update

Author: spatialthoughts

notebooks/geopandas_flood_frequncy.ipynb

@@ -313,8 +313,9 @@
     "    'date_diff_days': date_diff_days,\n",
     "})\n",
     "\n",
-    "ndays = 15 # This will be our threshold for \"nearby\" dates in the next step\n",
-    "# Filter to pairs whose dates are within ±15 days of each other\n",
+    "ndays = 7 # This will be our threshold for \"nearby\" dates in the next step\n",
+    "\n",
+    "# Filter to pairs whose dates are within ndays of each other\n",
     "nearby = pairs[pairs['date_diff_days'] <= ndays].reset_index(drop=True)\n",
     "print(f'Intersecting polygon pairs with dates within ±{ndays} days: {len(nearby):>10,}')\n",
     "nearby.head(10)"
@@ -363,12 +364,32 @@
     "gf"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "1a7b15e6",
+   "metadata": {},
+   "source": [
+    "## Aggregate Flood Events to Grid\n",
+    "\n",
+    "We now count the number of unique flood events per grid cell."
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "c56357b6",
    "metadata": {},
    "source": [
-    "Reproject the filtered flood points to EPSG:7755 — done once as a shared step before comparing both counting approaches."
+    "Reproject the filtered flood points to the chosen CRS."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "76c29140",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "gf_reprojected = gf.to_crs(grid_crs)"
    ]
   },
   {
@@ -386,7 +407,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "gf_reprojected = gf.to_crs(grid_crs)\n",
     "joined = gpd.sjoin(grid_india, gf_reprojected, how='inner', predicate='intersects')\n",
     "joined"
    ]
@@ -398,11 +418,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "counts_a = joined.groupby(joined.index)['flood_event'].nunique().rename('flood_count')\n",
+    "counts = joined.groupby(joined.index)['flood_event'].nunique().rename('flood_count')\n",
     "\n",
-    "grid_india_a = grid_india.copy()\n",
-    "grid_india_a['flood_count'] = grid_india_a.index.map(counts_a).fillna(0).astype(int)\n",
-    "print(grid_india_a['flood_count'].describe())"
+    "grid_country['flood_count'] = grid_country.index.map(counts).fillna(0).astype(int)\n",
+    "grid_country.head()"
    ]
   },
   {
@@ -415,31 +434,28 @@
     "from matplotlib.cm import ScalarMappable\n",
     "from matplotlib.patches import Patch\n",
     "\n",
-    "plot_data = grid_india_a.copy()\n",
-    "plot_data['flood_count'] = plot_data['flood_count'].fillna(0).astype(int)\n",
-    "\n",
     "fig, ax = plt.subplots(figsize=(10, 12))\n",
     "\n",
     "# Cells with no observations in neutral grey\n",
-    "plot_data[plot_data['flood_count'] == 0].plot(\n",
+    "grid_country[grid_country['flood_count'] == 0].plot(\n",
     "    ax=ax, color='#d9d9d9', linewidth=0\n",
     ")\n",
     "\n",
-    "# Cells with observations coloured on a log scale\n",
+    "# Cells with observations coloured by flood count\n",
     "# (flood counts are heavily right-skewed; log scale reveals spatial variation)\n",
-    "non_zero = plot_data[plot_data['flood_count'] > 0]\n",
-    "vmax = non_zero['flood_count'].max()\n",
+    "norm = mcolors.LogNorm(vmin=1, vmax=grid_country['flood_count'].max())\n",
     "cmap = 'YlOrRd'\n",
     "\n",
+    "non_zero = grid_country[grid_country['flood_count'] > 0]\n",
     "non_zero.plot(\n",
     "    ax=ax, column='flood_count',\n",
-    "    cmap=cmap, vmin=1, vmax=50, linewidth=0\n",
+    "    cmap=cmap, norm=norm, linewidth=0\n",
     ")\n",
     "\n",
     "# Country boundary\n",
     "country_proj.plot(ax=ax, facecolor='none', edgecolor='#333333', linewidth=0.8)\n",
     "\n",
-    "ax.set_title('Flood Risk Map — India (2000–2025)\\n10 km × 10 km grid, observation count per cell', fontsize=13, pad=12)\n",
+    "ax.set_title('Flood Frequency Map (2000–2025)', fontsize=13, pad=12)\n",
     "ax.axis('off')\n",
     "plt.tight_layout()\n",
     "plt.show()"