Finding geohashes of certain length within radius from a point

Question

I have points with a given latlong and a distance around them - e.g. { 40.6826048,-74.0288632 : 20 miles, 51.5007825,-0.1258957 : 100 miles}. If I pick a fixed geohash lengt

Answer 1

This is how I would try to do:

Input: Point of interest(lat, long), Query Radius

Step 1: Find the 'MINIMUM' BOUNDING RECTANGLE(MBR) which completely contains the QUERY CIRCLE

Step 2: To create the minimum bounding rectangle, first calculate its minimum and maximum lat, long using the input parameters. Please refer to section 3.1 and 3.3 of Computing the Minimum and Maximum Latitude Longitude – the Correct Way

Step 3: Using (minLat, minLon), (maxLat, maxLon) calculate the four corners of the MBR NorthWest(maxLat, minLon), SouthWest(minLat, minLon), SouthEast(minLat, maxLon), NorthEast(maxLat, maxLon)

Step 4: Calculate the GeoHash of all four corners of MBR

Ex: for a point in NYC, say (40.75798, -73.991516), distance: 800 Meters and GeoHash length: 12

• NorthWest : dr5ruj4477kd
• SouthWest : dr5ru46ne2ux
• SouthEast : dr5ru6ryw0cp
• NorthEast : dr5rumpfq534

Step 5: From these GeoHashes, calculate the Query Bounding Box(MBR) Prefix: dr5ru

This would give you the coarser GeoHash which completely contains our MBR and hence the query region. In other words, all points indexed by dr5ru, yielding with 32 GeoHashes from dr5ru0 - dr5ruz

Final Step:

To find the exact grids (or) GeoHashes that correspond to our Query Circle(Square(MBR) to be precise), we should pick from these 32 GeoHashes by representing a recurring (4X8) Matrix using 2D Array.

In our example: we get dr5ru + J, M, H, K, 5, 7, 4, 6. All these GeoHashes represent the points that are within 800 meters from the Central Query Point, Except very few GeoHashes, which could not be avoided, because of considering MBR instead of a perfect circle.

---

THE OVERALL PROCESS IN A SINGLE GIF: (Step 1- 5)

---

FINAL STEP:

Important: Please find the use of 4 x 8 Grid for GeoHash. It varies for each character along the length of GeoHash. For ODD lengths it is 8 x 4, for even its transpose 4 X 8. In our case, we are inside dr5ru(5 + 1, 6th resolution) and hence we use 4 X 8

Answer 2

Have a look at this -> ProximityHash.

ProximityHash generates a set of geohashes that cover a circular area, given the center coordinates and the radius. It also has an additional option to use GeoRaptor that creates the best combination of geohashes across various levels to represent the circle, starting from the highest level and iterating till the optimal blend is brewed. Result accuracy remains the same as that of the starting geohash level, but data size reduces considerably, thereby improving speed and performance.