Here's a shorter, data-driven version of your program that produces the same output as yours, and should be more efficient too:

#!/usr/bin/perl -w
use strict;
use warnings;

print "Content-type: text/html\n\n";

my @locations = qw(LA LB LC LD LE);


my %loc = (
    LA  => { NE => 1, E  => 1, SE => 1, S  => 1, SW => 1 },
    LB  => { NE => 1, SW => 1, W  => 1, NW => 1},
    LC  => { SE => 1, S  => 1, SW  => 1},
    LD  => { SE => 1, S  => 1, SW => 1},
);

my %results = (
    LA => '1LA= NE|E|SE|S|SW',
    LB => '2LB= NE|SW|W|NW',
    LC => '3LC= SE|S|SW',
    LD => '4LD= SE|S|SW',
);

my $c_position = '';

foreach my $location (@locations) {
    my $test_loc = "Park: SW at 10 position"; #this could be for LA
    my $locs = substr $test_loc, 6, 2;
    $locs=~s/\s+//;

    $c_position .= ' ';
    if ($loc{$location}{$locs}) {
        $c_position .= $results{$location};
    } else {
        $c_position .= '5ALL OTHERS HERE.';
    }
}
print "Results= $c_position\n\n";
[download]

This is not quite perfect; for example one could unify %results %loc (pretty easy if the order of letters in the values of %results doesn't matter), but it's a start.

It also makes it possible to read the hashes from config files, so that a long list of conditions requires no code change.