):
     def _validate_model(self) -> Self:
         if self.start_time and self.end_time and self.start_time > self.end_time:
             raise ValueError("start_time must be before end_time")
+        if not self.waveform_dirs and not self.persistent:
+            raise ValueError("no waveform directories or persistent selection provided")
         return self
 
-    @field_validator("waveform_dirs")
-    def check_dirs(cls, dirs: list[Path]) -> list[Path]:  # noqa: N805
-        if not dirs:
-            raise ValueError("no waveform directories provided!")
-        return dirs
-
     def get_squirrel(self) -> Squirrel:
         if not self._squirrel:
-            logger.info("initializing squirrel waveform provider")
-            squirrel = Squirrel(str(self.environment.expanduser()))
+            logger.info(
+                "initializing squirrel waveform provider in environment %s",
+                self.environment,
+            )
+            squirrel = Squirrel(
+                env=str(self.environment.expanduser()) if self.environment else None,
+                persistent=self.persistent,
+            )
             paths = []
             for path in self.waveform_dirs:
                 if "**" in str(path):
@@ -173,7 +178,7 @@ def get_squirrel(self) -> Squirrel:
             squirrel.add(paths, check=False)
             if self._stations:
                 for path in self._stations.station_xmls:
-                    logger.info("loading responses from %s", path)
+                    logger.info("loading StationXML responses from %s", path)
                     squirrel.add(str(path), check=False)
             self._squirrel = squirrel
         return self._squirrel