How do we fix the network model when it doesn't match reality?

The paper identifies real world issues where the engineering calculations used to compute the variables that go into a power system network model can fall short of reality. This not only impacts the viability of the simulation tool used, but worse the accuracy of the data it provides to configure and set today's modern IED protection relays. The faulty result is often hidden until a fault of circumstance occurs to reveal the issue. But what data of the model needs correction?