From df5dba964ef2403cacc8ea6871e9fa8ab95dc464 Mon Sep 17 00:00:00 2001
From: Slesa Adhikari <slesaad@gmail.com>
Date: Tue, 22 Sep 2026 16:32:55 -0500
Subject: [PATCH] Add Air4US datasets vector ingest configs

---
 .../vector-config/aqs-ozone-first.json        | 30 +++++++++++++++
 .../staging/vector-config/aqs-ozone-last.json | 38 +++++++++++++++++++
 .../staging/vector-config/hms-fire.json       | 18 +++++++++
 .../staging/vector-config/hms-smoke.json      | 18 +++++++++
 4 files changed, 104 insertions(+)
 create mode 100644 ingestion-data/staging/vector-config/aqs-ozone-first.json
 create mode 100644 ingestion-data/staging/vector-config/aqs-ozone-last.json
 create mode 100644 ingestion-data/staging/vector-config/hms-fire.json
 create mode 100644 ingestion-data/staging/vector-config/hms-smoke.json

diff --git a/ingestion-data/staging/vector-config/aqs-ozone-first.json b/ingestion-data/staging/vector-config/aqs-ozone-first.json
new file mode 100644
index 0000000..baa34b5
--- /dev/null
+++ b/ingestion-data/staging/vector-config/aqs-ozone-first.json
@@ -0,0 +1,30 @@
+{
+  "bucket": "airquality-data-store-develop",
+  "prefix": "aqs/transformed/hourly/ozone/",
+  "vector": true,
+  "chunk_size": 1,
+  "worker_concurrency": 1,
+  "collection": "ozone_hourly_bt_test_slesa2",
+  "filename_regex": ".*hourly_44201_2020\\.csv$",
+  "extra_flags": [
+    "-append",
+    "-nln",
+    "ozone_hourly_bt_test_slesa2",
+    "-oo",
+    "X_POSSIBLE_NAMES=longitude",
+    "-oo",
+    "Y_POSSIBLE_NAMES=latitude",
+    "-oo",
+    "AUTODETECT_TYPE=YES",
+    "-oo",
+    "EMPTY_STRING_AS_NULL=YES",
+    "--config",
+    "PG_USE_COPY",
+    "YES",
+    "--debug",
+    "on"
+  ],
+  "source_projection": "EPSG:4326",
+  "target_projection": "EPSG:4326",
+  "invalidate_cloudfront": false
+}
\ No newline at end of file
diff --git a/ingestion-data/staging/vector-config/aqs-ozone-last.json b/ingestion-data/staging/vector-config/aqs-ozone-last.json
new file mode 100644
index 0000000..07a776b
--- /dev/null
+++ b/ingestion-data/staging/vector-config/aqs-ozone-last.json
@@ -0,0 +1,38 @@
+{
+  "bucket": "airquality-data-store-develop",
+  "prefix": "aqs/transformed/hourly/ozone/",
+  "vector": true,
+  "chunk_size": 1,
+  "worker_concurrency": 1,
+  "collection": "ozone_hourly_bt_test_slesa2",
+  "filename_regex": ".*hourly_44201_202[1-4]\\.csv$",
+  "extra_flags": [
+    "-append",
+    "-nln",
+    "ozone_hourly_bt_test_slesa2",
+    "-oo",
+    "X_POSSIBLE_NAMES=longitude",
+    "-oo",
+    "Y_POSSIBLE_NAMES=latitude",
+    "-oo",
+    "AUTODETECT_TYPE=YES",
+    "-oo",
+    "EMPTY_STRING_AS_NULL=YES",
+    "--config",
+    "PG_USE_COPY",
+    "YES",
+    "--debug",
+    "on"
+  ],
+  "table_config": {
+    "schema": "public",
+    "indexes": [
+      {"columns": ["datetime_gmt_utc"], "method": "btree", "concurrently": true},
+      {"columns": ["state_code", "county_code", "site_num", "poc", "datetime_gmt_utc"], "method": "btree", "concurrently": true}
+    ],
+    "analyze": true
+  },
+  "source_projection": "EPSG:4326",
+  "target_projection": "EPSG:4326",
+  "invalidate_cloudfront": true
+}
\ No newline at end of file
diff --git a/ingestion-data/staging/vector-config/hms-fire.json b/ingestion-data/staging/vector-config/hms-fire.json
new file mode 100644
index 0000000..8d4de9a
--- /dev/null
+++ b/ingestion-data/staging/vector-config/hms-fire.json
@@ -0,0 +1,18 @@
+{
+  "collection": "hms_fire_sample",
+  "bucket": "airquality-data-store-develop",
+  "prefix": "hms_fire/featuresdb/",
+  "filename_regex": ".*\\.gpkg$",
+  "vector": true,
+  "source_projection": "EPSG:4326",
+  "target_projection": "EPSG:4326",
+  "extra_flags": ["-overwrite", "-progress", "-nlt", "POINT", "-lco", "GEOMETRY_NAME=geom"],
+  "table_config": {
+    "schema": "public",
+    "indexes": [
+      {"columns": ["datetime"], "method": "btree", "concurrently": true}
+    ],
+    "analyze": true
+  },
+  "invalidate_cloudfront": true
+}
\ No newline at end of file
diff --git a/ingestion-data/staging/vector-config/hms-smoke.json b/ingestion-data/staging/vector-config/hms-smoke.json
new file mode 100644
index 0000000..09652d3
--- /dev/null
+++ b/ingestion-data/staging/vector-config/hms-smoke.json
@@ -0,0 +1,18 @@
+{
+  "collection": "hms_smoke",
+  "bucket": "airquality-data-store-develop",
+  "prefix": "hms_smoke/featuresdb/",
+  "filename_regex": ".*\\.gpkg$",
+  "vector": true,
+  "source_projection": "EPSG:4326",
+  "target_projection": "EPSG:4326",
+  "extra_flags": ["-overwrite", "-progress", "-nlt", "MULTIPOLYGON", "-lco", "GEOMETRY_NAME=geom"],
+  "invalidate_cloudfront": true,
+  "table_config": {
+    "schema": "public",
+    "indexes": [
+      {"columns": ["datetime"], "method": "btree", "concurrently": true}
+    ],
+    "analyze": true
+  }
+}
\ No newline at end of file