fix(dswx-s1): tile collection location

Refs #705

cluster_provisioning/modules/common/mozart.tf

@@ -580,7 +580,7 @@ resource "aws_instance" "mozart" {
 	  
       # For daac_data_subscriber utility tool
       mkdir ~/Downloads/
-      aws s3 cp  s3://opera-ancillaries/MGRS_tile_collection_v0.2.sqlite ~/Downloads/
+      aws s3 cp  s3://opera-ancillaries/mgrs_tiles/dswx_s1/MGRS_tile_collection_v0.3.sqlite ~/Downloads/
     EOT
     ]
   }

conf/settings.yaml

@@ -34,7 +34,7 @@ DISP_S1_PRODUCT_VERSION: "0.1"
 DSWX_S1_COVERAGE_TARGET: 99
 # Force DSWx-S1 processing of pending burst sets after X minutes, rather than waiting for additional data
 DSWX_S1_COLLECTION_GRACE_PERIOD_MINUTES: 210
-MGRS_TILE_COLLECTION_DB_S3PATH: "s3://opera-ancillaries/MGRS_tile_collection_v0.2.sqlite"
+MGRS_TILE_COLLECTION_DB_S3PATH: "s3://opera-ancillaries/mgrs_tiles/dswx_s1/MGRS_tile_collection_v0.3.sqlite"
 
 # The minimum coverage, defined as the percent of bursts in a bursts set, required to run DISP_S1
 #  Must be a number in the closed interval [0, 99].

data_subscriber/rtc/mgrs_bursts_collection_db_client.py

@@ -1,5 +1,6 @@
 import ast
 import logging
+import os
 import re
 from collections import defaultdict
 from functools import cache
@@ -54,7 +55,7 @@ def load_mgrs_burst_db(filter_land=True):
 
 def load_mgrs_burst_db_raw(filter_land=True) -> GeoDataFrame:
     """Loads the MGRS Tile Collection Database. On AWS environments, this will localize from a known S3 location."""
-    mtc_local_filepath = Path("~/Downloads/MGRS_tile_collection_v0.2.sqlite").expanduser()
+    mtc_local_filepath = Path(os.environ.get("MGRS_TILE_COLLECTION_DB_FILEPATH", "~/Downloads/MGRS_tile_collection_v0.3.sqlite")).expanduser()
     if mtc_local_filepath.exists():
         vector_gdf = gpd.read_file(mtc_local_filepath, crs="EPSG:4326")  # , bbox=(-230, 0, -10, 90))  # bbox=(-180, -90, 180, 90)  # global
     else:
@@ -70,7 +71,7 @@ def load_mgrs_burst_db_raw(filter_land=True) -> GeoDataFrame:
     # vector_gdf = vector_gdf.overlay(na_gdf, how="intersection")
     logger.info(f"{len(vector_gdf)=}")
     if filter_land:
-        vector_gdf = vector_gdf[vector_gdf["land_ocean_flag"] == "water/land"]  # filter out water (water == no relevant data)
+        vector_gdf = vector_gdf[vector_gdf["land_ocean_flag"] in ["water/land", "land"]]  # filter out water (water == no relevant data)
         logger.info(f"{len(vector_gdf)=}")
 
     return vector_gdf

tests/opera_chimera/test_precondition_functions.py

@@ -585,7 +585,7 @@ def test_get_dswx_s1_static_ancillary_files(self):
                 },
                 "mgrs_collection_database_file": {
                     oc_const.S3_BUCKET: "opera-ancillaries",
-                    oc_const.S3_KEY: "mgrs_tiles/dswx_s1/MGRS_tile_collection_v0.2.1.sqlite"
+                    oc_const.S3_KEY: "mgrs_tiles/dswx_s1/MGRS_tile_collection_v0.3.sqlite"
                 }
             }
         }

tools/ops/README.md

@@ -73,7 +73,7 @@ See `data_subscriber/data_subscriber_client.py --help`, documentation comments,
 1. Python (see `.python-version`).
 1. A clone of the `opera-sds-pcm` repo.
 1. data_subscriber_client.py requires GDAL to be natively installed on the host machine. Installation instructions vary by operating system, but macOS users with Homebrew installed may install it using `brew install gdal` or as otherwise noted in `setup.py` in the `cmr_audit` dependency section. NOTE: native `gdal` is distinct from the `GDAL` python package dependency.
-1. data_subscriber_client.py requires the MGRS collection DB file to be present in `~/Downloads/MGRS_tile_collection_v0.2.sqlite` This file is found in the ancillary S3 and may also be in the opera-pcm repo.
+1. data_subscriber_client.py requires the MGRS collection DB file to be present in `~/Downloads/MGRS_tile_collection_v0.3.sqlite` or the path defined in environment variable `MGRS_TILE_COLLECTION_DB_FILEPATH`. This file is found in the ancillary S3 and may also be in the opera-pcm repo.
 
 ### Installation