update for 4 hour rain forecast

2026-04-06 18:32:33 +10:00
parent fb50c8ed71
commit 3a7309b2cf
20 changed files with 716 additions and 132 deletions
@@ -112,6 +112,34 @@ CREATE INDEX IF NOT EXISTS idx_predictions_rain_1h_site_ts
 CREATE INDEX IF NOT EXISTS idx_predictions_rain_1h_pending_eval
  ON predictions_rain_1h(site, evaluated_at, ts DESC);

+-- Rain model predictions (next 4h)
+CREATE TABLE IF NOT EXISTS predictions_rain_4h (
+  ts                    TIMESTAMPTZ NOT NULL,
+  generated_at          TIMESTAMPTZ NOT NULL DEFAULT now(),
+  site                  TEXT NOT NULL,
+  model_name            TEXT NOT NULL,
+  model_version         TEXT NOT NULL,
+  threshold             DOUBLE PRECISION NOT NULL,
+  probability           DOUBLE PRECISION NOT NULL,
+  predict_rain          BOOLEAN NOT NULL,
+
+  rain_next_4h_mm_actual DOUBLE PRECISION,
+  rain_next_4h_actual    BOOLEAN,
+  evaluated_at           TIMESTAMPTZ,
+
+  metadata              JSONB,
+
+  PRIMARY KEY (site, model_name, model_version, ts)
+);
+
+SELECT create_hypertable('predictions_rain_4h', 'ts', if_not_exists => TRUE);
+
+CREATE INDEX IF NOT EXISTS idx_predictions_rain_4h_site_ts
+  ON predictions_rain_4h(site, ts DESC);
+
+CREATE INDEX IF NOT EXISTS idx_predictions_rain_4h_pending_eval
+  ON predictions_rain_4h(site, evaluated_at, ts DESC);
+
 -- Raw retention: 90 days
 DO $$
 BEGIN
@@ -100,44 +100,104 @@ FROM baseline;


 CREATE OR REPLACE VIEW rain_prediction_drift_daily AS
+WITH all_predictions AS (
+  SELECT
+    1::INT AS horizon_hours,
+    ts,
+    site,
+    model_name,
+    model_version,
+    threshold,
+    probability,
+    predict_rain
+  FROM predictions_rain_1h
+  UNION ALL
+  SELECT
+    4::INT AS horizon_hours,
+    ts,
+    site,
+    model_name,
+    model_version,
+    threshold,
+    probability,
+    predict_rain
+  FROM predictions_rain_4h
+)
 SELECT
  time_bucket(INTERVAL '1 day', ts) AS day,
  site,
  model_name,
  model_version,
+  horizon_hours,
  count(*) AS prediction_rows,
  avg(probability) AS probability_mean,
  stddev_samp(probability) AS probability_stddev,
  avg(CASE WHEN predict_rain THEN 1.0 ELSE 0.0 END) AS predicted_positive_rate,
  avg(threshold) AS threshold_mean
-FROM predictions_rain_1h
-GROUP BY 1,2,3,4;
+FROM all_predictions
+GROUP BY 1,2,3,4,5;


 CREATE OR REPLACE VIEW rain_calibration_drift_daily AS
+WITH all_predictions AS (
+  SELECT
+    1::INT AS horizon_hours,
+    ts,
+    site,
+    model_name,
+    model_version,
+    probability,
+    predict_rain,
+    rain_next_1h_actual AS rain_actual
+  FROM predictions_rain_1h
+  UNION ALL
+  SELECT
+    4::INT AS horizon_hours,
+    ts,
+    site,
+    model_name,
+    model_version,
+    probability,
+    predict_rain,
+    rain_next_4h_actual AS rain_actual
+  FROM predictions_rain_4h
+)
 SELECT
  time_bucket(INTERVAL '1 day', ts) AS day,
  site,
  model_name,
  model_version,
-  count(*) FILTER (WHERE rain_next_1h_actual IS NOT NULL) AS evaluated_rows,
-  avg(probability) FILTER (WHERE rain_next_1h_actual IS NOT NULL) AS mean_probability,
-  avg(CASE WHEN rain_next_1h_actual THEN 1.0 ELSE 0.0 END) FILTER (WHERE rain_next_1h_actual IS NOT NULL) AS observed_positive_rate,
-  avg(power(probability - CASE WHEN rain_next_1h_actual THEN 1.0 ELSE 0.0 END, 2.0))
-    FILTER (WHERE rain_next_1h_actual IS NOT NULL) AS brier_score,
+  horizon_hours,
+  count(*) FILTER (WHERE rain_actual IS NOT NULL) AS evaluated_rows,
+  avg(probability) FILTER (WHERE rain_actual IS NOT NULL) AS mean_probability,
+  avg(CASE WHEN rain_actual THEN 1.0 ELSE 0.0 END) FILTER (WHERE rain_actual IS NOT NULL) AS observed_positive_rate,
+  avg(power(probability - CASE WHEN rain_actual THEN 1.0 ELSE 0.0 END, 2.0))
+    FILTER (WHERE rain_actual IS NOT NULL) AS brier_score,
  avg(
    CASE
-      WHEN rain_next_1h_actual IS NULL THEN NULL
-      WHEN predict_rain = rain_next_1h_actual THEN 1.0
+      WHEN rain_actual IS NULL THEN NULL
+      WHEN predict_rain = rain_actual THEN 1.0
      ELSE 0.0
    END
  ) AS decision_accuracy
-FROM predictions_rain_1h
-GROUP BY 1,2,3,4;
+FROM all_predictions
+GROUP BY 1,2,3,4,5;


 CREATE OR REPLACE VIEW rain_pipeline_health AS
-WITH sites AS (
+WITH prediction_latest AS (
+  SELECT
+    site,
+    max(generated_at) AS prediction_generated_latest_ts,
+    max(evaluated_at) AS prediction_evaluated_latest_ts
+  FROM (
+    SELECT site, generated_at, evaluated_at FROM predictions_rain_1h
+    UNION ALL
+    SELECT site, generated_at, evaluated_at FROM predictions_rain_4h
+  ) p
+  GROUP BY site
+),
+sites AS (
  SELECT DISTINCT site FROM observations_ws90
  UNION
  SELECT DISTINCT site FROM observations_baro
@@ -145,12 +205,16 @@ WITH sites AS (
  SELECT DISTINCT site FROM forecast_openmeteo_hourly
  UNION
  SELECT DISTINCT site FROM predictions_rain_1h
+  UNION
+  SELECT DISTINCT site FROM predictions_rain_4h
 )
 SELECT
  s.site,
  (SELECT max(ts) FROM observations_ws90 w WHERE w.site = s.site) AS ws90_latest_ts,
  (SELECT max(ts) FROM observations_baro b WHERE b.site = s.site) AS baro_latest_ts,
  (SELECT max(ts) FROM forecast_openmeteo_hourly f WHERE f.site = s.site) AS forecast_latest_ts,
-  (SELECT max(generated_at) FROM predictions_rain_1h p WHERE p.site = s.site) AS prediction_generated_latest_ts,
-  (SELECT max(evaluated_at) FROM predictions_rain_1h p WHERE p.site = s.site) AS prediction_evaluated_latest_ts
-FROM sites s;
+  p.prediction_generated_latest_ts,
+  p.prediction_evaluated_latest_ts
+FROM sites s
+LEFT JOIN prediction_latest p
+  ON p.site = s.site;
@@ -0,0 +1,29 @@
+-- Add 4-hour rain prediction storage table.
+-- Safe to re-run.
+
+CREATE TABLE IF NOT EXISTS predictions_rain_4h (
+  ts                    TIMESTAMPTZ NOT NULL,
+  generated_at          TIMESTAMPTZ NOT NULL DEFAULT now(),
+  site                  TEXT NOT NULL,
+  model_name            TEXT NOT NULL,
+  model_version         TEXT NOT NULL,
+  threshold             DOUBLE PRECISION NOT NULL,
+  probability           DOUBLE PRECISION NOT NULL,
+  predict_rain          BOOLEAN NOT NULL,
+
+  rain_next_4h_mm_actual DOUBLE PRECISION,
+  rain_next_4h_actual    BOOLEAN,
+  evaluated_at           TIMESTAMPTZ,
+
+  metadata              JSONB,
+
+  PRIMARY KEY (site, model_name, model_version, ts)
+);
+
+SELECT create_hypertable('predictions_rain_4h', 'ts', if_not_exists => TRUE);
+
+CREATE INDEX IF NOT EXISTS idx_predictions_rain_4h_site_ts
+  ON predictions_rain_4h(site, ts DESC);
+
+CREATE INDEX IF NOT EXISTS idx_predictions_rain_4h_pending_eval
+  ON predictions_rain_4h(site, evaluated_at, ts DESC);