Update: filter configuration

Author: heptaflar

kaleidoscope/config/config.collect.json

@@ -7,43 +7,65 @@
   "dic": {
     "attrs": {
       "long_name": "standard uncertainty of Dissolved Inorganic Carbon in seawater"
+    },
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 1.0
     }
   },
   "DOC": {
     "attrs": {
       "long_name": "standard uncertainty of estimated DOC"
     },
-    "filter": true
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 4.0
+    }
   },
   "PIC": {
     "attrs": {
       "long_name": "standard uncertainty of PIC"
     },
-    "filter": true
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 4.0
+    }
   },
   "POC": {
     "attrs": {
       "long_name": "standard uncertainty of POC"
     },
-    "filter": true
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 4.0
+    }
   },
   "pc": {
     "attrs": {
       "long_name": "standard uncertainty of Phytoplankton Carbon concentration in seawater"
     },
-    "filter": true
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 4.0
+    }
   },
   "pp": {
     "attrs": {
       "long_name": "standard uncertainty of Phytoplankton Primary Production"
     },
-    "filter": true
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 4.0
+    }
   },
   "Predicted_DOC": {
     "attrs": {
       "long_name": "standard uncertainty of Predicted Dissolved Organic Carbon",
       "description": "standard uncertainty of Predicted DOC values"
     },
-    "filter": true
+    "filter": {
+      "kind": "Gaussian",
+      "fwhm": 4.0
+    }
   }
 }

kaleidoscope/config/config.writer.json

@@ -19,7 +19,7 @@
       "time": 12,
       "latitude": 0,
       "longitude": 0,
-      "depth": 0
+      "depth": 13
     },
     "esa-scope-doc": {
       "lat": 720,

kaleidoscope/operators/collectop.py

@@ -22,18 +22,23 @@
 from ..logger import get_logger
 
 
-def _filter(x: da.Array, dims: tuple) -> da.Array:
+def _filter(x: da.Array, dims: tuple, filter_config: dict) -> da.Array:
     """
     Applies a lateral low pass filter to suppress statistical
     fluctuations caused by finite sampling of error probability
     density functions.
     """
     return Gaussian(dtype=x.dtype, m=x.ndim, n=x.ndim).apply_to(
-        x, dims=dims, fwhm=4.0
+        x, dims=dims, fwhm=filter_config.get("fwhm", 1.0)
     )
 
 
-def _std(x: da.Array, dims: tuple, filtered: bool = False) -> da.Array:
+def _std(
+    x: da.Array,
+    dims: tuple,
+    filtered: bool = False,
+    filter_config: dict | None = None,
+) -> da.Array:
     """
     Returns the standard deviation of simulated errors.
 
@@ -44,7 +49,7 @@ def _std(x: da.Array, dims: tuple, filtered: bool = False) -> da.Array:
     Monte Carlo method.
     """
     return da.sqrt(
-        _filter(_mse(x[1:] - x[:1]), dims)
+        _filter(_mse(x[1:] - x[:1]), dims, filter_config)
         if filtered
         else _mse(x[1:] - x[:1])
     )
@@ -110,7 +115,7 @@ def run(self, source: Dataset) -> Dataset:  # noqa: D102
             if v not in config:
                 continue
             self.add_uncertainty(target, source, v, x)
-            if config[v].get("filter", False):
+            if config[v].get("filter", {}):
                 self.add_uncertainty(target, source, v, x, filtered=True)
         return target
 
@@ -131,7 +136,12 @@ def add_uncertainty(self, target, source, v, x, filtered: bool = False):
         if v_unc in target:
             return
         get_logger().info(f"starting graph for variable: {v_unc}")
-        x_unc = _std(decode(source[v].data, x.attrs), x.dims, filtered)
+        x_unc = _std(
+            decode(source[v].data, x.attrs),
+            x.dims,
+            filtered,
+            config[v].get("filter", {}),
+        )
         get_logger().info(f"finished graph for variable: {v_unc}")
         target[v_unc] = DataArray(
             data=encode(x_unc, x.attrs, x.dtype),