According to "Kaufman et al 1990", some peak heights ratio are needed for making comparison between different oil samples to reveal reservoir continuity. Based on which rules we should choose peaks? Which peak need to be selected for comparison?
What cause him to choose these peaks and connect them to each other( C,B,E,... )? What specific attributes or advantages they have?
This is really a correlation problem, and ratios of pairs of peaks is the simplest approach. It isn't necessary to know what compounds the peaks represent, although it helps in investigating what processes may be affecting composition (e.g. benzene will be more affected by water washing than say nC6). Usually, it's best to select pairs of peaks that elute near each other, and are resolved from other peaks, to avoid any variations in chromatography during analysis of the sample set. It can be more difficult to determine heights/areas of small peaks consistently accurately, so they are best avoided. Visual comparison of the chromatograms will help you pick peaks that show some variation - if there are any. It can also be useful to include a standard oil every few analyses so you can check that differences are due to genuine compositional variation among the study samples and not due to analytical effects. You can then examine the statistical relevance of any variations you observe using whatever software you have (there are some free simple multivariate packages available).