Because the edge of frosting is around 0.1*C (celsius). And it's usually well below that. For example -2*C.
Medium inside the exchanger gets closer to external temp. Therefore negative* temps for glycol fluid. Absolutely works that way too.
So heat exchanger inlet is for example -15*C and outlet is -5*C while external temperature is -2*C.
Whatever touches the fins will either get stuck (snow) or freeze (water droplet, air moisture, salty water droplet).
Love the idea of dropping off chunks of ice
*negative - I mean like freezing temps, or whatever word is used in imperial.
The amount of frozen material is not 1:1 compared to energy extracted.
For example, within 30 minutes of HP operation heat exchanger can extract 3kW of energy (average rate 6kW).
Then if "fully frosted". Defrost cycle required thawing 500ml of water (half a liter) from -2*C up to +0*C (incl phase change).
That requires much less energy than was extracted "in between freezing cycles".
0.5l of water temp to be raised for 2 degrees requires 0.5*2calories=1 calorie. 0.0005kWh+0,047kWh phase change.
Therefore we could round it to 0.05kWh per half liter of water (almost half kg).
So the reason to defrost is because defrosting required 0,05kWh of energy and after that heat exhanger can supply 3kWh
until it is fully frosted again.