output of geoparser gives locations with a latitude and longitude of the center of each place

• problem →some geographical entities are big, so plotting a point in the center of them might not be the best way to represent the information
• if we want to bring this data together with other information about the world - for example climate model data - we need a common geographical unit.
  • use grid cells, which divide the world into equally sized cells
    • with shapefiles → translate the places we have into our database into a polygon instead of a simple point → we can map our places we found in the studies to gridcells

coordinate grids

coordinate systems split the world up into 360 degrees of longitude (from 180° West to 180° East), and 180 degrees of latitude (from 90° South to 90° North). can be set up a grid like this with cells of a given size in degrees with numpy:

import numpy as np
def generate_grid_df(degrees):
    LON = np.linspace(-180+degrees*0.5,180-degrees*0.5,int(360/degrees))
    LAT = np.linspace(-90+degrees*0.5,90-degrees*0.5,int(180/degrees))
    lon_df, lat_df = np.meshgrid(LON,LAT)

    return pd.DataFrame({"LAT": lat_df.ravel(), "LON": lon_df.ravel()})
    
grid_df = generate_grid_df(2.5)
grid_df.head()

to check → merge with some other gridded data and plot

example:

• open population data
• round the grid points up to the grid cell size
• take the sum of larger grid cells

Shapefiles

to bridge between places and grid cells → get shape files that will give polygons for geographical entities

→ shapefiles from natural earth have different feature classes to the geographical entities in our dataset (used diff dictionary to map to the shapefile classes)