The use of incomplete global data for probabilistic event trees: challenges and strategies