The array is below the human body, inside the backpack.
I am trying to replicate the issue, I would say that when the human body is between the array and the transmitter, the kraken sdr gets less signal as you said. In such cases reception can be lost, and bearings can be erratic.
In the tests I did at home, when the human body is NOT between the array and the transmitter, but near the array, I would say the main lobe can skew about 25 degrees, it keeps pointing an approximate bearing. You should arrive to the transmitting site, only you could need more time.
Perhaps the human body is a little issue, and a bigger one is multipath. I can remember now some other tests in Madrid with the array over the body. In the end the transmitter was found, but multipath in a big city is a problem and you need to insist.