updated schema

Author: irm-codebase

config/config.yaml

@@ -30,10 +30,10 @@ imputation:
         upper: 300000
       minimums:
         porosity_ratio: 0.1
-        mean_depth_m: 800
+        depth_m: 800
         reservoir_thickness_m: 10
         seal_thickness_m: 0
-        permeability_mD: 0
+        permeability_md: 0
       remove_w_remarks:
         surface_issues: False
         subsurface_issues: False
@@ -44,10 +44,10 @@ imputation:
         upper: 300000
       minimums:
         porosity_ratio: 0
-        mean_depth_m: 800
+        depth_m: 800
         reservoir_thickness_m: 10
         seal_thickness_m: 0
-        permeability_mD: 0
+        permeability_md: 0
       remove_w_remarks:
         surface_issues: False
         subsurface_issues: False
@@ -58,10 +58,10 @@ imputation:
         upper: 300000
       minimums:
         porosity_ratio: 0
-        mean_depth_m: 800
+        depth_m: 800
         reservoir_thickness_m: 10
         seal_thickness_m: 0
-        permeability_mD: 0
+        permeability_md: 0
       remove_w_remarks:
         surface_issues: False
         subsurface_issues: False

workflow/Snakefile

@@ -10,7 +10,7 @@ configfile: workflow.source_path("../config/config.yaml")
 
 
 # Validate the configuration using the schema file.
-# validate(config, workflow.source_path("internal/config.schema.yaml"))
+validate(config, workflow.source_path("internal/config.schema.yaml"))
 
 # Load internal settings separately so users cannot modify them.
 with open(workflow.source_path("internal/settings.yaml"), "r") as f:

workflow/internal/config.schema.yaml

@@ -11,8 +11,8 @@ $defs:
       - type: string
       - type: integer
 
-  nonNegativeInteger:
-    type: integer
+  nonNegativeNumber:
+    type: number
     minimum: 0
 
   boundsMtco2:
@@ -22,50 +22,60 @@ $defs:
     required: [lower, upper]
     properties:
       lower:
-        $ref: "#/$defs/nonNegativeInteger"
+        $ref: "#/$defs/nonNegativeNumber"
       upper:
-        $ref: "#/$defs/nonNegativeInteger"
+        $ref: "#/$defs/nonNegativeNumber"
 
-properties:
-  crs:
-    description: "CRS values to use during projections."
+  aggregatedConfig:
+    description: "Settings for each aggregated CDR group."
     type: object
     additionalProperties: false
-    required: [geographic, projected]
+    required: [bounds_mtco2]
     properties:
-      geographic:
-        description: "Geographic CRS (for lat/lon operations)."
-        allOf:
-          - $ref: "#/$defs/crsValue"
-      projected:
-        description: "Projected CRS (for area operations)."
+      bounds_mtco2:
+        description: "These bounds will be applied to each shape_id case after aggregation."
         allOf:
-          - $ref: "#/$defs/crsValue"
+          - $ref: "#/$defs/boundsMtco2"
 
-  imputation:
-    description: "These settings alter how data is processed."
+  co2stopConfig:
+    description: "Settings for each CDR group in the CO2Stop dataset."
     type: object
     additionalProperties: false
-    required: [filter, bounds_mtco2]
+    required: [bounds_mtco2, minimums, remove_w_remarks]
     properties:
       bounds_mtco2:
+        description: "These bounds will be applied to each row in the CO2Stop dataset."
+        allOf:
+          - $ref: "#/$defs/boundsMtco2"
+      minimums:
         description: |
-          Lower and upper bounds for the CO2Stop polygons.
-          These are applied before aggregation.
+          Minimum values for key reservoir attributes. Cases below them will be dropped.
+          IMPORTANT: CO2Stop uses '0' for missing values, making NaN and actual 0 indistinguishable.
+          Some of these settings may result in a lot of removed data due to this dataset limitation!
         type: object
-        required: [co2stop_polygons, aggregated_polygons]
+        additionalProperties: false
+        required: [porosity_ratio, depth_m, reservoir_thickness_m, seal_thickness_m, permeability_md]
         properties:
-          co2stop_polygons:
-            description: "These bounds will be applied to each row in the CO2Stop dataset."
-            allOf:
-              - $ref: "#/$defs/boundsMtco2"
-          aggregated_polygons:
-            description: "These bounds will be applied to each shape requested for aggregation."
-            allOf:
-              - $ref: "#/$defs/boundsMtco2"
-      filter:
+          porosity_ratio:
+            description: Minimum mean porosity ratio
+            type: number
+            minimum: 0
+            maximum: 1
+          depth_m:
+            description: Minimum mean depth in metres
+            $ref: "#/$defs/nonNegativeNumber"
+          reservoir_thickness_m:
+            description: Minimum mean reservoir thickness in metres
+            $ref: "#/$defs/nonNegativeNumber"
+          seal_thickness_m:
+            description: Minimum seal thickness in metres
+            $ref: "#/$defs/nonNegativeNumber"
+          permeability_md:
+            description: Permeability in millidarcy (md)
+            $ref: "#/$defs/nonNegativeNumber"
+      remove_w_remarks:
         description: |
-          If enabled, these will remove CO2Stop polygons in certain cases.
+          Remove CO2Stop polygons in certain cases.
           Note that CO2Stop settings and comments are the only thing used for these filters.
           Quality may vary due to the inconsistent nature of the dataset.
         type: object
@@ -83,3 +93,56 @@ properties:
               These are cases where the authors state that polygons do not reflect the actual extent of storage.
               RECOMMENDED TO LEAVE AS 'True'.
             type: boolean
+
+properties:
+  crs:
+    description: "CRS values to use during projections."
+    type: object
+    additionalProperties: false
+    required: [geographic, projected]
+    properties:
+      geographic:
+        description: "Geographic CRS (for lat/lon operations)."
+        allOf:
+          - $ref: "#/$defs/crsValue"
+      projected:
+        description: "Projected CRS (for area operations)."
+        allOf:
+          - $ref: "#/$defs/crsValue"
+
+  imputation:
+    description: "These settings alter how data is processed."
+    type: object
+    additionalProperties: false
+    required: [aggregated, co2stop]
+    properties:
+      aggregated:
+        description: Settings applied after aggregating to the provided shapes file.
+        type: object
+        additionalProperties: false
+        required: [aquifer, gas, oil]
+        properties:
+          aquifer:
+            allOf:
+              - $ref: "#/$defs/aggregatedConfig"
+          gas:
+            allOf:
+              - $ref: "#/$defs/aggregatedConfig"
+          oil:
+            allOf:
+              - $ref: "#/$defs/aggregatedConfig"
+      co2stop:
+        description: Settings applied while preparing / standardising the CO2Stop dataset.
+        type: object
+        additionalProperties: false
+        required: [aquifer, gas, oil]
+        properties:
+          aquifer:
+            allOf:
+              - $ref: "#/$defs/co2stopConfig"
+          gas:
+            allOf:
+              - $ref: "#/$defs/co2stopConfig"
+          oil:
+            allOf:
+              - $ref: "#/$defs/co2stopConfig"

workflow/scripts/prepare_co2stop.py

@@ -25,10 +25,10 @@
 # Translate readable config names to CO2Stop columns
 MINIMUMS_CO2STOP = {
     "porosity_ratio": "POROSITY_MEAN",
-    "mean_depth_m": "DEPTH_MEAN",
+    "depth_m": "DEPTH_MEAN",
     "reservoir_thickness_m": "GROSS_THICK_MEAN",
     "seal_thickness_m": "MIN_SEAL_THICK",
-    "permeability_mD": "PERM_MEAN"
+    "permeability_md": "PERM_MEAN"
 }