gridding-py 0.4.2

Build geographical grids

gridding

Build geographical grids in Python

This is a Python library implementing the "carroyage" for geographical data.

Motivation

In order to be able to make analysis in the more granular level of the "carreau", this provides an easy reference to building custom grids by allowing to assign a unique tile / "carreau" ID to a geographical coordinate (latitude, longitude) or a French postal address.

Indeed, the result of the computation gives the ID according to the EU directive INSPIRE in the following format: WGS84|RES200m|N2471400|E0486123 that could be decomposed as a pipe-separated string with:

• the projection system, eg. WGS84;
• the resolution of the grid, ie. the size of the square, eg. RES200m for $200\text{ m}$;
• the bottom-left point of the grid's bounding box defined by:
  • its latitude in decimal degrees, eg. N2471400 for $\text{N }24.714000°$;
  • its longitude, eg. E0486123 for $\text{E }4.861230°$;

as well as the coordinate in the grid in the form of <row>.<column> from the initial bottom-left point.

It would need as input parameters:

• the coordinates of the initial point of the grid (by default the bottom-left point of the bounding box covering the French metropolitan territory: $\text{N }41.316666°$ / $\text{W }5.151110$);
• the size and scale of the grid, eg. 200m, 1km, ...;
• the projection system (default: WGS84).

Below are the calculation bases.

Geographic Information System (GIS) computations

Latitude

In WGS84, one degree of latitude is almost equivalent to $111\text{ km}$ everywhere on Earth.

The distance in meter, $D_{lat}$, for one degree in latitude is given by the following formula:

D_{lat} = \pi \cdot \frac{a (1 - e^2)}{(1 - e^2 \sin^2(\text{latitude}))^{3/2}} \times \frac{1}{180}

where

• $a$ is the semi-major axis of the Earth's ellipsoid (in WGS84, $a = 6378137$);
• $e$ is the excentricity of the ellipsoid, given by:

e = \sqrt{1 - \frac{b^2}{a^2}}

(in WGS84, $b = 6356752.3142$)

• $latitude$ is the latitude of the point in radians.

From here, one can compute the degree in latitude for any size of tile / "carreau", eg. in WGS84, a "carreau" of height $200\text{ m} \approx 0.001796°$:

\Delta latitude = \frac{distance}{D_{lat}}

where

• $distance$ is the vertical distance to convert in degrees of latitude, eg. $200$;
• $D_{lat}$ is the result of the computation above around the given $latitude$.

Longitude

The variation of longitude as a function of latitude is calculated by taking into account the decrease in the distance between the meridians as one moves away from the equator. This decrease is due to the fact that the Earth is spherical.

In WGS84, the formula that allows one to calculate the distance corresponding to a degree of longitude as a function of a latitude is the following:

\Delta{longitude} = \frac{distance}{D_{lon} \times \cos(latitude)}

where

• $distance$ is the horizontal distance to convert in degrees of longitude, eg. $200$;
• $latitude$ is the starting point expressed in radians, eg. $45° = \frac{\pi}{4}$;
• $D_{lon}$ is the result of the computation below at the given $latitude$:

D_\text{lon} = \pi \cdot a \cdot \cos(\text{latitude}) \cdot \frac{1}{180} \cdot \frac{1}{\sqrt{1 - e^2 \sin^2(\text{latitude})}}

XY coordinates

The UTM projection would be used to handle XY coordinates with the central longitude $\lambda_0$ being automatically calculated for the UTM zone with the longitude of the searched point, eg. $-3°$ for the UTM zone 30T covering France from $W 6°$ to $0°$.

The following data and formula would be used in that case: eg. for WGS84

• $a = 6378137$
• $e = 0.0818191908426$
• $k_0 = 0.9996$
• $\varphi = latitude \times \frac{\pi}{180}$
• $\lambda = longitude \times \frac{\pi}{180}$
• $N = \frac{a}{\sqrt{1 - e^2\sin^2(\phi)}}$
• $T = \tan^2(\varphi)$
• $C = \frac{e^2\cos^2(\varphi)}{1 - e^2}$
• $A = (\lambda - \lambda_0)\cos(\varphi)$
• $M = a \left( (1 - \frac{e^2}{4} - \frac{3 e^4}{64} - \frac{5 e^6}{256}) \varphi - (\frac{3 e^2}{8} + \frac{3 e^4}{32} + \frac{45 e^6}{1024}) \sin(2 \varphi) + (\frac{15 e^4}{256} + \frac{45 e^6}{1024}) \sin(4 \varphi) - (\frac{35 e^6}{3072}) \sin(6 \varphi) \right)$

One can now compute the XY coordinates as follows:

• $X = k_0 \left( N ( A + \frac{(1 - T + C) A^3}{6} + \frac{(5 - 18 T + T^2 + 72 C - 58 e^2) A^5}{120} ) \right) + 500000$
• $Y = k_0 \left( M + N \tan(\varphi) ( \frac{A^2}{2} + \frac{(5 - T + 9 C + 4 C^2) A^4}{24} + \frac{(61 - 58 T + T^2 + 600 C - 330 e^2) A^6}{720} ) \right)$

NB: If the searched point is in the south, add $10,000,000$ to $Y$ to avoid negative coordinates.

With all these formulas, one is now able to build a full system placing any GPS or XY coordinates into a unique "carreau". This library implements one way to do it.

Install

$ git clone https://github.com/cyrildever/gridding.git
$ cd gridding/packages/py/
$ python3 -m venv venv
$ source venv/bin/activate
$ pip install build twine

Usage

pip install gridding-py

If you want to use the from_address() feature, you need to build or upload the address repository.

Your one-line address must be compliant with the following guidelines:

• elements must be in the following order: <numero> <repetitor> <voie> <lieu_dit> <cp> <ville>;
• <numero> is the street number;
• <repetitor> is the eventual addition to the number, eg. bis, ter, ...;
• <voie> is the street name including its type;
• <lieu_dit> is a distinct locality (neither the city nor a postal box information);
• <cp> is the postal code in a full integer format (2A000 must be transformed in 20000 for Corsica for example);
• <ville> is the city name (ideally without further information, ie. no Arrondissement or Cedex mentions).

For now, only French addresses are fully covered, but the library can work with any other country providing you build the corresponding repository.
Feel free to add any normalizing tools through pull requests in the library to help enriching it.

1) Module

To get the tile / "carreau" from GPS coordinates:

from gridding import METER, WGS84, GPS, Grid, Resolution

grid = Grid(
    Resolution(200, METER),
    GPS(-5.151110, 41.316666),
    WGS84(),
)
carreau, tile = grid.from_gps(my_point)
print(f"This GPS point belongs to the carreau with code: {carreau}, and coordinate: {tile.to_string()}")

To get it from a French postal address:

from gridding import FR

carreau, tile = grid.from_address("9 boulevard Gouvion Saint-Cyr 75017 Paris", FR())
print(f"This address belongs to the carreau with code: {carreau}, and coordinate: {tile.to_string()}")

To get it from X/Y coordinates:

from gridding import XY

my_point = XY(647872.07, 5110548.44, "UTM", "WGS84")
carreau, tile = grid.from_xy(my_point)
print(f"This X/Y point belongs to the carreau with code: {carreau}, and coordinate: {tile.to_string()}")

You may want to use the "distance" between two tiles (which returns a sort of average radius in tiles from one tile to the other):

distance = Tile.Distance(tile1, tile2)
print(f"The radius between these two tiles is equal to {distance} tiles")

The idea here is to be able to easily know if a tile is within a certain tile distance of another, eg. two tiles away in each direction.

For now, only the WGS-84 system is available and can be used in conjunction with the UTM projection if need be to get X/Y coordinates.

IMPORTANT: when using negative coordinates for a pivot point, be sure to give it a 0 as 6th decimal.

2) Script

You may also use the main script to get the "carreau" from some GPS coordinates directly in a terminal, eg.

usage: python -m gridding [-h] -x LONGITUDE -y LATITUDE -r RESOLUTION [-o | --obfuscate | --no-obfuscate] [-t | --tile | --no-tile]

options:
  -h, --help            show this help message and exit
  -x LONGITUDE, --longitude LONGITUDE
                        the longitude in decimal degrees
  -y LATITUDE, --latitude LATITUDE
                        the latitude in decimal degrees
  -r RESOLUTION, --resolution RESOLUTION
                        the grid resolution, eg. '200m'
  -o, --obfuscate, --no-obfuscate
                        add to return a hashed result (default: no)
  -t, --tile, --no-tile
                        add to return the tile coordinates instead of the carreau id (default: no)

NB: The optional obfuscated result is a unique SHA-256 hexadecimal string.

Tests

$ pip install -e . && python3 -m unittest discover

License

This module is distributed under a MIT license.
See the LICENSE file.

---

© 2024 Cyril Dever. All rights reserved.