Filter geo-temporal data by region

Use the geo.filterRows function to filter geo-temporal data by geographic region:

1. Define a geographic region
2. Use strict or non-strict filtering

The following example uses the sample bird migration data and queries data points within 200km of Cairo, Egypt:

import "experimental/geo"

sampleGeoData
  |> geo.filterRows(
    region: {lat: 30.04, lon: 31.23, radius: 200.0},
    strict: true
  )

Define a geographic region

Many functions in the Geo package filter data based on geographic region. Define a geographic region using one of the the following shapes:

• box
• circle
• polygon

box

Define a box-shaped region by specifying a record containing the following properties:

• minLat: minimum latitude in decimal degrees (WGS 84) (Float)
• maxLat: maximum latitude in decimal degrees (WGS 84) (Float)
• minLon: minimum longitude in decimal degrees (WGS 84) (Float)
• maxLon: maximum longitude in decimal degrees (WGS 84) (Float)

Example box-shaped region

{
  minLat: 40.51757813,
  maxLat: 40.86914063,
  minLon: -73.65234375,
  maxLon: -72.94921875
}

circle

Define a circular region by specifying a record containing the following properties:

• lat: latitude of the circle center in decimal degrees (WGS 84) (Float)
• lon: longitude of the circle center in decimal degrees (WGS 84) (Float)
• radius: radius of the circle in kilometers (km) (Float)

Example circular region

{
  lat: 40.69335938,
  lon: -73.30078125,
  radius: 20.0
}

polygon

Define a polygonal region with a record containing the latitude and longitude for each point in the polygon:

• points: points that define the custom polygon (Array of records)

  Define each point with a record containing the following properties:

  - **lat**: latitude in decimal degrees (WGS 84) _(Float)_
  - **lon**: longitude in decimal degrees (WGS 84) _(Float)_
  

Example polygonal region

{
  points: [
    {lat: 40.671659, lon: -73.936631},
    {lat: 40.706543, lon: -73.749177},
    {lat: 40.791333, lon: -73.880327}
  ]
}

Strict and non-strict filtering

In most cases, the specified geographic region does not perfectly align with S2 grid cells.

• Non-strict filtering returns points that may be outside of the specified region but inside S2 grid cells partially covered by the region.
• Strict filtering returns only points inside the specified region.

Strict filtering is less performant, but more accurate than non-strict filtering.

S2 grid cell
Filter region
Returned point