Skip to main content

EarthAI Platform Support Portal

How to Query the EarthAI Catalog with OpenStreetMap Data

Brown, Jason
  • Updated

You can use a large variety of spatial data to query Earth OnDemand. This page shows you how to use the crowdsourced OpenStreetMap (OSM) to query Earth OnDemand for related imagery. To access OSM data, we will use the osmnx library. There are many ways to access OSM data, and this is useful for datasets like streets, buildings and points of interest.

Import Libraries

In [1]:
import osmnx
from earthai.init import *
Importing EarthAI libraries.
EarthAI version 1.6.0; RasterFrames version 0.9.0; PySpark version 2.4.7
Creating SparkSession...
SparkSession is available as `spark`.

Retrieve OpenStreetMap Data

We will extract all the libraries in the state of Vermont, USA.

In [2]:
lib_gdf = osmnx.geometries_from_place("Vermont, United States",
                                tags={"amenity": ["library"]})
print("There are "+str(len(lib_gdf))+" libraries in Vermont, USA.")
There are 244 libraries in Vermont, USA.

The data are returned as a GeoPandas GeoDataFrame. This enables easy plotting and reasoning over the spatial data. You can use the head method to view the top few rows, and the plot method to plot the geometries.

In [3]:
lib_gdf.head(5)
Out[3]:
  unique_id osmid element_type amenity name wheelchair wikidata geometry addr:state ele ... twitter name:en name:nl operator:type building:levels fixme opening_hours:url start_date height layer
0 node/366667053 366667053 node library Butterfield Library NaN Q37908869 POINT (-72.45863 43.06724) VT 94 ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
1 way/607980301 607980301 way library Putney Public Library NaN NaN POLYGON ((-72.52289 42.96962, -72.52289 42.96968, -72.52272 42.96967, -72.52273 42.96961, -72.52264 42.96961, -72.52264 42.96958, -72.52262 42.96958, -72.52262 42.96956, -72.52262 42.96952, -72.52264 42.96952, -72.52265 42.96952, -72.52265 42.96949, -72.52272 42.96949, -72.52273 42.96945, -72.52288 42.96945, -72.52288 42.96950, -72.52301 42.96950, -72.52300 42.96962, -72.52289 42.96962)) VT 114 ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
2 node/366667250 366667250 node library Springfield Hospital Information Center Library NaN NaN POINT (-72.49453 43.29813) VT 181 ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
3 way/445267247 445267247 way library NaN NaN NaN POLYGON ((-72.50874 43.14230, -72.50873 43.14222, -72.50866 43.14222, -72.50864 43.14198, -72.50857 43.14198, -72.50852 43.14198, -72.50853 43.14222, -72.50850 43.14222, -72.50847 43.14222, -72.50848 43.14232, -72.50874 43.14230)) NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
4 way/606852021 606852021 way library Windsor Public Library NaN NaN POLYGON ((-72.38837 43.48035, -72.38839 43.48027, -72.38846 43.48028, -72.38848 43.48023, -72.38857 43.48025, -72.38856 43.48029, -72.38862 43.48030, -72.38860 43.48038, -72.38851 43.48036, -72.38851 43.48038, -72.38847 43.48037, -72.38847 43.48036, -72.38837 43.48035)) VT NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN

5 rows × 59 columns

In [4]:
lib_gdf.plot()
Out[4]:

Query Earth OnDemand for Imagery

We pass the GeoDataFrame into our earth_ondemand.read_catalog function. Because some geometries are points and others are polygons, we create a small buffer to force them all to be polygons, since the EarthAI Catalog API requires all input geometries to be the same type.

In [5]:
lib_gdf.geom_type.value_counts()
Out[5]:
Polygon 141
Point 103
dtype: int64
In [6]:
lib_cat = earth_ondemand.read_catalog(lib_gdf.buffer(0.0001),
                                      '2019-06-01',
                                      '2019-06-11',
                                      collections='landsat8_l1tp'
                                      )

You can download a companion notebook that runs through these steps from the attachment below.

Related articles

Comments

0 comments

Please sign in to leave a comment.