how to categorize lat/lon to find nearest city

Question

I am playing with iris (really neat.) and I have a list of cities lat/lons I am interested to see average temperature over time. I have.netcdf files with air temperatures covering entire country, I would like to tag data points in a cube with lat/lons closest to my cities so then I can easily get values I need just for these cities. or export data just for these cities somewhere.

I imagine I need to use add_categorised_coord somehow? https://scitools.org.uk/iris/docs/latest/iris/iris/coord_categorisation.html#iris.coord_categorisation.add_categorised_coord

I will appreciate an example. Thanks!

Answer 1

Assuming you have a gridded dataset of air temperature, a better solution would be to interpolate the data to given coordinate points, instead of "tagging" data points in a cube.

This can be done by looping over cities and their coordinates and using cube.interpolate() method. See https://scitools.org.uk/iris/docs/latest/userguide/interpolation_and_regridding.html#cube-interpolation-and-regridding for examples.

A more optimised solution would be to interpolate the data to all city points at once using the trajectory module:

import iris
import iris.analysis.trajectory as itraj
import numpy as np

# Create some dummy data
nx = 10
ny = 20

lonc = iris.coords.DimCoord(
    np.linspace(-5, 10, nx), units="degrees", standard_name="longitude"
)
latc = iris.coords.DimCoord(
    np.linspace(45, 55, ny), units="degrees", standard_name="latitude"
)
cube = iris.cube.Cube(
    np.random.rand(ny, nx),
    dim_coords_and_dims=((latc, 0), (lonc, 1)),
    standard_name="x_wind",
    units="m s^-1",
    attributes=dict(title="dummy_data"),
)

# Create a Nx2 array of city coordinates
city_coords = np.array([[50.7184, -3.5339], [48.8566, 2.3522], [52.6401898, 1.2517445]])

# Interpolate the data to the given points
sample_points = [("latitude", city_coords[:, 0]), ("longitude", city_coords[:, 1])]
cube_values_in_cities = itraj.interpolate(cube, sample_points, "linear")

Hope this helps.