Upload 2 files

Russia_protected_areas_simplified.geoparquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aab10cc608c079b112df3655efb9694f891b6a6040a1fc11a91f0876b1af2c60
-size 7376561
+oid sha256:700d2be037950da938063c0c1f9fe3f4ecbfd872713c0f2f2e45e41cbe5d16e3
+size 3698919

app.py

@@ -3,133 +3,190 @@ import geopandas as gpd
 import plotly.express as px
 import plotly.graph_objects as go
 import pandas as pd
+from shapely.geometry import MultiPolygon
 
-# IUCN Category information (descriptions only, colors from Plotly)
+# ──────────────────────────────────────────────────────────────
+# IUCN Category reference information
+# ──────────────────────────────────────────────────────────────
 IUCN_CATEGORIES = {
-    'Ia': {
-        'name': 'Strict Nature Reserve',
-        'description': 'Strictly protected for biodiversity, geological/geomorphological features, and scientific research.'
+    "Ia": {
+        "name": "Strict Nature Reserve",
+        "description": (
+            "Strictly protected for biodiversity, geological/geomorphological"
+            " features, and scientific research."
+        ),
     },
-    'Ib': {
-        'name': 'Wilderness Area',
-        'description': 'Large, unmodified areas retaining natural character, without permanent habitation, managed to preserve natural condition.'
+    "Ib": {
+        "name": "Wilderness Area",
+        "description": (
+            "Large, unmodified areas retaining natural character, without"
+            " permanent habitation, managed to preserve natural condition."
+        ),
     },
-    'II': {
-        'name': 'National Park',
-        'description': 'Large natural/near-natural areas protecting large-scale ecological processes and species, while allowing compatible spiritual/recreational use.'
+    "II": {
+        "name": "National Park",
+        "description": (
+            "Large natural/near-natural areas protecting large-scale ecological"
+            " processes and species, while allowing compatible spiritual/"
+            "recreational use."
+        ),
     },
-    'III': {
-        'name': 'Natural Monument/Feature',
-        'description': 'Set aside to protect a specific natural monument, landform, sea mount, or geological feature.'
+    "III": {
+        "name": "Natural Monument/Feature",
+        "description": (
+            "Set aside to protect a specific natural monument, landform, sea"
+            " mount, or geological feature."
+        ),
     },
-    'IV': {
-        'name': 'Habitat/Species Management Area',
-        'description': 'Protects particular species or habitats, often requiring regular, active management interventions.'
+    "IV": {
+        "name": "Habitat/Species Management Area",
+        "description": (
+            "Protects particular species or habitats, often requiring regular,"
+            " active management interventions."
+        ),
+    },
+    "V": {
+        "name": "Protected Landscape/Seascape",
+        "description": (
+            "Protects areas where the interaction of people and nature over"
+            " time has produced a distinct character."
+        ),
     },
-    'V': {
-        'name': 'Protected Landscape/Seascape',
-        'description': 'Protects areas where the interaction of people and nature over time has produced a distinct character.'
+    "VI": {
+        "name": "Sustainable Use Area",
+        "description": (
+            "Conserves ecosystems and habitats together with associated"
+            " cultural values and traditional natural resource management"
+            " systems."
+        ),
     },
-    'VI': {
-        'name': 'Sustainable Use Area',
-        'description': 'Conserves ecosystems and habitats together with associated cultural values and traditional natural resource management systems.'
+    "Not Applicable": {
+        "name": "Not Applicable",
+        "description": "IUCN category is not applicable to this protected area.",
     },
-    'Not Applicable': {
-        'name': 'Not Applicable',
-        'description': 'IUCN category is not applicable to this protected area.'
+    "Not Assigned": {
+        "name": "Not Assigned",
+        "description": "IUCN category has not been assigned.",
     },
-    'Not Assigned': {
-        'name': 'Not Assigned',
-        'description': 'IUCN category has not been assigned.'
+    "Not Reported": {
+        "name": "Not Reported",
+        "description": "IUCN category has not been reported.",
     },
-    'Not Reported': {
-        'name': 'Not Reported',
-        'description': 'IUCN category has not been reported.'
-    }
 }
 
+
+# ──────────────────────────────────────────────────────────────
+# Helper: keep only the largest polygon from a MultiPolygon
+# ──────────────────────────────────────────────────────────────
+def largest_polygon(geom):
+    """Return the single largest polygon from a MultiPolygon."""
+    if isinstance(geom, MultiPolygon):
+        return max(geom.geoms, key=lambda g: g.area)
+    return geom
+
+
+# ──────────────────────────────────────────────────────────────
 # Load data
+# ───────────��──────────────────────────────────────────────────
 try:
-    # Load protected areas
-    protected_gdf = gpd.read_parquet('Russia_protected_areas_simplified.geoparquet')
+    protected_gdf = gpd.read_parquet(
+        "Russia_protected_areas_simplified.geoparquet"
+    )
     protected_gdf = protected_gdf.to_crs(epsg=4326)
     print(f"Loaded {len(protected_gdf)} protected areas")
-    
-    # Ensure IUCN_CAT is string type and handle missing values
-    if 'IUCN_CAT' in protected_gdf.columns:
-        protected_gdf['IUCN_CAT'] = protected_gdf['IUCN_CAT'].fillna('Not Reported').astype(str)
-        protected_gdf['IUCN_CAT'] = protected_gdf['IUCN_CAT'].apply(
-            lambda x: x if x in IUCN_CATEGORIES else 'Not Reported'
+
+    # Extra safety: ensure no MultiPolygons remain at runtime
+    protected_gdf["geometry"] = protected_gdf.geometry.apply(largest_polygon)
+
+    # Normalise IUCN_CAT
+    if "IUCN_CAT" in protected_gdf.columns:
+        protected_gdf["IUCN_CAT"] = (
+            protected_gdf["IUCN_CAT"]
+            .fillna("Not Reported")
+            .astype(str)
+            .apply(lambda x: x if x in IUCN_CATEGORIES else "Not Reported")
         )
-    
+
     # Keep only needed columns
-    columns_to_keep = ['NAME', 'DESIG', 'DESIG_ENG', 'DESIG_TYPE', 'IUCN_CAT', 'geometry']
-    available_columns = [col for col in columns_to_keep if col in protected_gdf.columns]
+    columns_to_keep = [
+        "NAME", "DESIG", "DESIG_ENG", "DESIG_TYPE", "IUCN_CAT", "geometry",
+    ]
+    available_columns = [
+        c for c in columns_to_keep if c in protected_gdf.columns
+    ]
     protected_gdf = protected_gdf[available_columns].copy()
-    
+
     print(f"Columns retained: {', '.join(protected_gdf.columns.tolist())}")
     print("\nIUCN Category distribution:")
-    print(protected_gdf['IUCN_CAT'].value_counts().sort_index())
-    
+    print(protected_gdf["IUCN_CAT"].value_counts().sort_index())
+
 except Exception as e:
     print(f"Error loading protected areas: {e}")
     protected_gdf = None
 
+
+# ──────────────────────────────────────────────────────────────
+# Helpers
+# ──────────────────────────────────────────────────────────────
 def format_value(value):
-    """Format value for tooltip"""
-    if pd.isna(value) or value == '' or value == 'None':
+    """Format a value for the hover tooltip."""
+    if pd.isna(value) or value in ("", "None"):
         return "N/A"
     return str(value)
 
+
 def create_legend_table():
-    """Create a DataFrame for the IUCN category legend"""
-    legend_data = []
-    for cat, info in IUCN_CATEGORIES.items():
-        legend_data.append({
-            'Category': cat,
-            'Name': info['name'],
-            'Primary Objective': info['description']
-        })
-    return pd.DataFrame(legend_data)
+    """Return a DataFrame describing each IUCN category."""
+    rows = [
+        {
+            "Category": cat,
+            "Name": info["name"],
+            "Primary Objective": info["description"],
+        }
+        for cat, info in IUCN_CATEGORIES.items()
+    ]
+    return pd.DataFrame(rows)
 
+
+# ──────────────────────────────────────────────────────────────
+# Map
+# ──────────────────────────────────────────────────────────────
 def create_map():
-    """Create the interactive map with protected areas using satellite style"""
-    
+    """Create the interactive Plotly choropleth map."""
+
     if protected_gdf is None:
         fig = go.Figure()
         fig.add_annotation(
             text="Error: Could not load data files",
             xref="paper", yref="paper",
-            x=0.5, y=0.5, showarrow=False
+            x=0.5, y=0.5, showarrow=False,
         )
         return fig
-    
-    # Build customdata columns directly in the dataframe
-    # so that px.choropleth_map keeps them aligned per-trace
-    protected_gdf_plot = protected_gdf.copy()
-    protected_gdf_plot['tooltip_name'] = protected_gdf_plot['NAME'].apply(format_value)
-    protected_gdf_plot['tooltip_desig'] = protected_gdf_plot['DESIG'].apply(format_value)
-    protected_gdf_plot['tooltip_desig_eng'] = protected_gdf_plot['DESIG_ENG'].apply(format_value)
-    protected_gdf_plot['tooltip_desig_type'] = protected_gdf_plot['DESIG_TYPE'].apply(format_value)
-    protected_gdf_plot['tooltip_iucn'] = protected_gdf_plot['IUCN_CAT'].apply(format_value)
-    
-    # Use hover_data to let Plotly handle per-trace customdata alignment
+
+    plot_gdf = protected_gdf.copy()
+    plot_gdf["tooltip_name"] = plot_gdf["NAME"].apply(format_value)
+    plot_gdf["tooltip_desig"] = plot_gdf["DESIG"].apply(format_value)
+    plot_gdf["tooltip_desig_eng"] = plot_gdf["DESIG_ENG"].apply(format_value)
+    plot_gdf["tooltip_desig_type"] = plot_gdf["DESIG_TYPE"].apply(format_value)
+    plot_gdf["tooltip_iucn"] = plot_gdf["IUCN_CAT"].apply(format_value)
+
     fig = px.choropleth_map(
-        protected_gdf_plot,
-        geojson=protected_gdf_plot.geometry,
-        locations=protected_gdf_plot.index,
-        color='IUCN_CAT',
+        plot_gdf,
+        geojson=plot_gdf.geometry,
+        locations=plot_gdf.index,
+        color="IUCN_CAT",
         color_discrete_sequence=px.colors.qualitative.Prism,
-        custom_data=['tooltip_name', 'tooltip_desig', 'tooltip_desig_eng',
-                     'tooltip_desig_type', 'tooltip_iucn'],
+        custom_data=[
+            "tooltip_name", "tooltip_desig", "tooltip_desig_eng",
+            "tooltip_desig_type", "tooltip_iucn",
+        ],
         map_style="satellite-streets",
         zoom=2,
         center={"lat": 60, "lon": 90},
         opacity=0.65,
-        labels={'IUCN_CAT': 'IUCN Category'}
+        labels={"IUCN_CAT": "IUCN Category"},
     )
-    
-    # Now update_traces is safe because Plotly already split customdata per trace
+
     fig.update_traces(
         hovertemplate=(
             "<b>Name:</b> %{customdata[0]}<br>"
@@ -139,84 +196,87 @@ def create_map():
             "<b>IUCN Category:</b> %{customdata[4]}"
             "<extra></extra>"
         ),
-        marker_line_width=0
+        marker_line_width=0,
     )
-    
+
     fig.update_layout(
         margin={"r": 0, "t": 0, "l": 0, "b": 0},
         height=800,
         legend=dict(
-            title=dict(text='IUCN Category'),
-            yanchor="top",
-            y=0.99,
-            xanchor="left",
-            x=0.01,
-            bgcolor='rgba(255, 255, 255, 0.9)',
-            bordercolor='rgba(0, 0, 0, 0.2)',
-            borderwidth=1
+            title=dict(text="IUCN Category"),
+            yanchor="top", y=0.99,
+            xanchor="left", x=0.01,
+            bgcolor="rgba(255, 255, 255, 0.9)",
+            bordercolor="rgba(0, 0, 0, 0.2)",
+            borderwidth=1,
         ),
-        uirevision='constant'
+        uirevision="constant",
     )
-    
+
     return fig
 
-# Build Gradio UI
+
+# ──────────────────────────────────────────────────────────────
+# Gradio UI
+# ──────────────────────────────────────────────────────────────
 with gr.Blocks(title="Russia Protected Areas") as demo:
     gr.Markdown("# Protected Areas of Russia")
-    gr.Markdown("Explore protected areas in Russia colored by their IUCN conservation category.")
-    
+    gr.Markdown(
+        "Explore protected areas in Russia colored by their IUCN"
+        " conservation category."
+    )
+
     with gr.Row():
         map_plot = gr.Plot(label="Interactive Map", value=create_map())
-    
+
     with gr.Row():
         gr.Markdown("## IUCN Protected Area Categories")
-    
+
     with gr.Row():
         legend_table = gr.DataFrame(
             value=create_legend_table(),
             label="Category Definitions",
             interactive=False,
-            wrap=True
+            wrap=True,
         )
-    
+
     # Statistics section
     if protected_gdf is not None:
         with gr.Row():
             gr.Markdown("## Statistics")
-        
+
         with gr.Row():
             total_areas = len(protected_gdf)
-            
             stats_data = []
-            for cat in list(IUCN_CATEGORIES.keys()):
-                count = len(protected_gdf[protected_gdf['IUCN_CAT'] == cat])
+            for cat in IUCN_CATEGORIES:
+                count = len(protected_gdf[protected_gdf["IUCN_CAT"] == cat])
                 if count > 0:
                     percentage = (count / total_areas) * 100
-                    stats_data.append({
-                        'IUCN Category': cat,
-                        'Number of Areas': f"{count:,}",
-                        'Percentage': f"{percentage:.1f}%"
-                    })
-            
-            stats_df = pd.DataFrame(stats_data)
-            
+                    stats_data.append(
+                        {
+                            "IUCN Category": cat,
+                            "Number of Areas": f"{count:,}",
+                            "Percentage": f"{percentage:.1f}%",
+                        }
+                    )
+
             gr.DataFrame(
-                value=stats_df,
+                value=pd.DataFrame(stats_data),
                 label=f"Distribution of {total_areas:,} Protected Areas",
-                interactive=False
+                interactive=False,
             )
-    
-    gr.Markdown("""
+
+    gr.Markdown(
+        """
     ---
     ### Data Sources & Attribution
+
+    **Protected Areas Data:** [Protected Planet - WDPA](https://www.protectedplanet.net/)
     
-    **Protected Areas Data:** [Protected Planet - WDPA](https://www.protectedplanet.net/)  
-    **Administrative Boundaries:** Satellite map streets layer  
     **IUCN Categories:** [IUCN Protected Area Categories System](https://www.iucn.org/theme/protected-areas/about/protected-area-categories)
-    
-    ---
-    *Note: The map shows protected area polygons only. Points and lines from the WDPA dataset are not included.*
-    """)
+
+    """
+    )
 
 if __name__ == "__main__":
     demo.launch()