Understanding Geohash: Principles, Implementation, and Applications

Geohash is an algorithm that converts latitude and longitude into a short hash string, enabling fast and precise location queries in applications such as ride‑hailing and bike‑sharing.

Typical naive approaches store raw coordinates and scan them, which is inefficient. Geohash encodes a point into a base‑32 string; points with the same prefix are geographically close, allowing simple LIKE 'wtw366%' or LEFT(geohash,6)='wtw366' queries.

The encoding process consists of:

Specify the latitude and longitude.

Iteratively bisect the latitude (‑90 to 90) and longitude (‑180 to 180) ranges, recording 0 or 1 for each half‑interval.

Interleave the bits (even positions for longitude, odd for latitude).

Group the binary string into 5‑bit chunks and convert each chunk to a decimal value.

Map the decimal values to the base‑32 alphabet 0123456789bcdefghjkmnpqrstuvwxyz to obtain the final geohash.

Example PHP implementation (excerpt):

$minLat = -90; $maxLat = 90;
$minLng = -180; $maxLng = 180;
$latLength = $lngLength = 0;
$latList = $lngList = [];
$originPrecision = 30;
while ($latLength < $originPrecision) {
    $middle = ($minLat + $maxLat) / 2;
    if ($lat >= $middle) { $latList[] = 1; $minLat = $middle; }
    else { $latList[] = 0; $maxLat = $middle; }
    $latLength++;
}
while ($lngLength < $originPrecision) {
    $middle = ($minLng + $maxLng) / 2;
    if ($lng >= $middle) { $lngList[] = 1; $minLng = $middle; }
    else { $lngList[] = 0; $maxLng = $middle; }
    $lngLength++;
}

After interleaving and converting, the Shanghai Tencent Building (31.1688749, 121.3975184) yields the geohash wtw366ngz5qt . The length of the geohash determines precision: 12 characters correspond to ~37 mm, while 6 characters cover ~1.2 km.

Geohash has a boundary issue: two points near the edge of a cell may belong to different cells despite being closer than points inside the same cell. A common mitigation is to query the eight neighboring cells together.

Redis 3.2+ provides native GEO commands that store locations as 52‑bit integers (a form of geohash) in sorted sets, enabling fast radius queries via GEOSEARCH -like operations.

Distance between two coordinates can be computed with the spherical law of cosines or the haversine formula; PHP examples are provided.