How to query coords in between values that we have in 2 custom fields?

Question

I have two custom fields on my posts with values that could be like

get_usp-custom-19 = 40.85150386578784
get_usp-custom-20 = 14.258907499999964

Then I receive a GET from a form with some values like this:

$lat = $_GET['usp-custom-19'];
$ln = $_GET['usp-custom-20'];

Now $lat and $ln have latitude and a longitude separated values and I create a query like:

$args = array(
    'post_type' => 'post',
    'meta_query' => array(
        array(
            'relation' => 'AND',
            array(
                'key' => 'get_usp-custom-19',
                'value' => $lat,
                'compare' => 'BETWEEN',
                'type' => 'NUMERIC',
            ),
            array(
                'key' => 'get_usp-custom-20',
                'value' => $ln,
                'compare' => 'BETWEEN',
                'type' => 'NUMERIC',
            ),
        ),
    ),
);

But I get wrong results as it isn't comparing in between, I also tried to set it type as CHAR or DECIMALS but still, wrong results. Also I could get negative coords like -9.258907499999964 and I was reading about cons and sen or using abs(), but I am getting very confused now

UPDATE

Here I try to create a radius

$lat = $_GET['usp-custom-19'];
$ln = $_GET['usp-custom-20'];

$args = get_posts( 
    array( 
        'post_type'      => 'post', 
        'posts_per_page' => -1, 
    ) 
);
$query = new WP_Query( $args );
if ( $query->have_posts() ) {
    while ( $query->have_posts() ) {
        $query->the_post();
        $customCoords = usp_get_meta(false, 'usp-custom-90');
        $arrayCoords = explode( ",", $customCoords );
        $radiusLn = +$arrayCoords[0] + 10;
        $radiusLat = +$arrayCoords[1] + 10;
        $args = array(
            'post_type' => 'post',
            'meta_query' => array(
                array(
                    'relation' => 'AND',
                    array(
                        'key' => 'get_usp-custom-19',
                        'value' => array($ln, $radiusLn),
                        'compare' => '>='
                    ),
                    array(
                        'key' => 'get_usp-custom-20',
                        'value' => array($lat, $radiusLat),
                        'compare' => '<='
                    ),
                ),
            ),
        );
        $query = new WP_Query( $args );
        if ( $query->have_posts() ) {
            while ( $query->have_posts() ) {
                $query->the_post();
                the_title();
            }
        }
    }
}

Answer 1

Basically, you want to make sure that the value of the key and the $lat and $ln are the same type and they can be compared, for example, decimal.

Then you can use >= and <= to compare.

$args = array(
  'post_type' => 'post',
  'meta_query' => array(
     array(
       'relation' => 'AND',
        array(
            'key' => 'get_usp-custom-19', // make sure the value of this field is the same as $lat; 
            'value' => $lat,
            'compare' => '>=',
            'type' => 'NUMERIC',
        ),
        array(
            'key' => 'get_usp-custom-20', // make sure the value of this field is the same as $ln; 
            'value' => $ln,
            'compare' => '<=',
            'type' => 'NUMERIC',
        ),
    ),
  ),
);

You might also want to change the type of $lat and $ln first.

This post explains the compare operators pretty well.

Answer 2

I am guessing that your are trying to find the closest posts with respect to lat and long. In this case I think a raw SQL query is easier to use:

$post_ids = $wpdb->get_col( <<<EOD
SELECT m.post_id FROM $wpdb->postmeta m, $wpdb->postmeta n
    WHERE m.post_id=n.post_id AND m.meta_key='get_usp-custom-19' AND n.meta_key='get_usp-custom-20'
        AND (POW( CAST(m.meta_value AS DECIMAL(7,4)) - $lat, 2 )
                + POW( COS(0.0175 * $lat) * ( CAST(n.meta_value AS DECIMAL(7,4)) - $ln ), 2 ) ) )
            < $rad * $rad
EOD
);

The rows in the SQL join have the lat and long and then the results are where the sum of the squares of the differences from a given ($lat,$ln) is less than $rad squared.

Addenda:

The original query omitted the COS correction which must be applied to longitude degrees. This uses the Pythagorean theorem which is only valid on flat surfaces. The two points need to be close enough so that the triangle is essentially on a flat surface. (Otherwise, you need a "Great Circle" formula.) Also, there will be a problem if the two points straddle the opposite of the Prime Meridian - the meridian of 180 degrees - International Date Line.

Response to Comments:

The query as given is indeed to inefficient for large databases. The problem is that the join is done over the entire postmeta table. To fix this you can do the join over derived tables where the left derived table is restricted to rows with latitudes close to the given latitude and the right derived table are restricted to rows with longitude close to the given longitude.

The Pythagorean theorem is indeed not valid on the surface of the sphere. However, the objective here is to find close points not compute completely accurate distances. Since, we are looking for close points the distortion of the Pythagorean theorem is very small as the points are essentially on a flat surface. In my opinion, the Haversine formula is computationally far more expensive and provides unnecessary additional accuracy with respect to this problem.