There are many possible answers, and surprisingly, nobody ever says how they create the trapezium. In your example, we can't even identify an actual person who probably created it. You can find a similar chart for Gaza City Arabic, which present both the classical vowel chart and a graph of F1 vs F2 in Bark, from which you might interpolate specific measurements and relate that to the trapezium. There are also similar charts for Ambel, Dagaare, Kalasha and Mono Lake Paiute (there are hundreds of Illustrations of the IPA). The manuscript guidelines (§4) don't say anything specific about the positioning of vowels.
The traditional approach is to position vowels by reference to the IPA auditory standard, which is sampled here. If the author of the Gaza Arabic article was following that practice, that means that he judged the vowel [i:] in the language to be somewhat retracted and lower that IPA [i]. The author of the Ambel article would appear to have declined to make any finer-grained judgments (the Wiki graph-creator on the other hand does not decline). The only way to know how the vowels were positioned on the trapezium is to write to the author. One could position the vowels using formant values, though this is usually not done (except in formant charts, which are a different thing).
When vowels align perfectly, you can assume that the author is not being more specific than to say "close mid front vowel" which is the coarsest level of analysis. The Kalasha article does say that the vowels positions are "based on" formants, though they do not state what the equation is (exercise for the reader, I suppose) that relates their formant graphs to the trapezium. The Wikipedia graphs don't actually align perfectly, but there is no information on how the points were positioned in those two cases.
