test / preprocess.py
cassiebuhler's picture
filling missing values, include all gap codes
0c88eb4
import ibis
from ibis import _
conn = ibis.duckdb.connect("tmp3", extensions=["spatial"])
# ca_parquet = "https://data.source.coop/cboettig/ca30x30/ca_areas.parquet"
# or use local copy:
ca_parquet = "ca_areas.parquet"
# negative buffer to account for overlapping boundaries.
buffer = -30 #30m buffer
tbl = (
conn.read_parquet(ca_parquet)
.cast({"SHAPE": "geometry"})
.rename(geom = "SHAPE")
.filter(_.reGAP < 3) # only gap 1 and 2 count towards 30x30
)
# polygons with release_year 2024 are a superset of release_year 2023.
# use anti_join to isolate the objects that are in release_year 2024 but not release_year 2023 (aka newly established).
tbl_2023 = tbl.filter(_.Release_Year == 2023).mutate(geom=_.geom.buffer(buffer))
tbl_2024 = tbl.filter(_.Release_Year == 2024)
intersects = tbl_2024.anti_join(tbl_2023, _.geom.intersects(tbl_2023.geom))
new2024 = intersects.select("OBJECTID").mutate(established = 2024) # saving IDs to join on
ca = (conn
.read_parquet(ca_parquet)
.cast({"SHAPE": "geometry"})
.mutate(area = _.SHAPE.area())
.filter(_.Release_Year == 2024) # having both 2023 and 2024 is redudant since 2024 is the superset.
.left_join(new2024, "OBJECTID") # newly established 2024 polygons
.mutate(established=_.established.fill_null(2023))
.mutate(geom = _.SHAPE.convert("epsg:3310","epsg:4326"))
.rename(name = "cpad_PARK_NAME", access_type = "cpad_ACCESS_TYP", manager = "cpad_MNG_AGENCY",
manager_type = "cpad_MNG_AG_LEV", id = "OBJECTID", type = "TYPE")
.mutate(manager = _.manager.substitute({"": "Unknown"}))
.mutate(manager_type = _.manager_type.substitute({"": "Unknown"}))
.mutate(access_type = _.access_type.substitute({"": "Unknown Access"}))
.mutate(name = _.name.substitute({"": "Unknown"}))
.select(_.established, _.reGAP, _.name, _.access_type, _.manager, _.manager_type,
_.Easement, _.Acres, _.id, _.type, _.geom)
)
ca2024 = ca.execute()
ca2024.to_parquet("ca2024-30m.parquet")
ca2024.to_file("ca2024-30m.geojson") # tippecanoe can't parse geoparquet :-(
## Upload to Huggingface
# https://huggingface.co/datasets/boettiger-lab/ca-30x30/
from huggingface_hub import HfApi, login
import streamlit as st
login(st.secrets["HF_TOKEN"])
api = HfApi()
def hf_upload(file):
info = api.upload_file(
path_or_fileobj=file,
path_in_repo=file,
repo_id="boettiger-lab/ca-30x30",
repo_type="dataset",
)
hf_upload("ca2024-30m.parquet")
import subprocess
import os
def generate_pmtiles(input_file, output_file, max_zoom=12):
# Ensure Tippecanoe is installed
if subprocess.call(["which", "tippecanoe"], stdout=subprocess.DEVNULL) != 0:
raise RuntimeError("Tippecanoe is not installed or not in PATH")
# Construct the Tippecanoe command
command = [
"tippecanoe",
"-o", output_file,
"-z", str(max_zoom),
"--drop-densest-as-needed",
"--extend-zooms-if-still-dropping",
"--force",
input_file
]
# Run Tippecanoe
try:
subprocess.run(command, check=True)
print(f"Successfully generated PMTiles file: {output_file}")
except subprocess.CalledProcessError as e:
print(f"Error running Tippecanoe: {e}")
generate_pmtiles("ca2024-30m.geojson", "ca2024-30m-tippe.pmtiles")
hf_upload("ca2024-30m-tippe.pmtiles")