Even on a quiet day, the magnetic field has a smooth variation. This variation is disregarded when one computes the K indices.
Then the difference between the maximum and minimum (for the three-hour interval in question) is computed for both horizontal components (normally X and Y meaning (geographical) north and east, or H and D meaning magnetic north and east).
The largest of the two differences is then compared to a table, different for different observatories. For the Lovö observatory it looks like this (upper limit inclusive):
|(This is my first table in HTML.)|