Source code for eodms_api_client.geo
import logging
from os.path import splitext
import fiona
import geopandas as gpd
import pandas as pd
from pyproj import CRS, Transformer
from shapely.geometry import Polygon, shape
from shapely.wkt import dumps as to_wkt
# add 'read' support for KML/KMZ - disabled by default
fiona.drvsupport.supported_drivers['KML'] = 'r'
fiona.drvsupport.supported_drivers['LIBKML'] = 'r'
LOGGER = logging.getLogger('eodmsapi.geo')
SRC_CRS = CRS('epsg:4326')
[docs]def transform_metadata_geometry(meta_geom, target_crs=None):
'''
Transform a given image footprint geometry from WGS84 to the desired crs
EODMS always returns geometries in WGS84, so we provide the option to transform
into something more useful through the use of a CLI flag (--t_srs)
Inputs:
- meta_geom: the footprint geometry as a shapely.geometry object
- target_crs: either a pyproj.CRS or properly-formatted string for the desired projection
Outputs:
- if target_crs is not supplied, simply return the passed geometry
- if target_crs is supplied, return the transformed geometry
'''
meta_geom = shape(meta_geom)
if target_crs is None:
return meta_geom
elif isinstance(target_crs, CRS):
dst_crs = target_crs
else:
dst_crs = CRS(target_crs)
trans = Transformer.from_crs(
crs_from=SRC_CRS,
crs_to=dst_crs,
always_xy=True
)
lons, lats = meta_geom.exterior.coords.xy
xs, ys = trans.transform(xx=lons, yy=lats)
return Polygon(zip(xs, ys))
[docs]def metadata_to_gdf(metadata, collection, target_crs=None):
'''
Instead of a jumbled JSON-looking dictionary, create a geopandas.GeoDataFrame from the
EODMS search query response
Inputs:
- metadata: dictionary of EODMS search query and image metadata response
- collection: string of EODMS Collection ID - used for column name standardization
- target_crs: pyproj.CRS instance or properly-formatted string for the desired projection
Outputs:
- df: geopandas.GeoDataFrame containing <metadata> projected to <target_crs>
'''
if target_crs is None:
crs = CRS('epsg:4326')
elif isinstance(target_crs, CRS):
crs = target_crs
else:
crs = CRS(target_crs)
df = gpd.GeoDataFrame(metadata, crs=crs)
# standardize some column names
if collection == 'RCMImageProducts':
df.rename(
{
'recordId': 'EODMS RecordId',
'title': 'Granule'
},
axis=1,
inplace=True
)
date_cols = ['Acquisition Start Date', 'Acquisition End Date']
int_cols = [
'EODMS RecordId', 'Number of Azimuth Looks', 'Number of Range Looks',
'SIP Size (MB)', 'Relative Orbit', 'Absolute Orbit', 'Beam Mode Version'
]
float_cols = [
'Incidence Angle (Low)', 'Incidence Angle (High)', 'Sampled Pixel Spacing',
'Spatial Resolution'
]
elif collection == 'Radarsat2':
df.rename(
{
'Sequence Id': 'EODMS RecordId',
'Supplier Order Number': 'Granule'
},
axis=1,
inplace=True
)
date_cols = ['Start Date', 'End Date']
int_cols = [
'EODMS RecordId', 'SIP Size (MB)', 'Absolute Orbit'
]
float_cols = [
'Incidence Angle (Low)', 'Incidence Angle (High)', 'Spatial Resolution'
]
elif collection == 'Radarsat1':
df.rename(
{
'Sequence Id': 'EODMS RecordId',
'Product Id': 'Granule',
'Start Date': 'End Date', # TODO: Check that this is indeed necessary!
'End Date': 'Start Date' # TODO: Check that this is indeed necessary!
},
axis=1,
inplace=True
)
# fix column ordering for RS1
df = df[[
'EODMS RecordId', 'Granule', 'Start Date', 'End Date', 'Position',
'Sensor', 'Sensor Mode', 'Beam', 'Polarization', 'Look Orientation',
'Incidence Angle (Low)', 'Incidence Angle (High)', 'Orbit Direction',
'Absolute Orbit', 'LUT Applied', 'Product Format', 'Product Type',
'Spatial Resolution', 'SIP Size (MB)', 'geometry'
]]
date_cols = ['Start Date', 'End Date']
int_cols = [
'EODMS RecordId', 'SIP Size (MB)', 'Absolute Orbit'
]
float_cols = [
'Incidence Angle (Low)', 'Incidence Angle (High)', 'Spatial Resolution'
]
elif collection in ['PlanetScope', 'NAPL']:
df.rename(
{
'Sequence Id': 'EODMS RecordId',
'Title': 'Granule'
},
axis=1,
inplace=True
)
date_cols = ['Start Date', 'End Date']
int_cols = [
'EODMS RecordId', 'SIP Size (MB)',
]
float_cols = [
'Incidence Angle (Low)', 'Incidence Angle (High)'
]
# convert strings to unsigned integer
# necessary to do one-by-one because if we apply to dataframe, any "failure" fields
# will cause all "valid" fields to not be converted
for int_col in int_cols:
df[int_col] = pd.to_numeric(df[int_col], downcast='unsigned', errors='ignore')
# convert strings to floats
for float_col in float_cols:
df[float_col] = pd.to_numeric(df[float_col], downcast='float', errors='ignore')
# convert strings to datetimes
df[date_cols] = df[date_cols].apply(pd.to_datetime, axis=1)
# sort by RecordId
df.sort_values(by='EODMS RecordId', inplace=True)
df.reset_index(drop=True, inplace=True)
return df
[docs]def load_search_aoi(geofile):
'''
Given a file, attempt to parse it as a set of vector features to be sent as part
of a request to the EODMS REST API. If the features are not already in WGS84, they
are transformed by this function.
Inputs:
- geofile: A file containing vector data (SHP, GEOJSON, GPKG, etc.)
Outputs:
- wkt: The Well-Known Text representation of the features within <geofile>
'''
# use vsizip for KMZ/Zipped shapefile
if splitext(geofile)[-1] in ['.kmz', '.zip']:
df = gpd.read_file(f'/vsizip/{geofile}')
else:
df = gpd.read_file(geofile)
if df.crs != SRC_CRS:
df = df.to_crs(SRC_CRS)
geometry = df.unary_union
if geometry.type == 'MultiPolygon':
n_vertices = sum([
len(poly.exterior.coords) - 1
for poly in geometry.geoms
])
elif geometry.type == 'Polygon':
n_vertices = len(geometry.exterior.coords) - 1
else:
raise NotImplementedError('Search geometry must be a polygon/multipolygon')
if n_vertices > 100:
LOGGER.warn('Search geometry is too complex (more than 100 vertices) - Simplifying with 0.01° tolerance')
geometry = geometry.simplify(tolerance=0.01)
# use 3-decimal precision (tens to hundreds of meters)
# drop Z dimension if it exists - causes 500-errors with EODMS
wkt = to_wkt(geometry, rounding_precision=3, output_dimension=2)
else:
# drop Z dimension if it exists - causes 500-errors with EODMS
wkt = to_wkt(geometry, output_dimension=2)
return wkt