MySQL self-join to get two geographically closest items with different properties? - mysql

Question

We have a single table of weather markers on a map, where each marker type is either wind or cloud.
For each cloud marker we want to find the "WindDir" field value for the closest wind marker. It seems like a self-join would do the trick, but constructing the query is proving difficult.
Below is our incorrect pseudo SQL. We know the Distance value will be what we want. We just need to how to make the query find the closest wind marker for each cloud marker, so that we can get the wind marker's "WindDir" value.
SELECT w1.ID AS ID, w1.Lat AS Lat, w1.Lng AS Lng,
w2.WindDir AS WindDir,
MIN(SQRT(POWER(ABS(w1.Lng - w2.Lng), 2) + POWER(ABS(w1.Lat - w2.Lat), 2))*60) AS Distance
FROM Weather w1 WHERE w1.Marker="Cloud"
LEFT JOIN Weather w2 WHERE w2.Marker="Wind"
USING ID
We'd be grateful for any advice on making a version that works!
-Ken

Answer 1

You're pretty close (basically just reordered for syntax)
SELECT w1.ID AS ID, w1.Lat AS Lat, w1.Lng AS Lng,
w2.WindDir AS WindDir,
(SQRT(POWER(ABS(w1.Lng - w2.Lng), 2) + POWER(ABS(w1.Lat - w2.Lat), 2))*60) AS Distance
FROM Weather w1 LEFT JOIN Weather w2 USING (ID)
WHERE w1.Marker="Cloud" AND w2.Marker="Wind"
ORDER BY Distance DESC
LIMIT 1
Assuming you just want the closest result, I added a LIMIT 1 to restrict what's returned.
On a side note, this method of finding distance between two lat/lon pairs is unlikely to give you good results for even fairly small changes. There are methods for approximating these distances...