UCL · RachelMurray-Watson · Jan 29, 2026 · Jan 29, 2026 · Jan 29, 2026 · Jan 29, 2026
diff --git a/...eFile_Precipitation_Disruptions_ssp126_highest_monthly_prediction_weather_by_facility.csv b/...eFile_Precipitation_Disruptions_ssp126_highest_monthly_prediction_weather_by_facility.csv
diff --git a/...ceFile_Precipitation_Disruptions_ssp126_highest_window_prediction_weather_by_facility.csv b/...ceFile_Precipitation_Disruptions_ssp126_highest_window_prediction_weather_by_facility.csv
diff --git a/...ceFile_Precipitation_Disruptions_ssp126_lowest_monthly_prediction_weather_by_facility.csv b/...ceFile_Precipitation_Disruptions_ssp126_lowest_monthly_prediction_weather_by_facility.csv
diff --git a/...urceFile_Precipitation_Disruptions_ssp126_mean_monthly_prediction_weather_by_facility.csv b/...urceFile_Precipitation_Disruptions_ssp126_mean_monthly_prediction_weather_by_facility.csv
diff --git a/...eFile_Precipitation_Disruptions_ssp245_highest_monthly_prediction_weather_by_facility.csv b/...eFile_Precipitation_Disruptions_ssp245_highest_monthly_prediction_weather_by_facility.csv
diff --git a/...ceFile_Precipitation_Disruptions_ssp245_highest_window_prediction_weather_by_facility.csv b/...ceFile_Precipitation_Disruptions_ssp245_highest_window_prediction_weather_by_facility.csv
diff --git a/...ceFile_Precipitation_Disruptions_ssp245_lowest_monthly_prediction_weather_by_facility.csv b/...ceFile_Precipitation_Disruptions_ssp245_lowest_monthly_prediction_weather_by_facility.csv
diff --git a/...rceFile_Precipitation_Disruptions_ssp245_lowest_window_prediction_weather_by_facility.csv b/...rceFile_Precipitation_Disruptions_ssp245_lowest_window_prediction_weather_by_facility.csv
diff --git a/...urceFile_Precipitation_Disruptions_ssp245_mean_monthly_prediction_weather_by_facility.csv b/...urceFile_Precipitation_Disruptions_ssp245_mean_monthly_prediction_weather_by_facility.csv
diff --git a/...ourceFile_Precipitation_Disruptions_ssp245_mean_window_prediction_weather_by_facility.csv b/...ourceFile_Precipitation_Disruptions_ssp245_mean_window_prediction_weather_by_facility.csv
diff --git a/...eFile_Precipitation_Disruptions_ssp585_highest_monthly_prediction_weather_by_facility.csv b/...eFile_Precipitation_Disruptions_ssp585_highest_monthly_prediction_weather_by_facility.csv
diff --git a/...ceFile_Precipitation_Disruptions_ssp585_highest_window_prediction_weather_by_facility.csv b/...ceFile_Precipitation_Disruptions_ssp585_highest_window_prediction_weather_by_facility.csv
diff --git a/...ceFile_Precipitation_Disruptions_ssp585_lowest_monthly_prediction_weather_by_facility.csv b/...ceFile_Precipitation_Disruptions_ssp585_lowest_monthly_prediction_weather_by_facility.csv
diff --git a/...rceFile_Precipitation_Disruptions_ssp585_lowest_window_prediction_weather_by_facility.csv b/...rceFile_Precipitation_Disruptions_ssp585_lowest_window_prediction_weather_by_facility.csv
diff --git a/...urceFile_Precipitation_Disruptions_ssp585_mean_monthly_prediction_weather_by_facility.csv b/...urceFile_Precipitation_Disruptions_ssp585_mean_monthly_prediction_weather_by_facility.csv
diff --git a/...ourceFile_Precipitation_Disruptions_ssp585_mean_window_prediction_weather_by_facility.csv b/...ourceFile_Precipitation_Disruptions_ssp585_mean_window_prediction_weather_by_facility.csv
diff --git a/resources/ResourceFile_WeatherDisruption/parameter_values.csv b/resources/ResourceFile_WeatherDisruption/parameter_values.csv
@@ -0,0 +1,37 @@
+parameter_name,value,param_label,prior_min,prior_max,prior_note,reference
+climate_ssp,ssp245,scenario,NA,NA,Structural assumption,
+climate_model_ensemble_model,mean,scenario,NA,NA,Structural assumption,
+year_effective_climate_disruptions,2025,scenario,NA,NA,Structural assumption,
+services_affected_precip,none,scenario,NA,NA,Structural assumption,
+delay_in_seeking_care_weather,28,undetermined,0,,Structural assumption,"None, uncertainty analysis"
+scale_factor_reseeking_healthcare_post_disruption,1,undetermined,0,,Structural assumption,
+scale_factor_prob_disruption,1,undetermined,0,,Structural assumption,
+scale_factor_severity_disruption_and_delay,1,undetermined,0,,Structural assumption,
+prop_supply_side_disruptions,0.3,undetermined,0,,Structural assumption,
+baseline_coef_year,0.002102417,local,0.00205256,0.002152274,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_month,0.0047,local,,,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_altitude,0.0000551,local,1.28E-06,0.00010902,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_min_distance,3.023354283,local,2.760548389,3.286160177,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_urban,1.302601059,local,1.255023345,1.350178772,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_central_west,0.12713551,local,0.072699365,0.181571655,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_northern,-0.582897987,local,-0.63408374,-0.531712235,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_south_east,0.108568778,local,0.054119891,0.163017665,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_south_west,0.047290058,local,-0.012948704,0.107528819,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_government,0.263573318,local,0.197132521,0.330014115,95% confidence interval,"Murray-Watson et al, 2025'"
+baseline_coef_private,-1.83913218,local,-2.011993238,-1.666271123,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_year,0.002138682,local,0.002092911,0.002184453,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_month,0.0048,local,,,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_altitude,0.0000402,local,-1.08E-05,9.12E-05,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_min_distance,3.013236002,local,2.752061982,3.274410021,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_urban,1.300494948,local,1.252983312,1.348006584,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_central_west,0.10664008,local,0.058978636,0.154301524,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_northern,-0.584470779,local,-0.629405521,-0.539536038,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_south_east,0.08240437,local,0.040770555,0.124038185,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_south_west,0,local,-0.012948704,0.107528819,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_government,0.261964142,local,0.195516568,0.328411715,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_private,-1.820902432,local,-1.992445895,-1.649358968,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_precip_monthly,-0.000439395,local,-0.000713725,-0.000165064,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_precip_5day,0.000551989,local,-6.46E-05,0.001168602,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_lag_4month,-0.000148841,local,-0.000281277,-1.64E-05,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_lag_9month,-0.000330814,local,-0.000670456,8.83E-06,95% confidence interval,"Murray-Watson et al, 2025'"
+precipitation_coef_lag_1_5day,0.000322579,local,3.18E-06,0.000641975,95% confidence interval,"Murray-Watson et al, 2025'"
diff --git a/resources/demography/ResourceFile_Facility_Characteristics.csv b/resources/demography/ResourceFile_Facility_Characteristics.csv
diff --git a/resources/demography/mwi_pd_2020_1km_ASCII_XYZ_worldpop.txt b/resources/demography/mwi_pd_2020_1km_ASCII_XYZ_worldpop.txt
diff --git a/resources/demography/worldpop_density_with_districts.dbf b/resources/demography/worldpop_density_with_districts.dbf
diff --git a/resources/demography/worldpop_density_with_districts.prj b/resources/demography/worldpop_density_with_districts.prj
diff --git a/resources/demography/worldpop_density_with_districts.shp b/resources/demography/worldpop_density_with_districts.shp
diff --git a/resources/demography/worldpop_density_with_districts.shx b/resources/demography/worldpop_density_with_districts.shx
diff --git a/src/scripts/weather_disruptions/local_run_analysis.py b/src/scripts/weather_disruptions/local_run_analysis.py
@@ -0,0 +1,276 @@
+"""
+run_weather_disruptions_analysis.py
+
+A single-run simulation script for the WeatherDisruptions module.
+Usage:
+    python run_weather_disruptions_analysis.py
+"""
+
+from pathlib import Path
+
+import matplotlib.dates as mdates
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+
+from tlo import Date, Simulation, logging
+from tlo.analysis.utils import parse_log_file
+from tlo.methods.fullmodel import fullmodel
+
+#  CONFIGURATION
+
+seed = 0
+
+log_config = {
+    "filename": "weather_disruptions_analysis",
+    "directory": "./outputs",
+    "custom_levels": {
+        "*": logging.WARNING,
+        "tlo.methods.weather_disruptions": logging.INFO,
+        "tlo.methods.weather_disruptions.summary": logging.INFO,
+        "tlo.methods.healthsystem.summary": logging.INFO,
+        "tlo.methods.demography": logging.CRITICAL,
+        "tlo.methods.healthsystem": logging.CRITICAL,
+    },
+}
+
+start_date = Date(2010, 1, 1)
+end_date = Date(2030, 1, 1)
+pop_size = 5_000
+
+resourcefilepath = Path("./resources")
+
+#  COLOURS
+
+COLOUR_DELAYED = "#CC7000"
+COLOUR_CANCELLED = "#A8102E"
+COLOUR_TOTAL = "#1E3A8A"
+COLOUR_RATE = "#5b3f8c"
+
+#  SIMULATION SETUP
+
+sim = Simulation(start_date=start_date, seed=seed, log_config=log_config, resourcefilepath=resourcefilepath)
+
+sim.register(
+    *fullmodel(
+        module_kwargs={
+            "WeatherDisruptions": {
+                "climate_ssp": "ssp245",
+                "climate_model_ensemble_model": "mean",
+                "year_effective_climate_disruptions": 2025,
+                "services_affected_precip": "all",
+                "scale_factor_prob_disruption": 10.0,
+                "delay_in_seeking_care_weather": 28.0,
+                "scale_factor_reseeking_healthcare_post_disruption": 1.0,
+                "scale_factor_appointment_urgency": 1.0,
+                "scale_factor_severity_disruption_and_delay": 1.0,
+                "prop_supply_side_disruptions": 0.5,
+            }
+        }
+    )
+)
+
+sim.make_initial_population(n=pop_size)
+sim.simulate(end_date=end_date)
+
+#  PARSE LOGS
+
+log_df = parse_log_file(sim.log_filepath)
+
+# Monthly summary from the WeatherDisruptions logger
+monthly = log_df["tlo.methods.weather_disruptions.summary"]["weather_disruptions_monthly"].copy()
+monthly["date"] = pd.to_datetime(monthly["date"])
+monthly = monthly.sort_values("date").set_index("date")
+
+# Individual event logs (one row per disrupted HSI)
+delayed_events = log_df["tlo.methods.weather_disruptions.summary"].get(
+    "Weather_delayed_HSI_Event_full_info", pd.DataFrame()
+)
+cancelled_events = log_df["tlo.methods.weather_disruptions.summary"].get(
+    "Weather_cancelled_HSI_Event_full_info", pd.DataFrame()
+)
+
+#  DERIVED QUANTITIES
+
+monthly["total_disrupted"] = monthly["cancelled"] + monthly["delayed"]
+monthly["disruption_rate"] = (
+    monthly["total_disrupted"] / (monthly["hsi_total"] + monthly["total_disrupted"])
+).clip(upper=1.0)
+
+cumulative_cancelled = monthly["cancelled"].cumsum()
+cumulative_delayed = monthly["delayed"].cumsum()
+cumulative_total = monthly["total_disrupted"].cumsum()
+
+# Annual rollup
+annual = monthly.resample("YE").agg(
+    hsi_total=("hsi_total", "sum"),
+    cancelled=("cancelled", "sum"),
+    delayed=("delayed", "sum"),
+    total_disrupted=("total_disrupted", "sum"),
+)
+annual["disruption_rate"] = (
+    annual["total_disrupted"] / (annual["hsi_total"] + annual["total_disrupted"])
+).clip(upper=1.0)
+
+# Per-treatment-type breakdown from event logs (top N most common)
+TOP_N = 15
+
+
+def _count_by_treatment(df):
+    if df.empty or "TREATMENT_ID" not in df.columns:
+        return pd.Series(dtype=float)
+    return df["TREATMENT_ID"].value_counts()
+
+
+cancelled_by_type = _count_by_treatment(cancelled_events)
+delayed_by_type = _count_by_treatment(delayed_events)
+
+all_types = cancelled_by_type.index.union(delayed_by_type.index)
+by_type = pd.DataFrame({
+    "cancelled": cancelled_by_type.reindex(all_types, fill_value=0),
+    "delayed": delayed_by_type.reindex(all_types, fill_value=0),
+})
+by_type["total"] = by_type["cancelled"] + by_type["delayed"]
+top_types = by_type.nlargest(TOP_N, "total")
+
+#  PLOT 1: Monthly counts — cancelled, delayed, total
+
+fig1, ax1 = plt.subplots(figsize=(12, 5))
+
+ax1.plot(monthly.index, monthly["delayed"], color=COLOUR_DELAYED, lw=1.5, alpha=0.8, label="Delayed")
+ax1.plot(monthly.index, monthly["cancelled"], color=COLOUR_CANCELLED, lw=1.5, alpha=0.8, label="Cancelled")
+ax1.plot(monthly.index, monthly["total_disrupted"], color=COLOUR_TOTAL, lw=2.5, label="Total disrupted")
+
+ax1.xaxis.set_major_formatter(mdates.DateFormatter("%Y-%m"))
+ax1.xaxis.set_major_locator(mdates.MonthLocator(interval=6))
+plt.setp(ax1.xaxis.get_majorticklabels(), rotation=45, ha="right")
+ax1.set_xlabel("Month", fontsize=11, fontweight="bold")
+ax1.set_ylabel("Number of HSIs disrupted", fontsize=11, fontweight="bold")
+ax1.set_title(
+    f"Monthly weather-disrupted HSIs — SSP2-4.5 mean (n={pop_size:,})",
+    fontsize=12, fontweight="bold",
+)
+ax1.set_ylim(bottom=0)
+ax1.legend(fontsize=10, framealpha=0.9)
+fig1.tight_layout()
+fig1.savefig("./outputs/weather_disruptions_monthly_counts.png", dpi=300, bbox_inches="tight")
+plt.show()
+plt.close(fig1)
+
+#  PLOT 2: Monthly disruption rate (%)
+
+fig2, ax2 = plt.subplots(figsize=(12, 5))
+
+ax2.plot(
+    monthly.index, monthly["disruption_rate"] * 100,
+    color=COLOUR_RATE, lw=2, label="Disruption rate (%)",
+)
+
+ax2.xaxis.set_major_formatter(mdates.DateFormatter("%Y-%m"))
+ax2.xaxis.set_major_locator(mdates.MonthLocator(interval=6))
+plt.setp(ax2.xaxis.get_majorticklabels(), rotation=45, ha="right")
+ax2.set_xlabel("Month", fontsize=11, fontweight="bold")
+ax2.set_ylabel("% HSIs disrupted", fontsize=11, fontweight="bold", color=COLOUR_RATE)
+ax2.set_ylim(bottom=0)
+ax2.set_title(
+    "Monthly disruption rate and total HSI volume",
+    fontsize=12, fontweight="bold",
+)
+
+lines_2, labels_2 = ax2.get_legend_handles_labels()
+ax2.legend(lines_2, labels_2, fontsize=10, framealpha=0.9)
+
+fig2.tight_layout()
+fig2.savefig("./outputs/weather_disruptions_monthly_rate.png", dpi=300, bbox_inches="tight")
+plt.show()
+plt.close(fig2)
+
+#  PLOT 3: Annual disruption counts and rate
+
+fig3, ax3a = plt.subplots(figsize=(10, 5))
+ax3b = ax3a.twinx()
+
+years = annual.index.year
+x = np.arange(len(years))
+width = 0.35
+
+ax3a.bar(x - width / 2, annual["delayed"], width, color=COLOUR_DELAYED, alpha=0.85, label="Delayed")
+ax3a.bar(x + width / 2, annual["cancelled"], width, color=COLOUR_CANCELLED, alpha=0.85, label="Cancelled")
+ax3b.plot(x, annual["disruption_rate"] * 100, "o--", color=COLOUR_RATE, lw=2, ms=6, label="Disruption rate (%)")
+
+ax3a.set_xticks(x)
+ax3a.set_xticklabels(years, rotation=45, ha="right")
+ax3a.set_xlabel("Year", fontsize=11, fontweight="bold")
+ax3a.set_ylabel("Number of HSIs disrupted", fontsize=11, fontweight="bold")
+ax3b.set_ylabel("% HSIs disrupted", fontsize=11, fontweight="bold", color=COLOUR_RATE)
+ax3a.set_title(
+    "Annual weather-disrupted HSIs and disruption rate",
+    fontsize=12, fontweight="bold",
+)
+
+lines_a, labels_a = ax3a.get_legend_handles_labels()
+lines_b, labels_b = ax3b.get_legend_handles_labels()
+ax3a.legend(lines_a + lines_b, labels_a + labels_b, fontsize=10, framealpha=0.9)
+
+fig3.tight_layout()
+fig3.savefig("./outputs/weather_disruptions_annual.png", dpi=300, bbox_inches="tight")
+plt.show()
+plt.close(fig3)
+
+#  PLOT 4: Top N HSI types disrupted — horizontal bar chart
+
+if not top_types.empty:
+    fig4, ax4 = plt.subplots(figsize=(10, max(5, TOP_N * 0.5)))
+
+    y_pos = np.arange(len(top_types))
+    ax4.barh(
+        y_pos, top_types["delayed"].values,
+        color=COLOUR_DELAYED, alpha=0.85, label="Delayed",
+    )
+    ax4.barh(
+        y_pos, top_types["cancelled"].values,
+        left=top_types["delayed"].values,
+        color=COLOUR_CANCELLED, alpha=0.85, label="Cancelled",
+    )
+
+    ax4.set_yticks(y_pos)
+    ax4.set_yticklabels(top_types.index, fontsize=9)
+    ax4.invert_yaxis()
+    ax4.set_xlabel("Number of HSI events disrupted", fontsize=11, fontweight="bold")
+    ax4.set_title(
+        f"Top {TOP_N} HSI types by weather disruption count ({start_date.year}–{end_date.year - 1})",
+        fontsize=12, fontweight="bold",
+    )
+    ax4.legend(fontsize=10, framealpha=0.9)
+
+    fig4.tight_layout()
+    fig4.savefig("./outputs/weather_disruptions_by_hsi_type.png", dpi=300, bbox_inches="tight")
+    plt.show()
+    plt.close(fig4)
+
+# ─────────────────────────────────────────────────────────────────────────────
+#  PRINTED SUMMARY
+# ─────────────────────────────────────────────────────────────────────────────
+
+total_cancelled = int(monthly["cancelled"].sum())
+total_delayed = int(monthly["delayed"].sum())
+total_disrupted = total_cancelled + total_delayed
+total_hsi_seen = int(monthly["hsi_total"].sum())
+mean_monthly_rate = monthly["disruption_rate"].mean() * 100
+peak_month = monthly["disruption_rate"].idxmax()
+peak_rate = monthly["disruption_rate"].max() * 100
+
+print("\n" + "=" * 60)
+print("  WEATHER DISRUPTIONS — SIMULATION SUMMARY")
+print("=" * 60)
+print("  Period            : {start_date} → {end_date}")
+print("  Population size   : {pop_size:,}")
+print("  SSP / model       : ssp245 / mean")
+print("-" * 60)
+print("Total HSIs seen   : {total_hsi_seen:,}")
+print("  Total disrupted   : {total_disrupted:,}")
+print("    Cancelled       : {total_cancelled:,}  ({total_cancelled / total_disrupted * 100:.1f}%)")
+print("    Delayed         : {total_delayed:,}  ({total_delayed / total_disrupted * 100:.1f}%)")
+print("  Mean monthly rate : {mean_monthly_rate:.2f}%")
+print("  Peak month        : {peak_month.strftime('%Y-%m')}  ({peak_rate:.2f}%)")
+print("=" * 60)