rename

notebooks/wp5/glacier_distribution.ipynb

@@ -77,7 +77,18 @@
     "    ds.to_pandas(),\n",
     "    geometry=gpd.points_from_xy(ds[\"CENLON\"], ds[\"CENLAT\"]),\n",
     "    crs=\"EPSG:4326\",\n",
-    ")"
+    ")\n",
+    "\n",
+    "# Convert dates from string to datetime, and add digitalization year\n",
+    "columns = []\n",
+    "for column in (\"BGNDATE\", \"ENDDATE\"):\n",
+    "    years = gdf[column].str[:4]\n",
+    "    months = gdf[column].str[4:6].replace(\"99\", \"01\")\n",
+    "    days = gdf[column].str[6:8].replace(\"99\", \"01\")\n",
+    "    series = years + months + days\n",
+    "    series = series.where(~series.str.startswith(\"-\"))\n",
+    "    columns.append(pd.to_datetime(series))\n",
+    "gdf[\"year\"] = pd.DataFrame(columns).mean().dt.year.astype(\"Int64\")"
    ]
   },
   {
@@ -180,12 +191,22 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "year = pd.DataFrame(\n",
-    "    [gdf[\"BGNDATE\"].str[:4].astype(int), gdf[\"ENDDATE\"].str[:4].astype(int)]\n",
-    ")\n",
-    "gdf[\"year\"] = year.where(year > 0).mean().round().astype(\"Int64\")\n",
     "size = gdf.set_index(\"year\").groupby(\"year\").size()\n",
-    "ax = size.plot.bar(figsize=(15, 5), grid=True, ylabel=\"Number of glaciers\")"
+    "size = size.reindex(range(gdf[\"year\"].min(), gdf[\"year\"].max()), fill_value=0)\n",
+    "missing = gdf[\"year\"].isnull().sum()\n",
+    "missing_perc = 100 * (missing / len(gdf[\"year\"]))\n",
+    "ax = size.plot.bar(\n",
+    "    figsize=(15, 5),\n",
+    "    grid=True,\n",
+    "    ylabel=\"Number of glaciers\",\n",
+    ")\n",
+    "_ = ax.text(\n",
+    "    0,\n",
+    "    1,\n",
+    "    f\"Date of digitization data are missing for {missing} glaciers or {missing_perc:.2f}% of the dataset.\",\n",
+    "    transform=ax.transAxes,\n",
+    "    bbox={\"facecolor\": \"white\", \"edgecolor\": \"black\"},\n",
+    ")"
    ]
   },
   {